diff --git a/.github/workflows/ci.yaml b/.github/workflows/ci.yaml
index 90d81636..5bdb2bed 100644
--- a/.github/workflows/ci.yaml
+++ b/.github/workflows/ci.yaml
@@ -53,6 +53,15 @@ jobs:
         with:
           command: check ${{ matrix.checks }}
 
+  js_tests:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v2
+      - name: Install wasm-pack
+        run: curl https://rustwasm.github.io/wasm-pack/installer/init.sh -sSf | sh
+      - name: run tests
+        run: ./scripts/ci/js_tests
+
   linux:
     runs-on: ubuntu-latest
     strategy:
diff --git a/automerge-js/.gitignore b/automerge-wasm/automerge-js/.gitignore
similarity index 100%
rename from automerge-js/.gitignore
rename to automerge-wasm/automerge-js/.gitignore
diff --git a/automerge-js/package.json b/automerge-wasm/automerge-js/package.json
similarity index 86%
rename from automerge-js/package.json
rename to automerge-wasm/automerge-js/package.json
index 17018429..f0e65a18 100644
--- a/automerge-js/package.json
+++ b/automerge-wasm/automerge-js/package.json
@@ -10,7 +10,7 @@
     "mocha": "^9.1.1"
   },
   "dependencies": {
-    "automerge-wasm": "file:../automerge-wasm",
+    "automerge-wasm": "file:../dev",
     "fast-sha256": "^1.3.0",
     "pako": "^2.0.4",
     "uuid": "^8.3"
diff --git a/automerge-js/src/columnar.js b/automerge-wasm/automerge-js/src/columnar.js
similarity index 100%
rename from automerge-js/src/columnar.js
rename to automerge-wasm/automerge-js/src/columnar.js
diff --git a/automerge-js/src/common.js b/automerge-wasm/automerge-js/src/common.js
similarity index 100%
rename from automerge-js/src/common.js
rename to automerge-wasm/automerge-js/src/common.js
diff --git a/automerge-js/src/constants.js b/automerge-wasm/automerge-js/src/constants.js
similarity index 100%
rename from automerge-js/src/constants.js
rename to automerge-wasm/automerge-js/src/constants.js
diff --git a/automerge-js/src/counter.js b/automerge-wasm/automerge-js/src/counter.js
similarity index 100%
rename from automerge-js/src/counter.js
rename to automerge-wasm/automerge-js/src/counter.js
diff --git a/automerge-js/src/encoding.js b/automerge-wasm/automerge-js/src/encoding.js
similarity index 100%
rename from automerge-js/src/encoding.js
rename to automerge-wasm/automerge-js/src/encoding.js
diff --git a/automerge-js/src/index.js b/automerge-wasm/automerge-js/src/index.js
similarity index 100%
rename from automerge-js/src/index.js
rename to automerge-wasm/automerge-js/src/index.js
diff --git a/automerge-js/src/numbers.js b/automerge-wasm/automerge-js/src/numbers.js
similarity index 100%
rename from automerge-js/src/numbers.js
rename to automerge-wasm/automerge-js/src/numbers.js
diff --git a/automerge-js/src/proxies.js b/automerge-wasm/automerge-js/src/proxies.js
similarity index 100%
rename from automerge-js/src/proxies.js
rename to automerge-wasm/automerge-js/src/proxies.js
diff --git a/automerge-js/src/sync.js b/automerge-wasm/automerge-js/src/sync.js
similarity index 100%
rename from automerge-js/src/sync.js
rename to automerge-wasm/automerge-js/src/sync.js
diff --git a/automerge-js/src/text.js b/automerge-wasm/automerge-js/src/text.js
similarity index 100%
rename from automerge-js/src/text.js
rename to automerge-wasm/automerge-js/src/text.js
diff --git a/automerge-js/src/uuid.js b/automerge-wasm/automerge-js/src/uuid.js
similarity index 100%
rename from automerge-js/src/uuid.js
rename to automerge-wasm/automerge-js/src/uuid.js
diff --git a/automerge-js/test/basic_test.js b/automerge-wasm/automerge-js/test/basic_test.js
similarity index 100%
rename from automerge-js/test/basic_test.js
rename to automerge-wasm/automerge-js/test/basic_test.js
diff --git a/automerge-js/test/columnar_test.js b/automerge-wasm/automerge-js/test/columnar_test.js
similarity index 100%
rename from automerge-js/test/columnar_test.js
rename to automerge-wasm/automerge-js/test/columnar_test.js
diff --git a/automerge-js/test/helpers.js b/automerge-wasm/automerge-js/test/helpers.js
similarity index 100%
rename from automerge-js/test/helpers.js
rename to automerge-wasm/automerge-js/test/helpers.js
diff --git a/automerge-js/test/legacy_tests.js b/automerge-wasm/automerge-js/test/legacy_tests.js
similarity index 100%
rename from automerge-js/test/legacy_tests.js
rename to automerge-wasm/automerge-js/test/legacy_tests.js
diff --git a/automerge-js/test/sync_test.js b/automerge-wasm/automerge-js/test/sync_test.js
similarity index 100%
rename from automerge-js/test/sync_test.js
rename to automerge-wasm/automerge-js/test/sync_test.js
diff --git a/automerge-js/test/text_test.js b/automerge-wasm/automerge-js/test/text_test.js
similarity index 100%
rename from automerge-js/test/text_test.js
rename to automerge-wasm/automerge-js/test/text_test.js
diff --git a/automerge-js/test/uuid_test.js b/automerge-wasm/automerge-js/test/uuid_test.js
similarity index 100%
rename from automerge-js/test/uuid_test.js
rename to automerge-wasm/automerge-js/test/uuid_test.js
diff --git a/automerge-wasm/src/lib.rs b/automerge-wasm/src/lib.rs
index 80a3d65f..f3bbda98 100644
--- a/automerge-wasm/src/lib.rs
+++ b/automerge-wasm/src/lib.rs
@@ -8,6 +8,7 @@ use std::collections::{HashMap, HashSet};
 use std::convert::TryFrom;
 use std::convert::TryInto;
 use std::fmt::Display;
+use std::str::FromStr;
 use wasm_bindgen::prelude::*;
 use wasm_bindgen::JsCast;
 
@@ -149,7 +150,7 @@ impl Automerge {
     }
 
     pub fn keys(&mut self, obj: JsValue, heads: JsValue) -> Result<Array, JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         let result = if let Some(heads) = get_heads(heads) {
             self.0.keys_at(obj, &heads)
         } else {
@@ -162,7 +163,7 @@ impl Automerge {
     }
 
     pub fn text(&mut self, obj: JsValue, heads: JsValue) -> Result<JsValue, JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         if let Some(heads) = get_heads(heads) {
             self.0.text_at(obj, &heads)
         } else {
@@ -179,7 +180,7 @@ impl Automerge {
         delete_count: JsValue,
         text: JsValue,
     ) -> Result<(), JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         let start = to_usize(start, "start")?;
         let delete_count = to_usize(delete_count, "deleteCount")?;
         let mut vals = vec![];
@@ -214,7 +215,7 @@ impl Automerge {
         value: JsValue,
         datatype: JsValue,
     ) -> Result<JsValue, JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         //let key = self.insert_pos_for_index(&obj, prop)?;
         let index: Result<_, JsValue> = index
             .as_f64()
@@ -235,7 +236,7 @@ impl Automerge {
         value: JsValue,
         datatype: JsValue,
     ) -> Result<JsValue, JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         let prop = self.import_prop(prop)?;
         let value = self.import_value(value, datatype)?;
         let opid = self.0.set(obj, prop, value).map_err(to_js_err)?;
@@ -246,7 +247,7 @@ impl Automerge {
     }
 
     pub fn inc(&mut self, obj: JsValue, prop: JsValue, value: JsValue) -> Result<(), JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         let prop = self.import_prop(prop)?;
         let value: f64 = value
             .as_f64()
@@ -257,7 +258,7 @@ impl Automerge {
     }
 
     pub fn value(&mut self, obj: JsValue, prop: JsValue, heads: JsValue) -> Result<Array, JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         let result = Array::new();
         let prop = to_prop(prop);
         let heads = get_heads(heads);
@@ -284,7 +285,7 @@ impl Automerge {
     }
 
     pub fn values(&mut self, obj: JsValue, arg: JsValue, heads: JsValue) -> Result<Array, JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         let result = Array::new();
         let prop = to_prop(arg);
         if let Ok(prop) = prop {
@@ -316,7 +317,7 @@ impl Automerge {
     }
 
     pub fn length(&mut self, obj: JsValue, heads: JsValue) -> Result<JsValue, JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         if let Some(heads) = get_heads(heads) {
             Ok((self.0.length_at(obj, &heads) as f64).into())
         } else {
@@ -325,7 +326,7 @@ impl Automerge {
     }
 
     pub fn del(&mut self, obj: JsValue, prop: JsValue) -> Result<(), JsValue> {
-        let obj = self.import(obj)?;
+        let obj: automerge::ObjId = self.import(obj)?;
         let prop = to_prop(prop)?;
         self.0.del(obj, prop).map_err(to_js_err)?;
         Ok(())
@@ -442,16 +443,18 @@ impl Automerge {
         }
     }
 
-    fn export<E: automerge::Exportable>(&self, val: E) -> JsValue {
-        self.0.export(val).into()
+    fn export<D: std::fmt::Display>(&self, val: D) -> JsValue {
+        val.to_string().into()
     }
 
-    fn import<I: automerge::Importable>(&self, id: JsValue) -> Result<I, JsValue> {
-        let id_str = id
+    fn import<F: FromStr>(&self, id: JsValue) -> Result<F, JsValue> 
+        where F::Err: std::fmt::Display
+    {
+        id
             .as_string()
-            .ok_or("invalid opid/objid/elemid")
-            .map_err(to_js_err)?;
-        self.0.import(&id_str).map_err(to_js_err)
+            .ok_or("invalid opid/objid/elemid")?
+            .parse::<F>()
+            .map_err(to_js_err)
     }
 
     fn import_prop(&mut self, prop: JsValue) -> Result<Prop, JsValue> {
diff --git a/automerge/src/change.rs b/automerge/src/change.rs
index 4d3984e5..80fe571c 100644
--- a/automerge/src/change.rs
+++ b/automerge/src/change.rs
@@ -7,8 +7,8 @@ use crate::decoding::{Decodable, InvalidChangeError};
 use crate::encoding::{Encodable, DEFLATE_MIN_SIZE};
 use crate::legacy as amp;
 use crate::{
-    ActorId, AutomergeError, ElemId, IndexedCache, Key, ObjId, Op, OpId, OpType, Transaction, HEAD,
-    ROOT,
+    types::{ObjId, OpId},
+    ActorId, AutomergeError, ElemId, IndexedCache, Key, Op, OpType, Transaction, HEAD,
 };
 use core::ops::Range;
 use flate2::{
@@ -417,10 +417,9 @@ fn increment_range_map(ranges: &mut HashMap<u32, Range<usize>>, len: usize) {
 }
 
 fn export_objid(id: &ObjId, actors: &IndexedCache<ActorId>) -> amp::ObjectId {
-    if id.0 == ROOT {
-        amp::ObjectId::Root
-    } else {
-        export_opid(&id.0, actors).into()
+    match id {
+        ObjId::Root => amp::ObjectId::Root,
+        ObjId::Op(op) => export_opid(op, actors).into() 
     }
 }
 
@@ -433,7 +432,7 @@ fn export_elemid(id: &ElemId, actors: &IndexedCache<ActorId>) -> amp::ElementId
 }
 
 fn export_opid(id: &OpId, actors: &IndexedCache<ActorId>) -> amp::OpId {
-    amp::OpId(id.0, actors.get(id.1).clone())
+    amp::OpId(id.counter(), actors.get(id.actor()).clone())
 }
 
 fn export_op(op: &Op, actors: &IndexedCache<ActorId>, props: &IndexedCache<String>) -> amp::Op {
diff --git a/automerge/src/clock.rs b/automerge/src/clock.rs
index 979885b3..7edab530 100644
--- a/automerge/src/clock.rs
+++ b/automerge/src/clock.rs
@@ -1,4 +1,4 @@
-use crate::OpId;
+use crate::types::OpId;
 use fxhash::FxBuildHasher;
 use std::cmp;
 use std::collections::HashMap;
@@ -19,8 +19,8 @@ impl Clock {
     }
 
     pub fn covers(&self, id: &OpId) -> bool {
-        if let Some(val) = self.0.get(&id.1) {
-            val >= &id.0
+        if let Some(val) = self.0.get(&id.actor()) {
+            val >= &id.counter()
         } else {
             false
         }
@@ -38,15 +38,15 @@ mod tests {
         clock.include(1, 20);
         clock.include(2, 10);
 
-        assert!(clock.covers(&OpId(10, 1)));
-        assert!(clock.covers(&OpId(20, 1)));
-        assert!(!clock.covers(&OpId(30, 1)));
+        assert!(clock.covers(&OpId::new(10, 1)));
+        assert!(clock.covers(&OpId::new(20, 1)));
+        assert!(!clock.covers(&OpId::new(30, 1)));
 
-        assert!(clock.covers(&OpId(5, 2)));
-        assert!(clock.covers(&OpId(10, 2)));
-        assert!(!clock.covers(&OpId(15, 2)));
+        assert!(clock.covers(&OpId::new(5, 2)));
+        assert!(clock.covers(&OpId::new(10, 2)));
+        assert!(!clock.covers(&OpId::new(15, 2)));
 
-        assert!(!clock.covers(&OpId(1, 3)));
-        assert!(!clock.covers(&OpId(100, 3)));
+        assert!(!clock.covers(&OpId::new(1, 3)));
+        assert!(!clock.covers(&OpId::new(100, 3)));
     }
 }
diff --git a/automerge/src/columnar.rs b/automerge/src/columnar.rs
index 3a1df3cb..3339a68e 100644
--- a/automerge/src/columnar.rs
+++ b/automerge/src/columnar.rs
@@ -11,8 +11,7 @@ use std::{
     str,
 };
 
-use crate::ROOT;
-use crate::{ActorId, ElemId, Key, ObjId, ObjType, OpId, OpType, ScalarValue};
+use crate::{ActorId, ElemId, Key, ObjType, OpType, ScalarValue, types::{ObjId, OpId}};
 
 use crate::legacy as amp;
 use amp::SortedVec;
@@ -686,15 +685,15 @@ impl KeyEncoder {
                 self.ctr.append_null();
                 self.str.append_value(props[i].clone());
             }
-            Key::Seq(ElemId(OpId(0, 0))) => {
+            Key::Seq(ElemId(opid)) if opid.actor() == 0 && opid.counter() == 0 => {
                 // HEAD
                 self.actor.append_null();
                 self.ctr.append_value(0);
                 self.str.append_null();
             }
-            Key::Seq(ElemId(OpId(ctr, actor))) => {
-                self.actor.append_value(actors[actor]);
-                self.ctr.append_value(ctr);
+            Key::Seq(ElemId(opid)) => {
+                self.actor.append_value(actors[opid.actor()]);
+                self.ctr.append_value(opid.counter());
                 self.str.append_null();
             }
         }
@@ -773,8 +772,8 @@ impl SuccEncoder {
     fn append(&mut self, succ: &[OpId], actors: &[usize]) {
         self.num.append_value(succ.len());
         for s in succ.iter() {
-            self.ctr.append_value(s.0);
-            self.actor.append_value(actors[s.1]);
+            self.ctr.append_value(s.counter());
+            self.actor.append_value(actors[s.actor()]);
         }
     }
 
@@ -845,14 +844,14 @@ impl ObjEncoder {
     }
 
     fn append(&mut self, obj: &ObjId, actors: &[usize]) {
-        match obj.0 {
-            ROOT => {
+        match obj {
+            ObjId::Root => {
                 self.actor.append_null();
                 self.ctr.append_null();
             }
-            OpId(ctr, actor) => {
-                self.actor.append_value(actors[actor]);
-                self.ctr.append_value(ctr);
+            ObjId::Op(opid) => {
+                self.actor.append_value(actors[opid.actor()]);
+                self.ctr.append_value(opid.counter());
             }
         }
     }
diff --git a/automerge/src/external_types.rs b/automerge/src/external_types.rs
new file mode 100644
index 00000000..edaeef7d
--- /dev/null
+++ b/automerge/src/external_types.rs
@@ -0,0 +1,109 @@
+use std::{borrow::Cow, fmt::Display, str::FromStr};
+
+use crate::{op_tree::OpSetMetadata, types::OpId, ActorId};
+
+const ROOT_STR: &str = "_root";
+
+#[derive(Debug, PartialEq, Clone, Hash, Eq)]
+pub struct ExternalOpId {
+    counter: u64,
+    actor: ActorId,
+}
+
+impl ExternalOpId {
+    pub(crate) fn from_internal(opid: &OpId, metadata: &OpSetMetadata) -> Option<ExternalOpId> {
+        metadata
+            .actors
+            .get_safe(opid.actor())
+            .map(|actor| ExternalOpId {
+                counter: opid.counter(),
+                actor: actor.clone(),
+            })
+    }
+
+    pub(crate) fn counter(&self) -> u64 {
+        self.counter
+    }
+
+    pub(crate) fn actor(&self) -> &ActorId {
+        &self.actor
+    }
+}
+
+#[derive(Debug, PartialEq, Clone, Hash, Eq)]
+pub enum ExternalObjId<'a> {
+    Root,
+    Op(Cow<'a, ExternalOpId>),
+}
+
+impl<'a> ExternalObjId<'a> {
+    pub fn into_owned(self) -> ExternalObjId<'static> {
+        match self {
+            Self::Root => ExternalObjId::Root,
+            Self::Op(cow) => ExternalObjId::Op(Cow::<'static, _>::Owned(cow.into_owned().into())),
+        }
+    }
+}
+
+impl<'a> From<&'a ExternalOpId> for ExternalObjId<'a> {
+    fn from(op: &'a ExternalOpId) -> Self {
+        ExternalObjId::Op(Cow::Borrowed(op))
+    }
+}
+
+impl From<ExternalOpId> for ExternalObjId<'static> {
+    fn from(op: ExternalOpId) -> Self {
+        ExternalObjId::Op(Cow::Owned(op))
+    }
+}
+
+#[derive(thiserror::Error, Debug)]
+pub enum ParseError {
+    #[error("op IDs should have the format <counter>@<hex encoded actor>")]
+    BadFormat,
+    #[error("the counter of an opid should be a positive integer")]
+    InvalidCounter,
+    #[error("the actor of an opid should be valid hex encoded bytes")]
+    InvalidActor,
+}
+
+impl FromStr for ExternalOpId {
+    type Err = ParseError;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        let mut parts = s.split("@");
+        let first_part = parts.next().ok_or(ParseError::BadFormat)?;
+        let second_part = parts.next().ok_or(ParseError::BadFormat)?;
+        let counter: u64 = first_part.parse().map_err(|_| ParseError::InvalidCounter)?;
+        let actor: ActorId = second_part.parse().map_err(|_| ParseError::InvalidActor)?;
+        Ok(ExternalOpId { counter, actor })
+    }
+}
+
+impl FromStr for ExternalObjId<'static> {
+    type Err = ParseError;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        if s == ROOT_STR {
+            Ok(ExternalObjId::Root)
+        } else {
+            let op = s.parse::<ExternalOpId>()?.into();
+            Ok(ExternalObjId::Op(Cow::Owned(op)))
+        }
+    }
+}
+
+impl Display for ExternalOpId {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}@{}", self.counter, self.actor)
+    }
+}
+
+impl<'a> Display for ExternalObjId<'a> {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Self::Root => write!(f, "{}", ROOT_STR),
+            Self::Op(op) => write!(f, "{}", op),
+        }
+    }
+}
diff --git a/automerge/src/indexed_cache.rs b/automerge/src/indexed_cache.rs
index 21ffd75b..43a0fa2c 100644
--- a/automerge/src/indexed_cache.rs
+++ b/automerge/src/indexed_cache.rs
@@ -43,6 +43,11 @@ where
         &self.cache[index]
     }
 
+    // Todo replace all uses of `get` with this
+    pub fn get_safe(&self, index: usize) -> Option<&T> {
+        self.cache.get(index)
+    }
+
     pub fn sorted(&self) -> IndexedCache<T> {
         let mut sorted = Self::new();
         self.cache.iter().sorted().cloned().for_each(|item| {
diff --git a/automerge/src/lib.rs b/automerge/src/lib.rs
index c2595c68..dab21f23 100644
--- a/automerge/src/lib.rs
+++ b/automerge/src/lib.rs
@@ -40,28 +40,28 @@ mod sync;
 mod visualisation;
 
 mod error;
+mod external_types;
 mod op_set;
 mod op_tree;
 mod query;
 mod types;
 mod value;
+pub use external_types::{ExternalObjId as ObjId, ExternalOpId as OpId};
 
 use change::{encode_document, export_change};
 use clock::Clock;
 use indexed_cache::IndexedCache;
 use op_set::OpSet;
 use std::collections::{HashMap, HashSet, VecDeque};
-use types::{ElemId, Key, ObjId, Op, HEAD};
+use types::{ElemId, Key, ObjId as InternalObjId, Op, HEAD};
 use unicode_segmentation::UnicodeSegmentation;
 
 pub use change::{decode_change, Change};
 pub use error::AutomergeError;
 pub use legacy::Change as ExpandedChange;
 pub use sync::{BloomFilter, SyncHave, SyncMessage, SyncState};
-pub use types::{
-    ActorId, ChangeHash, Export, Exportable, Importable, ObjType, OpId, OpType, Patch, Peer, Prop,
-    ROOT,
-};
+pub use types::{ActorId, ChangeHash, ObjType, OpType, Patch, Peer, Prop};
+use types::{Export, Exportable, Importable, OpId as InternalOpId};
 pub use value::{ScalarValue, Value};
 
 #[derive(Debug, Clone)]
@@ -96,25 +96,25 @@ impl Automerge {
 
     pub fn set_actor(&mut self, actor: ActorId) {
         self.ensure_transaction_closed();
-        self.actor = Some(self.ops.m.actors.cache(actor))
+        self.actor = Some(self.ops.m.borrow_mut().actors.cache(actor))
     }
 
     fn random_actor(&mut self) -> ActorId {
         let actor = ActorId::from(uuid::Uuid::new_v4().as_bytes().to_vec());
-        self.actor = Some(self.ops.m.actors.cache(actor.clone()));
+        self.actor = Some(self.ops.m.borrow_mut().actors.cache(actor.clone()));
         actor
     }
 
     pub fn get_actor(&mut self) -> ActorId {
         if let Some(actor) = self.actor {
-            self.ops.m.actors[actor].clone()
+            self.ops.m.borrow().actors[actor].clone()
         } else {
             self.random_actor()
         }
     }
 
     pub fn maybe_get_actor(&self) -> Option<ActorId> {
-        self.actor.map(|i| self.ops.m.actors[i].clone())
+        self.actor.map(|i| self.ops.m.borrow().actors[i].clone())
     }
 
     fn get_actor_index(&mut self) -> usize {
@@ -139,7 +139,7 @@ impl Automerge {
             max_op: 0,
             transaction: None,
         };
-        am.actor = Some(am.ops.m.actors.cache(actor));
+        am.actor = Some(am.ops.m.borrow_mut().actors.cache(actor));
         am
     }
 
@@ -199,7 +199,9 @@ impl Automerge {
 
     pub fn ensure_transaction_closed(&mut self) {
         if let Some(tx) = self.transaction.take() {
-            self.update_history(export_change(&tx, &self.ops.m.actors, &self.ops.m.props));
+            let change =
+                export_change(&tx, &self.ops.m.borrow().actors, &self.ops.m.borrow().props);
+            self.update_history(change);
         }
     }
 
@@ -224,9 +226,9 @@ impl Automerge {
         }
     }
 
-    fn next_id(&mut self) -> OpId {
+    fn next_id(&mut self) -> InternalOpId {
         let tx = self.tx();
-        OpId(tx.start_op + tx.operations.len() as u64, tx.actor)
+        InternalOpId::new(tx.start_op + tx.operations.len() as u64, tx.actor)
     }
 
     fn insert_local_op(&mut self, op: Op, pos: usize, succ_pos: &[usize]) {
@@ -262,22 +264,26 @@ impl Automerge {
     // PropAt::()
     // NthAt::()
 
-    pub fn keys(&self, obj: OpId) -> Vec<String> {
+    pub fn keys<'a, O: Into<ObjId<'a>>>(&self, obj: O) -> Vec<String> {
+        let obj = self.import_objid(obj);
         let q = self.ops.search(obj.into(), query::Keys::new());
         q.keys.iter().map(|k| self.export(*k)).collect()
     }
 
-    pub fn keys_at(&self, obj: OpId, heads: &[ChangeHash]) -> Vec<String> {
+    pub fn keys_at<'a, O: Into<ObjId<'a>>>(&mut self, obj: O, heads: &[ChangeHash]) -> Vec<String> {
+        let obj = self.import_objid(obj.into());
         let clock = self.clock_at(heads);
         let q = self.ops.search(obj.into(), query::KeysAt::new(clock));
         q.keys.iter().map(|k| self.export(*k)).collect()
     }
 
-    pub fn length(&self, obj: OpId) -> usize {
+    pub fn length<'a, O: Into<ObjId<'a>>>(&self, obj: O) -> usize {
+        let obj = self.import_objid(obj.into());
         self.ops.search(obj.into(), query::Len::new(obj.into())).len
     }
 
-    pub fn length_at(&self, obj: OpId, heads: &[ChangeHash]) -> usize {
+    pub fn length_at<'a, O: Into<ObjId<'a>>>(&self, obj: O, heads: &[ChangeHash]) -> usize {
+        let obj = self.import_objid(obj.into());
         let clock = self.clock_at(heads);
         self.ops.search(obj.into(), query::LenAt::new(clock)).len
     }
@@ -300,23 +306,35 @@ impl Automerge {
     /// - The object does not exist
     /// - The key is the wrong type for the object
     /// - The key does not exist in the object
-    pub fn set<P: Into<Prop>, V: Into<Value>>(
+    pub fn set<'a, O: Into<ObjId<'a>>, P: Into<Prop>, V: Into<Value>>(
         &mut self,
-        obj: OpId,
+        obj: O,
         prop: P,
         value: V,
     ) -> Result<Option<OpId>, AutomergeError> {
         let value = value.into();
+        let obj = self.import_objid(obj.into());
         self.local_op(obj.into(), prop.into(), value.into())
+            .map(|o| o.map(|o| self.export_opid(&o).unwrap()))
     }
 
-    pub fn insert<V: Into<Value>>(
+    pub fn insert<'a, O: Into<ObjId<'a>>, V: Into<Value>>(
         &mut self,
-        obj: OpId,
+        obj: O,
         index: usize,
         value: V,
     ) -> Result<OpId, AutomergeError> {
-        let obj = obj.into();
+        let obj = self.import_objid(obj.into()).into();
+        let internal_op = self.insert_internal(obj, index, value)?;
+        Ok(self.export_opid(&internal_op).unwrap())
+    }
+
+    fn insert_internal<V: Into<Value>>(
+        &mut self,
+        obj: InternalObjId,
+        index: usize,
+        value: V,
+    ) -> Result<InternalOpId, AutomergeError> {
         let id = self.next_id();
 
         let query = self.ops.search(obj, query::InsertNth::new(index));
@@ -341,23 +359,41 @@ impl Automerge {
         Ok(id)
     }
 
-    pub fn inc<P: Into<Prop>>(
+    pub fn inc<'a, O: Into<ObjId<'a>>, P: Into<Prop>>(
         &mut self,
-        obj: OpId,
+        obj: O,
         prop: P,
         value: i64,
     ) -> Result<OpId, AutomergeError> {
-        match self.local_op(obj.into(), prop.into(), OpType::Inc(value))? {
-            Some(opid) => Ok(opid),
+        match self.local_op(
+            self.import_objid(obj.into()).into(),
+            prop.into(),
+            OpType::Inc(value),
+        )? {
+            Some(opid) => Ok(self.export_opid(&opid).unwrap()),
             None => {
                 panic!("increment should always create a new op")
             }
         }
     }
 
-    pub fn del<P: Into<Prop>>(&mut self, obj: OpId, prop: P) -> Result<OpId, AutomergeError> {
+    pub fn del<'a, O: Into<ObjId<'a>>, P: Into<Prop>>(
+        &mut self,
+        obj: O,
+        prop: P,
+    ) -> Result<OpId, AutomergeError> {
+        let obj = self.import_objid(obj);
+        self.del_internal(obj, prop)
+            .map(|o| self.export_opid(&o).unwrap())
+    }
+
+    fn del_internal<P: Into<Prop>>(
+        &mut self,
+        obj: InternalObjId,
+        prop: P,
+    ) -> Result<InternalOpId, AutomergeError> {
         // TODO: Should we also no-op multiple delete operations?
-        match self.local_op(obj.into(), prop.into(), OpType::Del)? {
+        match self.local_op(obj, prop.into(), OpType::Del)? {
             Some(opid) => Ok(opid),
             None => {
                 panic!("delete should always create a new op")
@@ -367,27 +403,32 @@ impl Automerge {
 
     /// Splice new elements into the given sequence. Returns a vector of the OpIds used to insert
     /// the new elements
-    pub fn splice(
+    pub fn splice<'a, O: Into<ObjId<'a>>>(
         &mut self,
-        obj: OpId,
+        obj: O,
         mut pos: usize,
         del: usize,
         vals: Vec<Value>,
     ) -> Result<Vec<OpId>, AutomergeError> {
+        let obj = self.import_objid(obj);
         for _ in 0..del {
-            self.del(obj, pos)?;
+            self.del_internal(obj, pos)?;
         }
         let mut result = Vec::with_capacity(vals.len());
         for v in vals {
-            result.push(self.insert(obj, pos, v)?);
+            result.push(self.insert_internal(obj, pos, v)?);
             pos += 1;
         }
-        Ok(result)
+        let exported = result
+            .into_iter()
+            .map(|o| self.export_opid(&o).unwrap())
+            .collect::<Vec<_>>();
+        Ok(exported)
     }
 
-    pub fn splice_text(
+    pub fn splice_text<'a, O: Into<ObjId<'a>>>(
         &mut self,
-        obj: OpId,
+        obj: O,
         pos: usize,
         del: usize,
         text: &str,
@@ -399,8 +440,8 @@ impl Automerge {
         self.splice(obj, pos, del, vals)
     }
 
-    pub fn text(&self, obj: OpId) -> Result<String, AutomergeError> {
-        let obj = obj.into();
+    pub fn text<'a, O: Into<ObjId<'a>>>(&self, obj: O) -> Result<String, AutomergeError> {
+        let obj = self.import_objid(obj.into()).into();
         let query = self.ops.search(obj, query::ListVals::new(obj));
         let mut buffer = String::new();
         for q in &query.ops {
@@ -411,9 +452,13 @@ impl Automerge {
         Ok(buffer)
     }
 
-    pub fn text_at(&self, obj: OpId, heads: &[ChangeHash]) -> Result<String, AutomergeError> {
+    pub fn text_at<'a, O: Into<ObjId<'a>>>(
+        &self,
+        obj: O,
+        heads: &[ChangeHash],
+    ) -> Result<String, AutomergeError> {
         let clock = self.clock_at(heads);
-        let obj = obj.into();
+        let obj = self.import_objid(obj.into()).into();
         let query = self.ops.search(obj, query::ListValsAt::new(clock));
         let mut buffer = String::new();
         for q in &query.ops {
@@ -427,38 +472,38 @@ impl Automerge {
     // TODO - I need to return these OpId's here **only** to get
     // the legacy conflicts format of { [opid]: value }
     // Something better?
-    pub fn value<P: Into<Prop>>(
+    pub fn value<'a, O: Into<ObjId<'a>>, P: Into<Prop>>(
         &self,
-        obj: OpId,
+        obj: O,
         prop: P,
     ) -> Result<Option<(Value, OpId)>, AutomergeError> {
         Ok(self.values(obj, prop.into())?.first().cloned())
     }
 
-    pub fn value_at<P: Into<Prop>>(
+    pub fn value_at<'a, O: Into<ObjId<'a>>, P: Into<Prop>>(
         &self,
-        obj: OpId,
+        obj: O,
         prop: P,
         heads: &[ChangeHash],
     ) -> Result<Option<(Value, OpId)>, AutomergeError> {
         Ok(self.values_at(obj, prop, heads)?.first().cloned())
     }
 
-    pub fn values<P: Into<Prop>>(
+    pub fn values<'a, O: Into<ObjId<'a>>, P: Into<Prop>>(
         &self,
-        obj: OpId,
+        obj: O,
         prop: P,
     ) -> Result<Vec<(Value, OpId)>, AutomergeError> {
-        let obj = obj.into();
+        let obj = self.import_objid(obj.into()).into();
         let result = match prop.into() {
             Prop::Map(p) => {
-                let prop = self.ops.m.props.lookup(p);
+                let prop = self.ops.m.borrow().props.lookup(p);
                 if let Some(p) = prop {
                     self.ops
                         .search(obj, query::Prop::new(obj, p))
                         .ops
                         .into_iter()
-                        .map(|o| o.into())
+                        .map(|o| self.labelled_value(&o))
                         .collect()
                 } else {
                     vec![]
@@ -469,30 +514,30 @@ impl Automerge {
                 .search(obj, query::Nth::new(n))
                 .ops
                 .into_iter()
-                .map(|o| o.into())
+                .map(|o| self.labelled_value(&o))
                 .collect(),
         };
         Ok(result)
     }
 
-    pub fn values_at<P: Into<Prop>>(
+    pub fn values_at<'a, O: Into<ObjId<'a>>, P: Into<Prop>>(
         &self,
-        obj: OpId,
+        obj: O,
         prop: P,
         heads: &[ChangeHash],
     ) -> Result<Vec<(Value, OpId)>, AutomergeError> {
         let prop = prop.into();
-        let obj = obj.into();
+        let obj = self.import_objid(obj).into();
         let clock = self.clock_at(heads);
         let result = match prop {
             Prop::Map(p) => {
-                let prop = self.ops.m.props.lookup(p);
+                let prop = self.ops.m.borrow().props.lookup(p);
                 if let Some(p) = prop {
                     self.ops
                         .search(obj, query::PropAt::new(p, clock))
                         .ops
                         .into_iter()
-                        .map(|o| o.into())
+                        .map(|o| self.labelled_value(&o))
                         .collect()
                 } else {
                     vec![]
@@ -503,7 +548,7 @@ impl Automerge {
                 .search(obj, query::NthAt::new(n, clock))
                 .ops
                 .into_iter()
-                .map(|o| o.into())
+                .map(|o| self.labelled_value(&o))
                 .collect(),
         };
         Ok(result)
@@ -552,10 +597,10 @@ impl Automerge {
 
     fn local_op(
         &mut self,
-        obj: ObjId,
+        obj: InternalObjId,
         prop: Prop,
         action: OpType,
-    ) -> Result<Option<OpId>, AutomergeError> {
+    ) -> Result<Option<InternalOpId>, AutomergeError> {
         match prop {
             Prop::Map(s) => self.local_map_op(obj, s, action),
             Prop::Seq(n) => self.local_list_op(obj, n, action),
@@ -564,16 +609,16 @@ impl Automerge {
 
     fn local_map_op(
         &mut self,
-        obj: ObjId,
+        obj: InternalObjId,
         prop: String,
         action: OpType,
-    ) -> Result<Option<OpId>, AutomergeError> {
+    ) -> Result<Option<InternalOpId>, AutomergeError> {
         if prop.is_empty() {
             return Err(AutomergeError::EmptyStringKey);
         }
 
         let id = self.next_id();
-        let prop = self.ops.m.props.cache(prop);
+        let prop = self.ops.m.borrow_mut().props.cache(prop);
         let query = self.ops.search(obj, query::Prop::new(obj, prop));
 
         match (&query.ops[..], &action) {
@@ -611,10 +656,10 @@ impl Automerge {
 
     fn local_list_op(
         &mut self,
-        obj: ObjId,
+        obj: InternalObjId,
         index: usize,
         action: OpType,
-    ) -> Result<Option<OpId>, AutomergeError> {
+    ) -> Result<Option<InternalOpId>, AutomergeError> {
         let query = self.ops.search(obj, query::Nth::new(index));
 
         let id = self.next_id();
@@ -675,17 +720,24 @@ impl Automerge {
             .iter_ops()
             .enumerate()
             .map(|(i, c)| {
-                let actor = self.ops.m.actors.cache(change.actor_id().clone());
-                let id = OpId(change.start_op + i as u64, actor);
+                let actor = self
+                    .ops
+                    .m
+                    .borrow_mut()
+                    .actors
+                    .cache(change.actor_id().clone());
+                let id = InternalOpId::new(change.start_op + i as u64, actor);
                 // FIXME dont need to_string()
-                let obj: ObjId = self.import(&c.obj.to_string()).unwrap();
+                let obj: InternalObjId = self.import(&c.obj.to_string()).unwrap();
                 let pred = c
                     .pred
                     .iter()
                     .map(|i| self.import(&i.to_string()).unwrap())
                     .collect();
                 let key = match &c.key {
-                    legacy::Key::Map(n) => Key::Map(self.ops.m.props.cache(n.to_string())),
+                    legacy::Key::Map(n) => {
+                        Key::Map(self.ops.m.borrow_mut().props.cache(n.to_string()))
+                    }
                     legacy::Key::Seq(legacy::ElementId::Head) => Key::Seq(HEAD),
                     // FIXME dont need to_string()
                     legacy::Key::Seq(legacy::ElementId::Id(i)) => {
@@ -727,8 +779,8 @@ impl Automerge {
         let bytes = encode_document(
             &c,
             ops.as_slice(),
-            &self.ops.m.actors,
-            &self.ops.m.props.cache,
+            &self.ops.m.borrow().actors,
+            &self.ops.m.borrow().props.cache,
         );
         if bytes.is_ok() {
             self.saved = self.get_heads().iter().copied().collect();
@@ -899,7 +951,7 @@ impl Automerge {
     pub fn get_last_local_change(&mut self) -> Option<&Change> {
         self.ensure_transaction_closed();
         if let Some(actor) = &self.actor {
-            let actor = &self.ops.m.actors[*actor];
+            let actor = &self.ops.m.borrow().actors[*actor];
             return self.history.iter().rev().find(|c| c.actor_id() == actor);
         }
         None
@@ -930,7 +982,13 @@ impl Automerge {
                         to_see.push(*h);
                     }
                 }
-                let actor = self.ops.m.actors.lookup(c.actor_id().clone()).unwrap();
+                let actor = self
+                    .ops
+                    .m
+                    .borrow()
+                    .actors
+                    .lookup(c.actor_id().clone())
+                    .unwrap();
                 clock.include(actor, c.max_op());
                 seen.insert(hash);
             }
@@ -1006,7 +1064,13 @@ impl Automerge {
         let history_index = self.history.len();
 
         self.states
-            .entry(self.ops.m.actors.cache(change.actor_id().clone()))
+            .entry(
+                self.ops
+                    .m
+                    .borrow_mut()
+                    .actors
+                    .cache(change.actor_id().clone()),
+            )
             .or_default()
             .push(history_index);
 
@@ -1023,7 +1087,7 @@ impl Automerge {
         self.deps.insert(change.hash);
     }
 
-    pub fn import<I: Importable>(&self, s: &str) -> Result<I, AutomergeError> {
+    pub(crate) fn import<I: Importable>(&self, s: &str) -> Result<I, AutomergeError> {
         if let Some(x) = I::from(s) {
             Ok(x)
         } else {
@@ -1037,17 +1101,40 @@ impl Automerge {
             let actor = self
                 .ops
                 .m
+                .borrow()
                 .actors
                 .lookup(actor)
                 .ok_or_else(|| AutomergeError::InvalidOpId(s.to_owned()))?;
-            Ok(I::wrap(OpId(counter, actor)))
+            Ok(I::wrap(InternalOpId::new(counter, actor)))
         }
     }
 
-    pub fn export<E: Exportable>(&self, id: E) -> String {
+    fn import_opid(&self, opid: &OpId) -> InternalOpId {
+        self.ops.m.borrow_mut().import_opid(opid)
+    }
+
+    fn export_opid(&self, opid: &InternalOpId) -> Option<OpId> {
+        OpId::from_internal(opid, &self.ops.m.borrow_mut())
+    }
+
+    fn import_objid<'a, A: Into<ObjId<'a>>>(&self, objid: A) -> InternalObjId {
+        match objid.into() {
+            ObjId::Root => InternalObjId::Root,
+            ObjId::Op(external_op) => {
+                let op = self.import_opid(&external_op);
+                InternalObjId::Op(op)
+            }
+        }
+    }
+
+    pub(crate) fn export<E: Exportable>(&self, id: E) -> String {
         match id.export() {
-            Export::Id(id) => format!("{}@{}", id.counter(), self.ops.m.actors[id.actor()]),
-            Export::Prop(index) => self.ops.m.props[index].clone(),
+            Export::Id(id) => format!(
+                "{}@{}",
+                id.counter(),
+                self.ops.m.borrow().actors[id.actor()]
+            ),
+            Export::Prop(index) => self.ops.m.borrow().props[index].clone(),
             Export::Special(s) => s,
         }
     }
@@ -1066,7 +1153,7 @@ impl Automerge {
             let id = self.export(i.id);
             let obj = self.export(i.obj);
             let key = match i.key {
-                Key::Map(n) => self.ops.m.props[n].clone(),
+                Key::Map(n) => self.ops.m.borrow().props[n].clone(),
                 Key::Seq(n) => self.export(n),
             };
             let value: String = match &i.action {
@@ -1093,6 +1180,16 @@ impl Automerge {
     pub fn visualise_optree(&self) -> String {
         self.ops.visualise()
     }
+
+    fn labelled_value(&self, op: &Op) -> (Value, OpId) {
+        let id = self.export_opid(&op.id).unwrap();
+        let value = match &op.action {
+            OpType::Make(obj_type) => Value::Object(*obj_type),
+            OpType::Set(scalar) => Value::Scalar(scalar.clone()),
+            _ => panic!("expected a make or set op"),
+        };
+        (value, id)
+    }
 }
 
 #[derive(Debug, Clone)]
@@ -1123,9 +1220,9 @@ mod tests {
     fn insert_op() -> Result<(), AutomergeError> {
         let mut doc = Automerge::new();
         doc.set_actor(ActorId::random());
-        doc.set(ROOT, "hello", "world")?;
+        doc.set(ObjId::Root, "hello", "world")?;
         assert!(doc.pending_ops() == 1);
-        doc.value(ROOT, "hello")?;
+        doc.value(ObjId::Root, "hello")?;
         Ok(())
     }
 
@@ -1133,17 +1230,20 @@ mod tests {
     fn test_list() -> Result<(), AutomergeError> {
         let mut doc = Automerge::new();
         doc.set_actor(ActorId::random());
-        let list_id = doc.set(ROOT, "items", Value::list())?.unwrap();
-        doc.set(ROOT, "zzz", "zzzval")?;
-        assert!(doc.value(ROOT, "items")?.unwrap().1 == list_id);
-        doc.insert(list_id, 0, "a")?;
-        doc.insert(list_id, 0, "b")?;
-        doc.insert(list_id, 2, "c")?;
-        doc.insert(list_id, 1, "d")?;
-        assert!(doc.value(list_id, 0)?.unwrap().0 == "b".into());
-        assert!(doc.value(list_id, 1)?.unwrap().0 == "d".into());
-        assert!(doc.value(list_id, 2)?.unwrap().0 == "a".into());
-        assert!(doc.value(list_id, 3)?.unwrap().0 == "c".into());
+        let list_id = doc
+            .set(ObjId::Root, "items", Value::list())?
+            .unwrap()
+            .into();
+        doc.set(ObjId::Root, "zzz", "zzzval")?;
+        assert!(doc.value(ObjId::Root, "items")?.unwrap().1 == list_id);
+        doc.insert(&list_id, 0, "a")?;
+        doc.insert(&list_id, 0, "b")?;
+        doc.insert(&list_id, 2, "c")?;
+        doc.insert(&list_id, 1, "d")?;
+        assert!(doc.value(&list_id, 0)?.unwrap().0 == "b".into());
+        assert!(doc.value(&list_id, 1)?.unwrap().0 == "d".into());
+        assert!(doc.value(&list_id, 2)?.unwrap().0 == "a".into());
+        assert!(doc.value(&list_id, 3)?.unwrap().0 == "c".into());
         assert!(doc.length(list_id) == 4);
         doc.save()?;
         Ok(())
@@ -1153,22 +1253,24 @@ mod tests {
     fn test_del() -> Result<(), AutomergeError> {
         let mut doc = Automerge::new();
         doc.set_actor(ActorId::random());
-        doc.set(ROOT, "xxx", "xxx")?;
-        assert!(!doc.values(ROOT, "xxx")?.is_empty());
-        doc.del(ROOT, "xxx")?;
-        assert!(doc.values(ROOT, "xxx")?.is_empty());
+        doc.set(ObjId::Root, "xxx", "xxx")?;
+        assert!(!doc.values(ObjId::Root, "xxx")?.is_empty());
+        doc.del(ObjId::Root, "xxx")?;
+        assert!(doc.values(ObjId::Root, "xxx")?.is_empty());
         Ok(())
     }
 
     #[test]
     fn test_inc() -> Result<(), AutomergeError> {
         let mut doc = Automerge::new();
-        let id = doc.set(ROOT, "counter", Value::counter(10))?.unwrap();
-        assert!(doc.value(ROOT, "counter")? == Some((Value::counter(10), id)));
-        doc.inc(ROOT, "counter", 10)?;
-        assert!(doc.value(ROOT, "counter")? == Some((Value::counter(20), id)));
-        doc.inc(ROOT, "counter", -5)?;
-        assert!(doc.value(ROOT, "counter")? == Some((Value::counter(15), id)));
+        let id = doc
+            .set(ObjId::Root, "counter", Value::counter(10))?
+            .unwrap();
+        assert!(doc.value(ObjId::Root, "counter")? == Some((Value::counter(10), id.clone())));
+        doc.inc(ObjId::Root, "counter", 10)?;
+        assert!(doc.value(ObjId::Root, "counter")? == Some((Value::counter(20), id.clone())));
+        doc.inc(ObjId::Root, "counter", -5)?;
+        assert!(doc.value(ObjId::Root, "counter")? == Some((Value::counter(15), id.clone())));
         Ok(())
     }
 
@@ -1176,15 +1278,15 @@ mod tests {
     fn test_save_incremental() -> Result<(), AutomergeError> {
         let mut doc = Automerge::new();
 
-        doc.set(ROOT, "foo", 1)?;
+        doc.set(ObjId::Root, "foo", 1)?;
 
         let save1 = doc.save().unwrap();
 
-        doc.set(ROOT, "bar", 2)?;
+        doc.set(ObjId::Root, "bar", 2)?;
 
         let save2 = doc.save_incremental();
 
-        doc.set(ROOT, "baz", 3)?;
+        doc.set(ObjId::Root, "baz", 3)?;
 
         let save3 = doc.save_incremental();
 
@@ -1202,7 +1304,7 @@ mod tests {
         let mut doc_a = Automerge::load(&save_a)?;
         let mut doc_b = Automerge::load(&save_b)?;
 
-        assert!(doc_a.values(ROOT, "baz")? == doc_b.values(ROOT, "baz")?);
+        assert!(doc_a.values(ObjId::Root, "baz")? == doc_b.values(ObjId::Root, "baz")?);
 
         assert!(doc_a.save().unwrap() == doc_b.save().unwrap());
 
@@ -1212,17 +1314,17 @@ mod tests {
     #[test]
     fn test_save_text() -> Result<(), AutomergeError> {
         let mut doc = Automerge::new();
-        let text = doc.set(ROOT, "text", Value::text())?.unwrap();
+        let text = doc.set(ObjId::Root, "text", Value::text())?.unwrap();
         let heads1 = doc.commit(None, None);
-        doc.splice_text(text, 0, 0, "hello world")?;
+        doc.splice_text(&text, 0, 0, "hello world")?;
         let heads2 = doc.commit(None, None);
-        doc.splice_text(text, 6, 0, "big bad ")?;
+        doc.splice_text(&text, 6, 0, "big bad ")?;
         let heads3 = doc.commit(None, None);
 
-        assert!(&doc.text(text)? == "hello big bad world");
-        assert!(&doc.text_at(text, &heads1)?.is_empty());
-        assert!(&doc.text_at(text, &heads2)? == "hello world");
-        assert!(&doc.text_at(text, &heads3)? == "hello big bad world");
+        assert!(&doc.text(&text)? == "hello big bad world");
+        assert!(&doc.text_at(&text, &heads1)?.is_empty());
+        assert!(&doc.text_at(&text, &heads2)? == "hello world");
+        assert!(&doc.text_at(&text, &heads3)? == "hello big bad world");
 
         Ok(())
     }
@@ -1231,50 +1333,50 @@ mod tests {
     fn test_props_vals_at() -> Result<(), AutomergeError> {
         let mut doc = Automerge::new();
         doc.set_actor("aaaa".try_into().unwrap());
-        doc.set(ROOT, "prop1", "val1")?;
+        doc.set(ObjId::Root, "prop1", "val1")?;
         doc.commit(None, None);
         let heads1 = doc.get_heads();
-        doc.set(ROOT, "prop1", "val2")?;
+        doc.set(ObjId::Root, "prop1", "val2")?;
         doc.commit(None, None);
         let heads2 = doc.get_heads();
-        doc.set(ROOT, "prop2", "val3")?;
+        doc.set(ObjId::Root, "prop2", "val3")?;
         doc.commit(None, None);
         let heads3 = doc.get_heads();
-        doc.del(ROOT, "prop1")?;
+        doc.del(ObjId::Root, "prop1")?;
         doc.commit(None, None);
         let heads4 = doc.get_heads();
-        doc.set(ROOT, "prop3", "val4")?;
+        doc.set(ObjId::Root, "prop3", "val4")?;
         doc.commit(None, None);
         let heads5 = doc.get_heads();
-        assert!(doc.keys_at(ROOT, &heads1) == vec!["prop1".to_owned()]);
-        assert!(doc.value_at(ROOT, "prop1", &heads1)?.unwrap().0 == Value::str("val1"));
-        assert!(doc.value_at(ROOT, "prop2", &heads1)? == None);
-        assert!(doc.value_at(ROOT, "prop3", &heads1)? == None);
+        assert!(doc.keys_at(ObjId::Root, &heads1) == vec!["prop1".to_owned()]);
+        assert!(doc.value_at(ObjId::Root, "prop1", &heads1)?.unwrap().0 == Value::str("val1"));
+        assert!(doc.value_at(ObjId::Root, "prop2", &heads1)? == None);
+        assert!(doc.value_at(ObjId::Root, "prop3", &heads1)? == None);
 
-        assert!(doc.keys_at(ROOT, &heads2) == vec!["prop1".to_owned()]);
-        assert!(doc.value_at(ROOT, "prop1", &heads2)?.unwrap().0 == Value::str("val2"));
-        assert!(doc.value_at(ROOT, "prop2", &heads2)? == None);
-        assert!(doc.value_at(ROOT, "prop3", &heads2)? == None);
+        assert!(doc.keys_at(ObjId::Root, &heads2) == vec!["prop1".to_owned()]);
+        assert!(doc.value_at(ObjId::Root, "prop1", &heads2)?.unwrap().0 == Value::str("val2"));
+        assert!(doc.value_at(ObjId::Root, "prop2", &heads2)? == None);
+        assert!(doc.value_at(ObjId::Root, "prop3", &heads2)? == None);
 
-        assert!(doc.keys_at(ROOT, &heads3) == vec!["prop1".to_owned(), "prop2".to_owned()]);
-        assert!(doc.value_at(ROOT, "prop1", &heads3)?.unwrap().0 == Value::str("val2"));
-        assert!(doc.value_at(ROOT, "prop2", &heads3)?.unwrap().0 == Value::str("val3"));
-        assert!(doc.value_at(ROOT, "prop3", &heads3)? == None);
+        assert!(doc.keys_at(ObjId::Root, &heads3) == vec!["prop1".to_owned(), "prop2".to_owned()]);
+        assert!(doc.value_at(ObjId::Root, "prop1", &heads3)?.unwrap().0 == Value::str("val2"));
+        assert!(doc.value_at(ObjId::Root, "prop2", &heads3)?.unwrap().0 == Value::str("val3"));
+        assert!(doc.value_at(ObjId::Root, "prop3", &heads3)? == None);
 
-        assert!(doc.keys_at(ROOT, &heads4) == vec!["prop2".to_owned()]);
-        assert!(doc.value_at(ROOT, "prop1", &heads4)? == None);
-        assert!(doc.value_at(ROOT, "prop2", &heads4)?.unwrap().0 == Value::str("val3"));
-        assert!(doc.value_at(ROOT, "prop3", &heads4)? == None);
+        assert!(doc.keys_at(ObjId::Root, &heads4) == vec!["prop2".to_owned()]);
+        assert!(doc.value_at(ObjId::Root, "prop1", &heads4)? == None);
+        assert!(doc.value_at(ObjId::Root, "prop2", &heads4)?.unwrap().0 == Value::str("val3"));
+        assert!(doc.value_at(ObjId::Root, "prop3", &heads4)? == None);
 
-        assert!(doc.keys_at(ROOT, &heads5) == vec!["prop2".to_owned(), "prop3".to_owned()]);
-        assert!(doc.value_at(ROOT, "prop1", &heads5)? == None);
-        assert!(doc.value_at(ROOT, "prop2", &heads5)?.unwrap().0 == Value::str("val3"));
-        assert!(doc.value_at(ROOT, "prop3", &heads5)?.unwrap().0 == Value::str("val4"));
+        assert!(doc.keys_at(ObjId::Root, &heads5) == vec!["prop2".to_owned(), "prop3".to_owned()]);
+        assert!(doc.value_at(ObjId::Root, "prop1", &heads5)? == None);
+        assert!(doc.value_at(ObjId::Root, "prop2", &heads5)?.unwrap().0 == Value::str("val3"));
+        assert!(doc.value_at(ObjId::Root, "prop3", &heads5)?.unwrap().0 == Value::str("val4"));
 
-        assert!(doc.keys_at(ROOT, &[]).is_empty());
-        assert!(doc.value_at(ROOT, "prop1", &[])? == None);
-        assert!(doc.value_at(ROOT, "prop2", &[])? == None);
-        assert!(doc.value_at(ROOT, "prop3", &[])? == None);
+        assert!(doc.keys_at(ObjId::Root, &[]).is_empty());
+        assert!(doc.value_at(ObjId::Root, "prop1", &[])? == None);
+        assert!(doc.value_at(ObjId::Root, "prop2", &[])? == None);
+        assert!(doc.value_at(ObjId::Root, "prop3", &[])? == None);
         Ok(())
     }
 
@@ -1283,47 +1385,47 @@ mod tests {
         let mut doc = Automerge::new();
         doc.set_actor("aaaa".try_into().unwrap());
 
-        let list = doc.set(ROOT, "list", Value::list())?.unwrap();
+        let list = doc.set(ObjId::Root, "list", Value::list())?.unwrap();
         let heads1 = doc.commit(None, None);
 
-        doc.insert(list, 0, Value::int(10))?;
+        doc.insert(&list, 0, Value::int(10))?;
         let heads2 = doc.commit(None, None);
 
-        doc.set(list, 0, Value::int(20))?;
-        doc.insert(list, 0, Value::int(30))?;
+        doc.set(&list, 0, Value::int(20))?;
+        doc.insert(&list, 0, Value::int(30))?;
         let heads3 = doc.commit(None, None);
 
-        doc.set(list, 1, Value::int(40))?;
-        doc.insert(list, 1, Value::int(50))?;
+        doc.set(&list, 1, Value::int(40))?;
+        doc.insert(&list, 1, Value::int(50))?;
         let heads4 = doc.commit(None, None);
 
-        doc.del(list, 2)?;
+        doc.del(&list, 2)?;
         let heads5 = doc.commit(None, None);
 
-        doc.del(list, 0)?;
+        doc.del(&list, 0)?;
         let heads6 = doc.commit(None, None);
 
-        assert!(doc.length_at(list, &heads1) == 0);
-        assert!(doc.value_at(list, 0, &heads1)?.is_none());
+        assert!(doc.length_at(&list, &heads1) == 0);
+        assert!(doc.value_at(&list, 0, &heads1)?.is_none());
 
-        assert!(doc.length_at(list, &heads2) == 1);
-        assert!(doc.value_at(list, 0, &heads2)?.unwrap().0 == Value::int(10));
+        assert!(doc.length_at(&list, &heads2) == 1);
+        assert!(doc.value_at(&list, 0, &heads2)?.unwrap().0 == Value::int(10));
 
-        assert!(doc.length_at(list, &heads3) == 2);
-        assert!(doc.value_at(list, 0, &heads3)?.unwrap().0 == Value::int(30));
-        assert!(doc.value_at(list, 1, &heads3)?.unwrap().0 == Value::int(20));
+        assert!(doc.length_at(&list, &heads3) == 2);
+        assert!(doc.value_at(&list, 0, &heads3)?.unwrap().0 == Value::int(30));
+        assert!(doc.value_at(&list, 1, &heads3)?.unwrap().0 == Value::int(20));
 
-        assert!(doc.length_at(list, &heads4) == 3);
-        assert!(doc.value_at(list, 0, &heads4)?.unwrap().0 == Value::int(30));
-        assert!(doc.value_at(list, 1, &heads4)?.unwrap().0 == Value::int(50));
-        assert!(doc.value_at(list, 2, &heads4)?.unwrap().0 == Value::int(40));
+        assert!(doc.length_at(&list, &heads4) == 3);
+        assert!(doc.value_at(&list, 0, &heads4)?.unwrap().0 == Value::int(30));
+        assert!(doc.value_at(&list, 1, &heads4)?.unwrap().0 == Value::int(50));
+        assert!(doc.value_at(&list, 2, &heads4)?.unwrap().0 == Value::int(40));
 
-        assert!(doc.length_at(list, &heads5) == 2);
-        assert!(doc.value_at(list, 0, &heads5)?.unwrap().0 == Value::int(30));
-        assert!(doc.value_at(list, 1, &heads5)?.unwrap().0 == Value::int(50));
+        assert!(doc.length_at(&list, &heads5) == 2);
+        assert!(doc.value_at(&list, 0, &heads5)?.unwrap().0 == Value::int(30));
+        assert!(doc.value_at(&list, 1, &heads5)?.unwrap().0 == Value::int(50));
 
-        assert!(doc.length_at(list, &heads6) == 1);
-        assert!(doc.value_at(list, 0, &heads6)?.unwrap().0 == Value::int(50));
+        assert!(doc.length_at(&list, &heads6) == 1);
+        assert!(doc.value_at(&list, 0, &heads6)?.unwrap().0 == Value::int(50));
 
         Ok(())
     }
diff --git a/automerge/src/op_set.rs b/automerge/src/op_set.rs
index 537cb80f..c814c069 100644
--- a/automerge/src/op_set.rs
+++ b/automerge/src/op_set.rs
@@ -1,9 +1,13 @@
 use crate::op_tree::OpTreeInternal;
 use crate::query::TreeQuery;
-use crate::{ActorId, IndexedCache, Key, ObjId, Op, OpId};
+use crate::{ActorId, IndexedCache, Key, types::{ObjId, OpId}, Op};
+use crate::external_types::ExternalOpId;
 use fxhash::FxBuildHasher;
 use std::cmp::Ordering;
 use std::collections::HashMap;
+use std::rc::Rc;
+use std::cell::RefCell;
+use std::fmt::Debug;
 
 pub(crate) type OpSet = OpSetInternal<16>;
 
@@ -12,7 +16,7 @@ pub(crate) struct OpSetInternal<const B: usize> {
     trees: HashMap<ObjId, OpTreeInternal<B>, FxBuildHasher>,
     objs: Vec<ObjId>,
     length: usize,
-    pub m: OpSetMetadata,
+    pub m: Rc<RefCell<OpSetMetadata>>,
 }
 
 impl<const B: usize> OpSetInternal<B> {
@@ -21,10 +25,11 @@ impl<const B: usize> OpSetInternal<B> {
             trees: Default::default(),
             objs: Default::default(),
             length: 0,
-            m: OpSetMetadata {
+            m: Rc::new(RefCell::new(OpSetMetadata {
                 actors: IndexedCache::new(),
                 props: IndexedCache::new(),
-            },
+                last_opid: None,
+            })),
         }
     }
 
@@ -41,7 +46,7 @@ impl<const B: usize> OpSetInternal<B> {
         Q: TreeQuery<B>,
     {
         if let Some(tree) = self.trees.get(&obj) {
-            tree.search(query, &self.m)
+            tree.search(query, &*self.m.borrow())
         } else {
             query
         }
@@ -83,7 +88,7 @@ impl<const B: usize> OpSetInternal<B> {
             .entry(element.obj)
             .or_insert_with(|| {
                 let pos = objs
-                    .binary_search_by(|probe| m.lamport_cmp(probe.0, element.obj.0))
+                    .binary_search_by(|probe| m.borrow().lamport_cmp(probe, &element.obj))
                     .unwrap_err();
                 objs.insert(pos, element.obj);
                 Default::default()
@@ -148,12 +153,16 @@ impl<'a, const B: usize> Iterator for Iter<'a, B> {
     }
 }
 
-#[derive(Clone, Debug)]
+#[derive(Debug, Clone)]
 pub(crate) struct OpSetMetadata {
     pub actors: IndexedCache<ActorId>,
     pub props: IndexedCache<String>,
+    // For the common case of many subsequent operations on the same object we cache the last
+    // object we looked up
+    last_opid: Option<(ExternalOpId, OpId)>,
 }
 
+
 impl OpSetMetadata {
     pub fn key_cmp(&self, left: &Key, right: &Key) -> Ordering {
         match (left, right) {
@@ -162,14 +171,54 @@ impl OpSetMetadata {
         }
     }
 
-    pub fn lamport_cmp(&self, left: OpId, right: OpId) -> Ordering {
-        match (left, right) {
-            (OpId(0, _), OpId(0, _)) => Ordering::Equal,
-            (OpId(0, _), OpId(_, _)) => Ordering::Less,
-            (OpId(_, _), OpId(0, _)) => Ordering::Greater,
-            // FIXME - this one seems backwards to me - why - is values() returning in the wrong order?
-            (OpId(a, x), OpId(b, y)) if a == b => self.actors[y].cmp(&self.actors[x]),
-            (OpId(a, _), OpId(b, _)) => a.cmp(&b),
+    pub fn lamport_cmp<S: SuccinctLamport>(&self, left: S, right: S) -> Ordering {
+        S::cmp(self, left, right)
+    }
+
+    pub fn import_opid(&mut self, ext_opid: &ExternalOpId) -> OpId {
+        if let Some((last_ext, last_int)) = &self.last_opid {
+            if last_ext == ext_opid {
+                return *last_int;
+            }
+        }
+        let actor = self.actors.cache(ext_opid.actor().clone());
+        let opid = OpId::new(ext_opid.counter(), actor);
+        self.last_opid = Some((ext_opid.clone(), opid));
+        opid
+    }
+}
+
+/// Lamport timestamps which don't contain their actor ID directly and therefore need access to
+/// some metadata to compare their actor ID parts
+pub(crate) trait SuccinctLamport {
+    fn cmp(m: &OpSetMetadata, left: Self, right: Self) -> Ordering;
+}
+
+impl SuccinctLamport for OpId {
+    fn cmp(m: &OpSetMetadata, left: Self, right: Self) -> Ordering {
+        match (left.counter(), right.counter()) {
+            (0, 0) => Ordering::Equal,
+            (0, _) => Ordering::Less,
+            (_, 0) => Ordering::Greater,
+            (a, b) if a == b => m.actors[right.actor()].cmp(&m.actors[left.actor()]),
+            (a, b) => a.cmp(&b),
         }
     }
 }
+
+impl SuccinctLamport for ObjId {
+    fn cmp(m: &OpSetMetadata, left: Self, right: Self) -> Ordering {
+        match (left, right) {
+            (ObjId::Root, ObjId::Root) => Ordering::Equal,
+            (ObjId::Root, ObjId::Op(_)) => Ordering::Less,
+            (ObjId::Op(_), ObjId::Root) => Ordering::Greater,
+            (ObjId::Op(left_op), ObjId::Op(right_op)) => <OpId as SuccinctLamport>::cmp(m, left_op, right_op),
+        }
+    }
+}
+
+impl SuccinctLamport for &ObjId {
+    fn cmp(m: &OpSetMetadata, left: Self, right: Self) -> Ordering {
+        <ObjId as SuccinctLamport>::cmp(m, *left, *right)
+    }
+}
diff --git a/automerge/src/op_tree.rs b/automerge/src/op_tree.rs
index 6142a7bf..a9d72353 100644
--- a/automerge/src/op_tree.rs
+++ b/automerge/src/op_tree.rs
@@ -6,7 +6,7 @@ use std::{
 
 pub(crate) use crate::op_set::OpSetMetadata;
 use crate::query::{Index, QueryResult, TreeQuery};
-use crate::{Op, OpId};
+use crate::types::{Op, OpId};
 use std::collections::HashSet;
 
 #[allow(dead_code)]
@@ -628,12 +628,12 @@ struct CounterData {
 #[cfg(test)]
 mod tests {
     use crate::legacy as amp;
-    use crate::{Op, OpId};
+    use crate::types::{Op, OpId};
 
     use super::*;
 
     fn op(n: usize) -> Op {
-        let zero = OpId(0, 0);
+        let zero = OpId::new(0, 0);
         Op {
             change: n,
             id: zero,
diff --git a/automerge/src/query.rs b/automerge/src/query.rs
index 15ac6fd6..662ad8ae 100644
--- a/automerge/src/query.rs
+++ b/automerge/src/query.rs
@@ -1,5 +1,5 @@
 use crate::op_tree::{OpSetMetadata, OpTreeNode};
-use crate::{Clock, ElemId, Op, OpId, OpType, ScalarValue};
+use crate::{Clock, ElemId, Op, ScalarValue, types::{OpId, OpType}};
 use fxhash::FxBuildHasher;
 use std::cmp::Ordering;
 use std::collections::{HashMap, HashSet};
diff --git a/automerge/src/query/len.rs b/automerge/src/query/len.rs
index 494b3515..b73e804a 100644
--- a/automerge/src/query/len.rs
+++ b/automerge/src/query/len.rs
@@ -1,6 +1,6 @@
 use crate::op_tree::OpTreeNode;
 use crate::query::{QueryResult, TreeQuery};
-use crate::ObjId;
+use crate::types::ObjId;
 use std::fmt::Debug;
 
 #[derive(Debug, Clone, PartialEq)]
diff --git a/automerge/src/query/list_vals.rs b/automerge/src/query/list_vals.rs
index c19ac4ad..aec8f878 100644
--- a/automerge/src/query/list_vals.rs
+++ b/automerge/src/query/list_vals.rs
@@ -1,6 +1,6 @@
 use crate::op_tree::{OpSetMetadata, OpTreeNode};
 use crate::query::{binary_search_by, is_visible, visible_op, QueryResult, TreeQuery};
-use crate::{ElemId, ObjId, Op};
+use crate::{ElemId, types::ObjId, Op};
 use std::fmt::Debug;
 
 #[derive(Debug, Clone, PartialEq)]
@@ -26,7 +26,7 @@ impl<const B: usize> TreeQuery<B> for ListVals {
         child: &OpTreeNode<B>,
         m: &OpSetMetadata,
     ) -> QueryResult {
-        let start = binary_search_by(child, |op| m.lamport_cmp(op.obj.0, self.obj.0));
+        let start = binary_search_by(child, |op| m.lamport_cmp(op.obj, self.obj));
         let mut counters = Default::default();
         for pos in start..child.len() {
             let op = child.get(pos).unwrap();
diff --git a/automerge/src/query/prop.rs b/automerge/src/query/prop.rs
index ac4b2bca..aa125c9d 100644
--- a/automerge/src/query/prop.rs
+++ b/automerge/src/query/prop.rs
@@ -1,6 +1,6 @@
 use crate::op_tree::{OpSetMetadata, OpTreeNode};
 use crate::query::{binary_search_by, is_visible, visible_op, QueryResult, TreeQuery};
-use crate::{Key, ObjId, Op};
+use crate::{Key, types::ObjId, Op};
 use std::fmt::Debug;
 
 #[derive(Debug, Clone, PartialEq)]
@@ -31,7 +31,7 @@ impl<const B: usize> TreeQuery<B> for Prop {
         m: &OpSetMetadata,
     ) -> QueryResult {
         let start = binary_search_by(child, |op| {
-            m.lamport_cmp(op.obj.0, self.obj.0)
+            m.lamport_cmp(op.obj, self.obj)
                 .then_with(|| m.key_cmp(&op.key, &self.key))
         });
         let mut counters = Default::default();
diff --git a/automerge/src/types.rs b/automerge/src/types.rs
index f00beed3..4adcf393 100644
--- a/automerge/src/types.rs
+++ b/automerge/src/types.rs
@@ -10,7 +10,6 @@ use std::str::FromStr;
 use tinyvec::{ArrayVec, TinyVec};
 
 pub(crate) const HEAD: ElemId = ElemId(OpId(0, 0));
-pub const ROOT: OpId = OpId(0, 0);
 
 const ROOT_STR: &str = "_root";
 const HEAD_STR: &str = "_head";
@@ -161,17 +160,17 @@ pub enum OpType {
 }
 
 #[derive(Debug)]
-pub enum Export {
+pub(crate) enum Export {
     Id(OpId),
     Special(String),
     Prop(usize),
 }
 
-pub trait Exportable {
+pub(crate) trait Exportable {
     fn export(&self) -> Export;
 }
 
-pub trait Importable {
+pub(crate) trait Importable {
     fn wrap(id: OpId) -> Self;
     fn from(s: &str) -> Option<Self>
     where
@@ -179,33 +178,32 @@ pub trait Importable {
 }
 
 impl OpId {
+    pub(crate) fn new(counter: u64, actor: usize) -> OpId {
+        OpId(counter, actor)
+    }
+
     #[inline]
     pub fn counter(&self) -> u64 {
         self.0
     }
     #[inline]
-    pub fn actor(&self) -> usize {
+    pub(crate) fn actor(&self) -> usize {
         self.1
     }
 }
 
 impl Exportable for ObjId {
     fn export(&self) -> Export {
-        if self.0 == ROOT {
-            Export::Special(ROOT_STR.to_owned())
-        } else {
-            Export::Id(self.0)
+        match self {
+            ObjId::Root => Export::Special(ROOT_STR.to_owned()),
+            ObjId::Op(o) => Export::Id(*o)
         }
     }
 }
 
 impl Exportable for &ObjId {
     fn export(&self) -> Export {
-        if self.0 == ROOT {
-            Export::Special(ROOT_STR.to_owned())
-        } else {
-            Export::Id(self.0)
-        }
+        (*self).export()
     }
 }
 
@@ -236,11 +234,11 @@ impl Exportable for Key {
 
 impl Importable for ObjId {
     fn wrap(id: OpId) -> Self {
-        ObjId(id)
+        ObjId::Op(id)
     }
     fn from(s: &str) -> Option<Self> {
         if s == ROOT_STR {
-            Some(ROOT.into())
+            Some(ObjId::Root)
         } else {
             None
         }
@@ -251,12 +249,8 @@ impl Importable for OpId {
     fn wrap(id: OpId) -> Self {
         id
     }
-    fn from(s: &str) -> Option<Self> {
-        if s == ROOT_STR {
-            Some(ROOT)
-        } else {
-            None
-        }
+    fn from(_s: &str) -> Option<Self> {
+        None
     }
 }
 
@@ -275,7 +269,10 @@ impl Importable for ElemId {
 
 impl From<OpId> for ObjId {
     fn from(o: OpId) -> Self {
-        ObjId(o)
+        match (o.counter(), o.actor()) {
+            (0,0) => ObjId::Root,
+            (_,_) => ObjId::Op(o),
+        }
     }
 }
 
@@ -352,10 +349,19 @@ impl Key {
 }
 
 #[derive(Debug, Clone, PartialOrd, Ord, Eq, PartialEq, Copy, Hash, Default)]
-pub struct OpId(pub u64, pub usize);
+pub(crate) struct OpId(u64, usize);
 
-#[derive(Debug, Clone, Copy, PartialOrd, Eq, PartialEq, Ord, Hash, Default)]
-pub(crate) struct ObjId(pub OpId);
+#[derive(Debug, Clone, Copy, PartialOrd, Eq, PartialEq, Ord, Hash)]
+pub(crate) enum ObjId{
+    Root,
+    Op(OpId),
+}
+
+impl Default for ObjId {
+    fn default() -> Self {
+        Self::Root
+    }
+}
 
 #[derive(Debug, Clone, Copy, PartialOrd, Eq, PartialEq, Ord, Hash, Default)]
 pub(crate) struct ElemId(pub OpId);
diff --git a/automerge/src/value.rs b/automerge/src/value.rs
index 333c1f53..2859073e 100644
--- a/automerge/src/value.rs
+++ b/automerge/src/value.rs
@@ -1,4 +1,4 @@
-use crate::{error, ObjType, Op, OpId, OpType};
+use crate::{error, ObjType, Op, types::OpId, OpType};
 use serde::{Deserialize, Serialize};
 use smol_str::SmolStr;
 use std::convert::TryFrom;
diff --git a/automerge/tests/helpers/mod.rs b/automerge/tests/helpers/mod.rs
index d93a211b..bbcad074 100644
--- a/automerge/tests/helpers/mod.rs
+++ b/automerge/tests/helpers/mod.rs
@@ -1,3 +1,5 @@
+use automerge::ObjId;
+
 use std::{collections::HashMap, convert::TryInto, hash::Hash};
 
 use serde::ser::{SerializeMap, SerializeSeq};
@@ -23,7 +25,7 @@ pub fn sorted_actors() -> (automerge::ActorId, automerge::ActorId) {
 
 /// This macro makes it easy to make assertions about a document. It is called with two arguments,
 /// the first is a reference to an `automerge::Automerge`, the second is an instance of
-/// `RealizedObject<ExportableOpId>`.
+/// `RealizedObject`.
 ///
 /// What - I hear you ask - is a `RealizedObject`? It's a fully hydrated version of the contents of
 /// an automerge document. You don't need to think about this too much though because you can
@@ -65,29 +67,17 @@ pub fn sorted_actors() -> (automerge::ActorId, automerge::ActorId) {
 ///     map!{
 ///         "field" => {
 ///             op1 => "one",
-///             op2.translate(&doc2) => "two"
+///             op2 => "two"
 ///         }
 ///     }
 /// );
 /// ```
-///
-/// ## Translating OpIds
-///
-/// One thing you may have noticed in the example above is the `op2.translate(&doc2)` call. What is
-/// that doing there? Well, the problem is that automerge OpIDs (in the current API) are specific
-/// to a document. Using an opid from one document in a different document will not work. Therefore
-/// this module defines an `OpIdExt` trait with a `translate` method on it. This method takes a
-/// document and converts the opid into something which knows how to be compared with opids from
-/// another document by using the document you pass to `translate`. Again, all you really need to
-/// know is that when constructing a document for comparison you should call `translate(fromdoc)`
-/// on opids which come from a document other than the one you pass to `assert_doc`.
 #[macro_export]
 macro_rules! assert_doc {
     ($doc: expr, $expected: expr) => {{
-        use $crate::helpers::{realize, ExportableOpId};
+        use $crate::helpers::realize;
         let realized = realize($doc);
-        let to_export: RealizedObject<ExportableOpId<'_>> = $expected.into();
-        let exported = to_export.export($doc);
+        let exported: RealizedObject = $expected.into();
         if realized != exported {
             let serde_right = serde_json::to_string_pretty(&realized).unwrap();
             let serde_left = serde_json::to_string_pretty(&exported).unwrap();
@@ -105,10 +95,9 @@ macro_rules! assert_doc {
 #[macro_export]
 macro_rules! assert_obj {
     ($doc: expr, $obj_id: expr, $prop: expr, $expected: expr) => {{
-        use $crate::helpers::{realize_prop, ExportableOpId};
+        use $crate::helpers::realize_prop;
         let realized = realize_prop($doc, $obj_id, $prop);
-        let to_export: RealizedObject<ExportableOpId<'_>> = $expected.into();
-        let exported = to_export.export($doc);
+        let exported: RealizedObject = $expected.into();
         if realized != exported {
             let serde_right = serde_json::to_string_pretty(&realized).unwrap();
             let serde_left = serde_json::to_string_pretty(&exported).unwrap();
@@ -145,9 +134,9 @@ macro_rules! map {
     (@inner { $($opid:expr => $value:expr),* }) => {
         {
             use std::collections::HashMap;
-            let mut inner: HashMap<ExportableOpId<'_>, RealizedObject<ExportableOpId<'_>>> = HashMap::new();
+            let mut inner: HashMap<ObjId, RealizedObject> = HashMap::new();
             $(
-                let _ = inner.insert($opid.into(), $value.into());
+                let _ = inner.insert(ObjId::from((&$opid)).into_owned(), $value.into());
             )*
             inner
         }
@@ -159,9 +148,8 @@ macro_rules! map {
     ($($key:expr => $inner:tt),*) => {
         {
             use std::collections::HashMap;
-            use crate::helpers::ExportableOpId;
             let _cap = map!(@count $($key),*);
-            let mut _map: HashMap<String, HashMap<ExportableOpId<'_>, RealizedObject<ExportableOpId<'_>>>> = ::std::collections::HashMap::with_capacity(_cap);
+            let mut _map: HashMap<String, HashMap<ObjId, RealizedObject>> = ::std::collections::HashMap::with_capacity(_cap);
             $(
                 let inner = map!(@inner $inner);
                 let _ = _map.insert($key.to_string(), inner);
@@ -194,9 +182,9 @@ macro_rules! list {
     (@inner { $($opid:expr => $value:expr),* }) => {
         {
             use std::collections::HashMap;
-            let mut inner: HashMap<ExportableOpId<'_>, RealizedObject<ExportableOpId<'_>>> = HashMap::new();
+            let mut inner: HashMap<ObjId, RealizedObject> = HashMap::new();
             $(
-                let _ = inner.insert($opid.into(), $value.into());
+                let _ = inner.insert(ObjId::from(&$opid).into_owned(), $value.into());
             )*
             inner
         }
@@ -204,9 +192,8 @@ macro_rules! list {
     ($($inner:tt,)+) => { list!($($inner),+) };
     ($($inner:tt),*) => {
         {
-            use crate::helpers::ExportableOpId;
             let _cap = list!(@count $($inner),*);
-            let mut _list: Vec<HashMap<ExportableOpId<'_>, RealizedObject<ExportableOpId<'_>>>> = Vec::new();
+            let mut _list: Vec<HashMap<ObjId, RealizedObject>> = Vec::new();
             $(
                 //println!("{}", stringify!($inner));
                 let inner = list!(@inner $inner);
@@ -217,26 +204,6 @@ macro_rules! list {
     }
 }
 
-/// Translate an op ID produced by one document to an op ID which can be understood by
-/// another
-///
-/// The current API of automerge exposes OpIds of the form (u64, usize) where the first component
-/// is the counter of an actors lamport timestamp and the second component is the index into an
-/// array of actor IDs stored by the document where the opid was generated. Obviously this is not
-/// portable between documents as the index of the actor array is unlikely to match between two
-/// documents. This function translates between the two representations.
-///
-/// At some point we will probably change the API to not be document specific but this function
-/// allows us to write tests first.
-pub fn translate_obj_id(
-    from: &automerge::Automerge,
-    to: &automerge::Automerge,
-    id: automerge::OpId,
-) -> automerge::OpId {
-    let exported = from.export(id);
-    to.import(&exported).unwrap()
-}
-
 pub fn mk_counter(value: i64) -> automerge::ScalarValue {
     automerge::ScalarValue::Counter(value)
 }
@@ -253,13 +220,13 @@ impl std::fmt::Display for ExportedOpId {
 /// A `RealizedObject` is a representation of all the current values in a document - including
 /// conflicts.
 #[derive(PartialEq, Debug)]
-pub enum RealizedObject<Oid: PartialEq + Eq + Hash> {
-    Map(HashMap<String, HashMap<Oid, RealizedObject<Oid>>>),
-    Sequence(Vec<HashMap<Oid, RealizedObject<Oid>>>),
+pub enum RealizedObject<'a> {
+    Map(HashMap<String, HashMap<ObjId<'a>, RealizedObject<'a>>>),
+    Sequence(Vec<HashMap<ObjId<'a>, RealizedObject<'a>>>),
     Value(automerge::ScalarValue),
 }
 
-impl serde::Serialize for RealizedObject<ExportedOpId> {
+impl serde::Serialize for RealizedObject<'static> {
     fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
     where
         S: serde::Serializer,
@@ -271,7 +238,7 @@ impl serde::Serialize for RealizedObject<ExportedOpId> {
                     let kvs_serded = kvs
                         .iter()
                         .map(|(opid, value)| (opid.to_string(), value))
-                        .collect::<HashMap<String, &RealizedObject<ExportedOpId>>>();
+                        .collect::<HashMap<String, &RealizedObject>>();
                     map_ser.serialize_entry(k, &kvs_serded)?;
                 }
                 map_ser.end()
@@ -282,7 +249,7 @@ impl serde::Serialize for RealizedObject<ExportedOpId> {
                     let kvs_serded = elem
                         .iter()
                         .map(|(opid, value)| (opid.to_string(), value))
-                        .collect::<HashMap<String, &RealizedObject<ExportedOpId>>>();
+                        .collect::<HashMap<String, &RealizedObject>>();
                     list_ser.serialize_element(&kvs_serded)?;
                 }
                 list_ser.end()
@@ -292,40 +259,40 @@ impl serde::Serialize for RealizedObject<ExportedOpId> {
     }
 }
 
-pub fn realize(doc: &automerge::Automerge) -> RealizedObject<ExportedOpId> {
-    realize_obj(doc, automerge::ROOT, automerge::ObjType::Map)
+pub fn realize<'a>(doc: &automerge::Automerge) -> RealizedObject<'a> {
+    realize_obj(doc, ObjId::Root, automerge::ObjType::Map)
 }
 
 pub fn realize_prop<P: Into<automerge::Prop>>(
     doc: &automerge::Automerge,
-    obj_id: automerge::OpId,
+    obj_id: automerge::ObjId,
     prop: P,
-) -> RealizedObject<ExportedOpId> {
+) -> RealizedObject<'static> {
     let (val, obj_id) = doc.value(obj_id, prop).unwrap().unwrap();
     match val {
-        automerge::Value::Object(obj_type) => realize_obj(doc, obj_id, obj_type),
+        automerge::Value::Object(obj_type) => realize_obj(doc, obj_id.into(), obj_type),
         automerge::Value::Scalar(v) => RealizedObject::Value(v),
     }
 }
 
 pub fn realize_obj(
     doc: &automerge::Automerge,
-    obj_id: automerge::OpId,
+    obj_id: automerge::ObjId,
     objtype: automerge::ObjType,
-) -> RealizedObject<ExportedOpId> {
+) -> RealizedObject<'static> {
     match objtype {
         automerge::ObjType::Map | automerge::ObjType::Table => {
             let mut result = HashMap::new();
-            for key in doc.keys(obj_id) {
-                result.insert(key.clone(), realize_values(doc, obj_id, key));
+            for key in doc.keys(obj_id.clone()) {
+                result.insert(key.clone(), realize_values(doc, obj_id.clone(), key));
             }
             RealizedObject::Map(result)
         }
         automerge::ObjType::List | automerge::ObjType::Text => {
-            let length = doc.length(obj_id);
+            let length = doc.length(obj_id.clone());
             let mut result = Vec::with_capacity(length);
             for i in 0..length {
-                result.push(realize_values(doc, obj_id, i));
+                result.push(realize_values(doc, obj_id.clone(), i));
             }
             RealizedObject::Sequence(result)
         }
@@ -334,55 +301,25 @@ pub fn realize_obj(
 
 fn realize_values<K: Into<automerge::Prop>>(
     doc: &automerge::Automerge,
-    obj_id: automerge::OpId,
+    obj_id: automerge::ObjId,
     key: K,
-) -> HashMap<ExportedOpId, RealizedObject<ExportedOpId>> {
-    let mut values_by_opid = HashMap::new();
+) -> HashMap<ObjId<'static>, RealizedObject<'static>> {
+    let mut values_by_objid: HashMap<ObjId, RealizedObject> = HashMap::new();
     for (value, opid) in doc.values(obj_id, key).unwrap() {
         let realized = match value {
-            automerge::Value::Object(objtype) => realize_obj(doc, opid, objtype),
+            automerge::Value::Object(objtype) => realize_obj(doc, opid.clone().into(), objtype),
             automerge::Value::Scalar(v) => RealizedObject::Value(v),
         };
-        let exported_opid = ExportedOpId(doc.export(opid));
-        values_by_opid.insert(exported_opid, realized);
+        values_by_objid.insert(opid.into(), realized);
     }
-    values_by_opid
+    values_by_objid
 }
 
-impl<'a> RealizedObject<ExportableOpId<'a>> {
-    pub fn export(self, doc: &automerge::Automerge) -> RealizedObject<ExportedOpId> {
-        match self {
-            Self::Map(kvs) => RealizedObject::Map(
-                kvs.into_iter()
-                    .map(|(k, v)| {
-                        (
-                            k,
-                            v.into_iter()
-                                .map(|(k, v)| (k.export(doc), v.export(doc)))
-                                .collect(),
-                        )
-                    })
-                    .collect(),
-            ),
-            Self::Sequence(values) => RealizedObject::Sequence(
-                values
-                    .into_iter()
-                    .map(|v| {
-                        v.into_iter()
-                            .map(|(k, v)| (k.export(doc), v.export(doc)))
-                            .collect()
-                    })
-                    .collect(),
-            ),
-            Self::Value(v) => RealizedObject::Value(v),
-        }
-    }
-}
 
-impl<'a, O: Into<ExportableOpId<'a>>, I: Into<RealizedObject<ExportableOpId<'a>>>>
-    From<HashMap<&str, HashMap<O, I>>> for RealizedObject<ExportableOpId<'a>>
+impl<'a, I: Into<RealizedObject<'a>>>
+    From<HashMap<&str, HashMap<ObjId<'a>, I>>> for RealizedObject<'a>
 {
-    fn from(values: HashMap<&str, HashMap<O, I>>) -> Self {
+    fn from(values: HashMap<&str, HashMap<ObjId<'a>, I>>) -> Self {
         let intoed = values
             .into_iter()
             .map(|(k, v)| {
@@ -396,107 +333,44 @@ impl<'a, O: Into<ExportableOpId<'a>>, I: Into<RealizedObject<ExportableOpId<'a>>
     }
 }
 
-impl<'a, O: Into<ExportableOpId<'a>>, I: Into<RealizedObject<ExportableOpId<'a>>>>
-    From<Vec<HashMap<O, I>>> for RealizedObject<ExportableOpId<'a>>
+impl<'a, I: Into<RealizedObject<'a>>>
+    From<Vec<HashMap<ObjId<'a>, I>>> for RealizedObject<'a>
 {
-    fn from(values: Vec<HashMap<O, I>>) -> Self {
+    fn from(values: Vec<HashMap<ObjId<'a>, I>>) -> Self {
         RealizedObject::Sequence(
             values
                 .into_iter()
-                .map(|v| v.into_iter().map(|(k, v)| (k.into(), v.into())).collect())
+                .map(|v| v.into_iter().map(|(k, v)| (k, v.into())).collect())
                 .collect(),
         )
     }
 }
 
-impl From<bool> for RealizedObject<ExportableOpId<'_>> {
+impl From<bool> for RealizedObject<'static> {
     fn from(b: bool) -> Self {
         RealizedObject::Value(b.into())
     }
 }
 
-impl From<usize> for RealizedObject<ExportableOpId<'_>> {
+impl From<usize> for RealizedObject<'static> {
     fn from(u: usize) -> Self {
         let v = u.try_into().unwrap();
         RealizedObject::Value(automerge::ScalarValue::Int(v))
     }
 }
 
-impl From<automerge::ScalarValue> for RealizedObject<ExportableOpId<'_>> {
+impl From<automerge::ScalarValue> for RealizedObject<'static> {
     fn from(s: automerge::ScalarValue) -> Self {
         RealizedObject::Value(s)
     }
 }
 
-impl From<&str> for RealizedObject<ExportableOpId<'_>> {
+impl From<&str> for RealizedObject<'static> {
     fn from(s: &str) -> Self {
         RealizedObject::Value(automerge::ScalarValue::Str(s.into()))
     }
 }
 
-#[derive(Eq, PartialEq, Hash)]
-pub enum ExportableOpId<'a> {
-    Native(automerge::OpId),
-    Translate(Translate<'a>),
-}
-
-impl<'a> ExportableOpId<'a> {
-    fn export(self, doc: &automerge::Automerge) -> ExportedOpId {
-        let oid = match self {
-            Self::Native(oid) => oid,
-            Self::Translate(Translate { from, opid }) => translate_obj_id(from, doc, opid),
-        };
-        ExportedOpId(doc.export(oid))
-    }
-}
-
-pub struct Translate<'a> {
-    from: &'a automerge::Automerge,
-    opid: automerge::OpId,
-}
-
-impl<'a> PartialEq for Translate<'a> {
-    fn eq(&self, other: &Self) -> bool {
-        self.from.maybe_get_actor().unwrap() == other.from.maybe_get_actor().unwrap()
-            && self.opid == other.opid
-    }
-}
-
-impl<'a> Eq for Translate<'a> {}
-
-impl<'a> Hash for Translate<'a> {
-    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
-        self.from.maybe_get_actor().unwrap().hash(state);
-        self.opid.hash(state);
-    }
-}
-
-pub trait OpIdExt {
-    fn native(self) -> ExportableOpId<'static>;
-    fn translate(self, doc: &automerge::Automerge) -> ExportableOpId<'_>;
-}
-
-impl OpIdExt for automerge::OpId {
-    /// Use this opid directly when exporting
-    fn native(self) -> ExportableOpId<'static> {
-        ExportableOpId::Native(self)
-    }
-
-    /// Translate this OpID from `doc` when exporting
-    fn translate(self, doc: &automerge::Automerge) -> ExportableOpId<'_> {
-        ExportableOpId::Translate(Translate {
-            from: doc,
-            opid: self,
-        })
-    }
-}
-
-impl From<automerge::OpId> for ExportableOpId<'_> {
-    fn from(oid: automerge::OpId) -> Self {
-        ExportableOpId::Native(oid)
-    }
-}
-
 /// Pretty print the contents of a document
 #[allow(dead_code)]
 pub fn pretty_print(doc: &automerge::Automerge) {
diff --git a/automerge/tests/test.rs b/automerge/tests/test.rs
index 8dcc51df..4be44b97 100644
--- a/automerge/tests/test.rs
+++ b/automerge/tests/test.rs
@@ -1,16 +1,16 @@
-use automerge::Automerge;
+use automerge::{Automerge, ObjId};
 
 mod helpers;
 #[allow(unused_imports)]
 use helpers::{
     mk_counter, new_doc, new_doc_with_actor, pretty_print, realize, realize_obj, sorted_actors,
-    translate_obj_id, OpIdExt, RealizedObject,
+    RealizedObject,
 };
 #[test]
 fn no_conflict_on_repeated_assignment() {
     let mut doc = Automerge::new();
-    doc.set(automerge::ROOT, "foo", 1).unwrap();
-    let op = doc.set(automerge::ROOT, "foo", 2).unwrap().unwrap();
+    doc.set(ObjId::Root, "foo", 1).unwrap();
+    let op = doc.set(ObjId::Root, "foo", 2).unwrap().unwrap();
     assert_doc!(
         &doc,
         map! {
@@ -22,19 +22,19 @@ fn no_conflict_on_repeated_assignment() {
 #[test]
 fn no_change_on_repeated_map_set() {
     let mut doc = new_doc();
-    doc.set(automerge::ROOT, "foo", 1).unwrap();
-    assert!(doc.set(automerge::ROOT, "foo", 1).unwrap().is_none());
+    doc.set(ObjId::Root, "foo", 1).unwrap();
+    assert!(doc.set(ObjId::Root, "foo", 1).unwrap().is_none());
 }
 
 #[test]
 fn no_change_on_repeated_list_set() {
     let mut doc = new_doc();
     let list_id = doc
-        .set(automerge::ROOT, "list", automerge::Value::list())
+        .set(ObjId::Root, "list", automerge::Value::list())
         .unwrap()
-        .unwrap();
-    doc.insert(list_id, 0, 1).unwrap();
-    doc.set(list_id, 0, 1).unwrap();
+        .unwrap().into();
+    doc.insert(&list_id, 0, 1).unwrap();
+    doc.set(&list_id, 0, 1).unwrap();
     assert!(doc.set(list_id, 0, 1).unwrap().is_none());
 }
 
@@ -42,25 +42,25 @@ fn no_change_on_repeated_list_set() {
 fn no_change_on_list_insert_followed_by_set_of_same_value() {
     let mut doc = new_doc();
     let list_id = doc
-        .set(automerge::ROOT, "list", automerge::Value::list())
+        .set(ObjId::Root, "list", automerge::Value::list())
         .unwrap()
         .unwrap();
-    doc.insert(list_id, 0, 1).unwrap();
-    assert!(doc.set(list_id, 0, 1).unwrap().is_none());
+    doc.insert(&list_id, 0, 1).unwrap();
+    assert!(doc.set(&list_id, 0, 1).unwrap().is_none());
 }
 
 #[test]
 fn repeated_map_assignment_which_resolves_conflict_not_ignored() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
-    doc1.set(automerge::ROOT, "field", 123).unwrap();
+    doc1.set(ObjId::Root, "field", 123).unwrap();
     doc2.merge(&mut doc1);
-    doc2.set(automerge::ROOT, "field", 456).unwrap();
-    doc1.set(automerge::ROOT, "field", 789).unwrap();
+    doc2.set(ObjId::Root, "field", 456).unwrap();
+    doc1.set(ObjId::Root, "field", 789).unwrap();
     doc1.merge(&mut doc2);
-    assert_eq!(doc1.values(automerge::ROOT, "field").unwrap().len(), 2);
+    assert_eq!(doc1.values(ObjId::Root, "field").unwrap().len(), 2);
 
-    let op = doc1.set(automerge::ROOT, "field", 123).unwrap().unwrap();
+    let op = doc1.set(ObjId::Root, "field", 123).unwrap().unwrap();
     assert_doc!(
         &doc1,
         map! {
@@ -76,15 +76,14 @@ fn repeated_list_assignment_which_resolves_conflict_not_ignored() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
     let list_id = doc1
-        .set(automerge::ROOT, "list", automerge::Value::list())
+        .set(ObjId::Root, "list", automerge::Value::list())
         .unwrap()
         .unwrap();
-    doc1.insert(list_id, 0, 123).unwrap();
+    doc1.insert(&list_id, 0, 123).unwrap();
     doc2.merge(&mut doc1);
-    let list_id_in_doc2 = translate_obj_id(&doc1, &doc2, list_id);
-    doc2.set(list_id_in_doc2, 0, 456).unwrap().unwrap();
+    doc2.set(&list_id, 0, 456).unwrap().unwrap();
     doc1.merge(&mut doc2);
-    let doc1_op = doc1.set(list_id, 0, 789).unwrap().unwrap();
+    let doc1_op = doc1.set(&list_id, 0, 789).unwrap().unwrap();
 
     assert_doc!(
         &doc1,
@@ -102,13 +101,13 @@ fn repeated_list_assignment_which_resolves_conflict_not_ignored() {
 fn list_deletion() {
     let mut doc = new_doc();
     let list_id = doc
-        .set(automerge::ROOT, "list", automerge::Value::list())
+        .set(ObjId::Root, "list", automerge::Value::list())
         .unwrap()
         .unwrap();
-    let op1 = doc.insert(list_id, 0, 123).unwrap();
-    doc.insert(list_id, 1, 456).unwrap();
-    let op3 = doc.insert(list_id, 2, 789).unwrap();
-    doc.del(list_id, 1).unwrap();
+    let op1 = doc.insert(&list_id, 0, 123).unwrap();
+    doc.insert(&list_id, 1, 456).unwrap();
+    let op3 = doc.insert(&list_id.clone(), 2, 789).unwrap();
+    doc.del(&list_id, 1).unwrap();
     assert_doc!(
         &doc,
         map! {
@@ -124,28 +123,28 @@ fn list_deletion() {
 fn merge_concurrent_map_prop_updates() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
-    let op1 = doc1.set(automerge::ROOT, "foo", "bar").unwrap().unwrap();
+    let op1 = doc1.set(ObjId::Root, "foo", "bar").unwrap().unwrap();
     let hello = doc2
-        .set(automerge::ROOT, "hello", "world")
+        .set(ObjId::Root, "hello", "world")
         .unwrap()
         .unwrap();
     doc1.merge(&mut doc2);
     assert_eq!(
-        doc1.value(automerge::ROOT, "foo").unwrap().unwrap().0,
+        doc1.value(ObjId::Root, "foo").unwrap().unwrap().0,
         "bar".into()
     );
     assert_doc!(
         &doc1,
         map! {
             "foo" => { op1 => "bar" },
-            "hello" => { hello.translate(&doc2) => "world" },
+            "hello" => { hello => "world" },
         }
     );
     doc2.merge(&mut doc1);
     assert_doc!(
         &doc2,
         map! {
-            "foo" => { op1.translate(&doc1) => "bar" },
+            "foo" => { op1 => "bar" },
             "hello" => { hello => "world" },
         }
     );
@@ -157,12 +156,12 @@ fn add_concurrent_increments_of_same_property() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
     let counter_id = doc1
-        .set(automerge::ROOT, "counter", mk_counter(0))
+        .set(ObjId::Root, "counter", mk_counter(0))
         .unwrap()
         .unwrap();
     doc2.merge(&mut doc1);
-    doc1.inc(automerge::ROOT, "counter", 1).unwrap();
-    doc2.inc(automerge::ROOT, "counter", 2).unwrap();
+    doc1.inc(ObjId::Root, "counter", 1).unwrap();
+    doc2.inc(ObjId::Root, "counter", 2).unwrap();
     doc1.merge(&mut doc2);
     assert_doc!(
         &doc1,
@@ -181,17 +180,17 @@ fn add_increments_only_to_preceeded_values() {
 
     // create a counter in doc1
     let doc1_counter_id = doc1
-        .set(automerge::ROOT, "counter", mk_counter(0))
+        .set(ObjId::Root, "counter", mk_counter(0))
         .unwrap()
         .unwrap();
-    doc1.inc(automerge::ROOT, "counter", 1).unwrap();
+    doc1.inc(ObjId::Root, "counter", 1).unwrap();
 
     // create a counter in doc2
     let doc2_counter_id = doc2
-        .set(automerge::ROOT, "counter", mk_counter(0))
+        .set(ObjId::Root, "counter", mk_counter(0))
         .unwrap()
         .unwrap();
-    doc2.inc(automerge::ROOT, "counter", 3).unwrap();
+    doc2.inc(ObjId::Root, "counter", 3).unwrap();
 
     // The two values should be conflicting rather than added
     doc1.merge(&mut doc2);
@@ -200,8 +199,8 @@ fn add_increments_only_to_preceeded_values() {
         &doc1,
         map! {
             "counter" => {
-                doc1_counter_id.native() => mk_counter(1),
-                doc2_counter_id.translate(&doc2) => mk_counter(3),
+                doc1_counter_id => mk_counter(1),
+                doc2_counter_id => mk_counter(3),
             }
         }
     );
@@ -211,8 +210,8 @@ fn add_increments_only_to_preceeded_values() {
 fn concurrent_updates_of_same_field() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
-    let set_one_opid = doc1.set(automerge::ROOT, "field", "one").unwrap().unwrap();
-    let set_two_opid = doc2.set(automerge::ROOT, "field", "two").unwrap().unwrap();
+    let set_one_opid = doc1.set(ObjId::Root, "field", "one").unwrap().unwrap();
+    let set_two_opid = doc2.set(ObjId::Root, "field", "two").unwrap().unwrap();
 
     doc1.merge(&mut doc2);
 
@@ -220,8 +219,8 @@ fn concurrent_updates_of_same_field() {
         &doc1,
         map! {
             "field" => {
-                set_one_opid.native() => "one",
-                set_two_opid.translate(&doc2) => "two",
+                set_one_opid => "one",
+                set_two_opid => "two",
             }
         }
     );
@@ -232,14 +231,13 @@ fn concurrent_updates_of_same_list_element() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
     let list_id = doc1
-        .set(automerge::ROOT, "birds", automerge::Value::list())
+        .set(ObjId::Root, "birds", automerge::Value::list())
         .unwrap()
         .unwrap();
-    doc1.insert(list_id, 0, "finch").unwrap();
+    doc1.insert(list_id.clone(), 0, "finch").unwrap();
     doc2.merge(&mut doc1);
-    let set_one_op = doc1.set(list_id, 0, "greenfinch").unwrap().unwrap();
-    let list_id_in_doc2 = translate_obj_id(&doc1, &doc2, list_id);
-    let set_op_two = doc2.set(list_id_in_doc2, 0, "goldfinch").unwrap().unwrap();
+    let set_one_op = doc1.set(&list_id, 0, "greenfinch").unwrap().unwrap();
+    let set_op_two = doc2.set(&list_id, 0, "goldfinch").unwrap().unwrap();
 
     doc1.merge(&mut doc2);
 
@@ -248,8 +246,8 @@ fn concurrent_updates_of_same_list_element() {
         map! {
             "birds" => {
                 list_id => list![{
-                    set_one_op.native() => "greenfinch",
-                    set_op_two.translate(&doc2) => "goldfinch",
+                    set_one_op => "greenfinch",
+                    set_op_two => "goldfinch",
                 }]
             }
         }
@@ -262,15 +260,15 @@ fn assignment_conflicts_of_different_types() {
     let mut doc2 = new_doc();
     let mut doc3 = new_doc();
     let op_one = doc1
-        .set(automerge::ROOT, "field", "string")
+        .set(ObjId::Root, "field", "string")
         .unwrap()
         .unwrap();
     let op_two = doc2
-        .set(automerge::ROOT, "field", automerge::Value::list())
+        .set(ObjId::Root, "field", automerge::Value::list())
         .unwrap()
         .unwrap();
     let op_three = doc3
-        .set(automerge::ROOT, "field", automerge::Value::map())
+        .set(ObjId::Root, "field", automerge::Value::map())
         .unwrap()
         .unwrap();
 
@@ -281,9 +279,9 @@ fn assignment_conflicts_of_different_types() {
         &doc1,
         map! {
             "field" => {
-                op_one.native() => "string",
-                op_two.translate(&doc2) => list!{},
-                op_three.translate(&doc3) => map!{},
+                op_one => "string",
+                op_two => list!{},
+                op_three => map!{},
             }
         }
     );
@@ -294,24 +292,24 @@ fn changes_within_conflicting_map_field() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
     let op_one = doc1
-        .set(automerge::ROOT, "field", "string")
+        .set(ObjId::Root, "field", "string")
         .unwrap()
         .unwrap();
     let map_id = doc2
-        .set(automerge::ROOT, "field", automerge::Value::map())
+        .set(ObjId::Root, "field", automerge::Value::map())
         .unwrap()
         .unwrap();
-    let set_in_doc2 = doc2.set(map_id, "innerKey", 42).unwrap().unwrap();
+    let set_in_doc2 = doc2.set(&map_id, "innerKey", 42).unwrap().unwrap();
     doc1.merge(&mut doc2);
 
     assert_doc!(
         &doc1,
         map! {
             "field" => {
-                op_one.native() => "string",
-                map_id.translate(&doc2) => map!{
+                op_one => "string",
+                map_id => map!{
                     "innerKey" => {
-                        set_in_doc2.translate(&doc2) => 42,
+                        set_in_doc2 => 42,
                     }
                 }
             }
@@ -325,27 +323,26 @@ fn changes_within_conflicting_list_element() {
     let mut doc1 = new_doc_with_actor(actor1);
     let mut doc2 = new_doc_with_actor(actor2);
     let list_id = doc1
-        .set(automerge::ROOT, "list", automerge::Value::list())
+        .set(ObjId::Root, "list", automerge::Value::list())
         .unwrap()
         .unwrap();
-    doc1.insert(list_id, 0, "hello").unwrap();
+    doc1.insert(&list_id, 0, "hello").unwrap();
     doc2.merge(&mut doc1);
 
     let map_in_doc1 = doc1
-        .set(list_id, 0, automerge::Value::map())
+        .set(&list_id, 0, automerge::Value::map())
         .unwrap()
         .unwrap();
-    let set_map1 = doc1.set(map_in_doc1, "map1", true).unwrap().unwrap();
-    let set_key1 = doc1.set(map_in_doc1, "key", 1).unwrap().unwrap();
+    let set_map1 = doc1.set(&map_in_doc1, "map1", true).unwrap().unwrap();
+    let set_key1 = doc1.set(&map_in_doc1, "key", 1).unwrap().unwrap();
 
-    let list_id_in_doc2 = translate_obj_id(&doc1, &doc2, list_id);
     let map_in_doc2 = doc2
-        .set(list_id_in_doc2, 0, automerge::Value::map())
+        .set(&list_id, 0, automerge::Value::map())
         .unwrap()
         .unwrap();
     doc1.merge(&mut doc2);
-    let set_map2 = doc2.set(map_in_doc2, "map2", true).unwrap().unwrap();
-    let set_key2 = doc2.set(map_in_doc2, "key", 2).unwrap().unwrap();
+    let set_map2 = doc2.set(&map_in_doc2, "map2", true).unwrap().unwrap();
+    let set_key2 = doc2.set(&map_in_doc2, "key", 2).unwrap().unwrap();
 
     doc1.merge(&mut doc2);
 
@@ -355,13 +352,13 @@ fn changes_within_conflicting_list_element() {
             "list" => {
                 list_id => list![
                     {
-                        map_in_doc2.translate(&doc2) => map!{
-                            "map2" => { set_map2.translate(&doc2) => true },
-                            "key" => { set_key2.translate(&doc2) => 2 },
+                        map_in_doc2 => map!{
+                            "map2" => { set_map2 => true },
+                            "key" => { set_key2 => 2 },
                         },
-                        map_in_doc1.native() => map!{
-                            "key" => { set_key1.native() => 1 },
-                            "map1" => { set_map1.native() => true },
+                        map_in_doc1 => map!{
+                            "key" => { set_key1 => 1 },
+                            "map1" => { set_map1 => true },
                         }
                     }
                 ]
@@ -376,20 +373,20 @@ fn concurrently_assigned_nested_maps_should_not_merge() {
     let mut doc2 = new_doc();
 
     let doc1_map_id = doc1
-        .set(automerge::ROOT, "config", automerge::Value::map())
+        .set(ObjId::Root, "config", automerge::Value::map())
         .unwrap()
         .unwrap();
     let doc1_field = doc1
-        .set(doc1_map_id, "background", "blue")
+        .set(doc1_map_id.clone(), "background", "blue")
         .unwrap()
         .unwrap();
 
     let doc2_map_id = doc2
-        .set(automerge::ROOT, "config", automerge::Value::map())
+        .set(ObjId::Root, "config", automerge::Value::map())
         .unwrap()
         .unwrap();
     let doc2_field = doc2
-        .set(doc2_map_id, "logo_url", "logo.png")
+        .set(doc2_map_id.clone(), "logo_url", "logo.png")
         .unwrap()
         .unwrap();
 
@@ -399,11 +396,11 @@ fn concurrently_assigned_nested_maps_should_not_merge() {
         &doc1,
         map! {
             "config" => {
-                doc1_map_id.native() => map!{
-                    "background" => {doc1_field.native() => "blue"}
+                doc1_map_id => map!{
+                    "background" => {doc1_field => "blue"}
                 },
-                doc2_map_id.translate(&doc2) => map!{
-                    "logo_url" => {doc2_field.translate(&doc2) => "logo.png"}
+                doc2_map_id => map!{
+                    "logo_url" => {doc2_field => "logo.png"}
                 }
             }
         }
@@ -418,16 +415,15 @@ fn concurrent_insertions_at_different_list_positions() {
     assert!(doc1.maybe_get_actor().unwrap() < doc2.maybe_get_actor().unwrap());
 
     let list_id = doc1
-        .set(automerge::ROOT, "list", automerge::Value::list())
+        .set(ObjId::Root, "list", automerge::Value::list())
         .unwrap()
         .unwrap();
 
-    let one = doc1.insert(list_id, 0, "one").unwrap();
-    let three = doc1.insert(list_id, 1, "three").unwrap();
+    let one = doc1.insert(&list_id, 0, "one").unwrap();
+    let three = doc1.insert(&list_id, 1, "three").unwrap();
     doc2.merge(&mut doc1);
-    let two = doc1.splice(list_id, 1, 0, vec!["two".into()]).unwrap()[0];
-    let list_id_in_doc2 = translate_obj_id(&doc1, &doc2, list_id);
-    let four = doc2.insert(list_id_in_doc2, 2, "four").unwrap();
+    let two = doc1.splice(&list_id, 1, 0, vec!["two".into()]).unwrap()[0].clone();
+    let four = doc2.insert(&list_id, 2, "four").unwrap();
 
     doc1.merge(&mut doc2);
 
@@ -436,10 +432,10 @@ fn concurrent_insertions_at_different_list_positions() {
         map! {
             "list" => {
                 list_id => list![
-                    {one.native() => "one"},
-                    {two.native() => "two"},
-                    {three.native() => "three"},
-                    {four.translate(&doc2) => "four"},
+                    {one => "one"},
+                    {two => "two"},
+                    {three => "three"},
+                    {four => "four"},
                 ]
             }
         }
@@ -454,15 +450,14 @@ fn concurrent_insertions_at_same_list_position() {
     assert!(doc1.maybe_get_actor().unwrap() < doc2.maybe_get_actor().unwrap());
 
     let list_id = doc1
-        .set(automerge::ROOT, "birds", automerge::Value::list())
+        .set(ObjId::Root, "birds", automerge::Value::list())
         .unwrap()
         .unwrap();
-    let parakeet = doc1.insert(list_id, 0, "parakeet").unwrap();
+    let parakeet = doc1.insert(&list_id, 0, "parakeet").unwrap();
 
     doc2.merge(&mut doc1);
-    let list_id_in_doc2 = translate_obj_id(&doc1, &doc2, list_id);
-    let starling = doc1.insert(list_id, 1, "starling").unwrap();
-    let chaffinch = doc2.insert(list_id_in_doc2, 1, "chaffinch").unwrap();
+    let starling = doc1.insert(&list_id, 1, "starling").unwrap();
+    let chaffinch = doc2.insert(&list_id, 1, "chaffinch").unwrap();
     doc1.merge(&mut doc2);
 
     assert_doc!(
@@ -471,13 +466,13 @@ fn concurrent_insertions_at_same_list_position() {
             "birds" => {
                 list_id => list![
                     {
-                        parakeet.native() => "parakeet",
+                        parakeet => "parakeet",
                     },
                     {
-                        starling.native() => "starling",
+                        starling => "starling",
                     },
                     {
-                        chaffinch.translate(&doc2) => "chaffinch",
+                        chaffinch => "chaffinch",
                     },
                 ]
             },
@@ -489,11 +484,11 @@ fn concurrent_insertions_at_same_list_position() {
 fn concurrent_assignment_and_deletion_of_a_map_entry() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
-    doc1.set(automerge::ROOT, "bestBird", "robin").unwrap();
+    doc1.set(ObjId::Root, "bestBird", "robin").unwrap();
     doc2.merge(&mut doc1);
-    doc1.del(automerge::ROOT, "bestBird").unwrap();
+    doc1.del(ObjId::Root, "bestBird").unwrap();
     let set_two = doc2
-        .set(automerge::ROOT, "bestBird", "magpie")
+        .set(ObjId::Root, "bestBird", "magpie")
         .unwrap()
         .unwrap();
 
@@ -503,7 +498,7 @@ fn concurrent_assignment_and_deletion_of_a_map_entry() {
         &doc1,
         map! {
             "bestBird" => {
-                set_two.translate(&doc2) => "magpie",
+                set_two => "magpie",
             }
         }
     );
@@ -514,25 +509,24 @@ fn concurrent_assignment_and_deletion_of_list_entry() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
     let list_id = doc1
-        .set(automerge::ROOT, "birds", automerge::Value::list())
+        .set(ObjId::Root, "birds", automerge::Value::list())
         .unwrap()
         .unwrap();
-    let blackbird = doc1.insert(list_id, 0, "blackbird").unwrap();
-    doc1.insert(list_id, 1, "thrush").unwrap();
-    let goldfinch = doc1.insert(list_id, 2, "goldfinch").unwrap();
+    let blackbird = doc1.insert(&list_id, 0, "blackbird").unwrap();
+    doc1.insert(&list_id, 1, "thrush").unwrap();
+    let goldfinch = doc1.insert(&list_id, 2, "goldfinch").unwrap();
     doc2.merge(&mut doc1);
 
-    let starling = doc1.set(list_id, 1, "starling").unwrap().unwrap();
+    let starling = doc1.set(&list_id, 1, "starling").unwrap().unwrap();
 
-    let list_id_in_doc2 = translate_obj_id(&doc1, &doc2, list_id);
-    doc2.del(list_id_in_doc2, 1).unwrap();
+    doc2.del(&list_id, 1).unwrap();
 
     assert_doc!(
         &doc2,
         map! {
-            "birds" => {list_id.translate(&doc1) => list![
-                { blackbird.translate(&doc1) => "blackbird"},
-                { goldfinch.translate(&doc1) => "goldfinch"},
+            "birds" => {list_id => list![
+                { blackbird => "blackbird"},
+                { goldfinch => "goldfinch"},
             ]}
         }
     );
@@ -540,9 +534,9 @@ fn concurrent_assignment_and_deletion_of_list_entry() {
     assert_doc!(
         &doc1,
         map! {
-            "birds" => {list_id => list![
+            "birds" => {list_id.clone() => list![
                 { blackbird => "blackbird" },
-                { starling => "starling" },
+                { starling.clone() => "starling" },
                 { goldfinch => "goldfinch" },
             ]}
         }
@@ -567,22 +561,21 @@ fn insertion_after_a_deleted_list_element() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
     let list_id = doc1
-        .set(automerge::ROOT, "birds", automerge::Value::list())
+        .set(ObjId::Root, "birds", automerge::Value::list())
         .unwrap()
         .unwrap();
 
-    let blackbird = doc1.insert(list_id, 0, "blackbird").unwrap();
-    doc1.insert(list_id, 1, "thrush").unwrap();
-    doc1.insert(list_id, 2, "goldfinch").unwrap();
+    let blackbird = doc1.insert(list_id.clone(), 0, "blackbird").unwrap();
+    doc1.insert(&list_id, 1, "thrush").unwrap();
+    doc1.insert(&list_id, 2, "goldfinch").unwrap();
 
     doc2.merge(&mut doc1);
 
-    doc1.splice(list_id, 1, 2, Vec::new()).unwrap();
+    doc1.splice(&list_id, 1, 2, Vec::new()).unwrap();
 
-    let list_id_in_doc2 = translate_obj_id(&doc1, &doc2, list_id);
     let starling = doc2
-        .splice(list_id_in_doc2, 2, 0, vec!["starling".into()])
-        .unwrap()[0];
+        .splice(&list_id, 2, 0, vec!["starling".into()])
+        .unwrap()[0].clone();
 
     doc1.merge(&mut doc2);
 
@@ -590,8 +583,8 @@ fn insertion_after_a_deleted_list_element() {
         &doc1,
         map! {
             "birds" => {list_id => list![
-                { blackbird.native() => "blackbird" },
-                { starling.translate(&doc2) => "starling" }
+                { blackbird => "blackbird" },
+                { starling => "starling" }
             ]}
         }
     );
@@ -600,9 +593,9 @@ fn insertion_after_a_deleted_list_element() {
     assert_doc!(
         &doc2,
         map! {
-            "birds" => {list_id.translate(&doc1) => list![
-                { blackbird.translate(&doc1) => "blackbird" },
-                { starling.native() => "starling" }
+            "birds" => {list_id => list![
+                { blackbird => "blackbird" },
+                { starling => "starling" }
             ]}
         }
     );
@@ -613,29 +606,28 @@ fn concurrent_deletion_of_same_list_element() {
     let mut doc1 = new_doc();
     let mut doc2 = new_doc();
     let list_id = doc1
-        .set(automerge::ROOT, "birds", automerge::Value::list())
+        .set(ObjId::Root, "birds", automerge::Value::list())
         .unwrap()
         .unwrap();
 
-    let albatross = doc1.insert(list_id, 0, "albatross").unwrap();
-    doc1.insert(list_id, 1, "buzzard").unwrap();
-    let cormorant = doc1.insert(list_id, 2, "cormorant").unwrap();
+    let albatross = doc1.insert(list_id.clone(), 0, "albatross").unwrap();
+    doc1.insert(&list_id, 1, "buzzard").unwrap();
+    let cormorant = doc1.insert(&list_id, 2, "cormorant").unwrap();
 
     doc2.merge(&mut doc1);
 
-    doc1.del(list_id, 1).unwrap();
+    doc1.del(&list_id, 1).unwrap();
 
-    let list_id_in_doc2 = translate_obj_id(&doc1, &doc2, list_id);
-    doc2.del(list_id_in_doc2, 1).unwrap();
+    doc2.del(&list_id, 1).unwrap();
 
     doc1.merge(&mut doc2);
 
     assert_doc!(
         &doc1,
         map! {
-            "birds" => {list_id => list![
-                { albatross => "albatross" },
-                { cormorant  => "cormorant" }
+            "birds" => {list_id.clone() => list![
+                { albatross.clone() => "albatross" },
+                { cormorant.clone()  => "cormorant" }
             ]}
         }
     );
@@ -644,9 +636,9 @@ fn concurrent_deletion_of_same_list_element() {
     assert_doc!(
         &doc2,
         map! {
-            "birds" => {list_id.translate(&doc1) => list![
-                { albatross.translate(&doc1) => "albatross" },
-                { cormorant.translate(&doc1) => "cormorant" }
+            "birds" => {list_id => list![
+                { albatross => "albatross" },
+                { cormorant => "cormorant" }
             ]}
         }
     );
@@ -658,33 +650,32 @@ fn concurrent_updates_at_different_levels() {
     let mut doc2 = new_doc();
 
     let animals = doc1
-        .set(automerge::ROOT, "animals", automerge::Value::map())
+        .set(ObjId::Root, "animals", automerge::Value::map())
         .unwrap()
         .unwrap();
     let birds = doc1
-        .set(animals, "birds", automerge::Value::map())
+        .set(&animals, "birds", automerge::Value::map())
         .unwrap()
         .unwrap();
-    doc1.set(birds, "pink", "flamingo").unwrap().unwrap();
-    doc1.set(birds, "black", "starling").unwrap().unwrap();
+    doc1.set(&birds, "pink", "flamingo").unwrap().unwrap();
+    doc1.set(&birds, "black", "starling").unwrap().unwrap();
 
     let mammals = doc1
-        .set(animals, "mammals", automerge::Value::list())
+        .set(&animals, "mammals", automerge::Value::list())
         .unwrap()
         .unwrap();
-    let badger = doc1.insert(mammals, 0, "badger").unwrap();
+    let badger = doc1.insert(&mammals, 0, "badger").unwrap();
 
     doc2.merge(&mut doc1);
 
-    doc1.set(birds, "brown", "sparrow").unwrap().unwrap();
+    doc1.set(&birds, "brown", "sparrow").unwrap().unwrap();
 
-    let animals_in_doc2 = translate_obj_id(&doc1, &doc2, animals);
-    doc2.del(animals_in_doc2, "birds").unwrap();
+    doc2.del(&animals, "birds").unwrap();
     doc1.merge(&mut doc2);
 
     assert_obj!(
         &doc1,
-        automerge::ROOT,
+        ObjId::Root,
         "animals",
         map! {
             "mammals" => {
@@ -695,11 +686,11 @@ fn concurrent_updates_at_different_levels() {
 
     assert_obj!(
         &doc2,
-        automerge::ROOT,
+        ObjId::Root,
         "animals",
         map! {
             "mammals" => {
-                mammals.translate(&doc1) => list![{ badger.translate(&doc1) => "badger" }],
+                mammals => list![{ badger => "badger" }],
             }
         }
     );
@@ -711,21 +702,20 @@ fn concurrent_updates_of_concurrently_deleted_objects() {
     let mut doc2 = new_doc();
 
     let birds = doc1
-        .set(automerge::ROOT, "birds", automerge::Value::map())
+        .set(ObjId::Root, "birds", automerge::Value::map())
         .unwrap()
         .unwrap();
     let blackbird = doc1
-        .set(birds, "blackbird", automerge::Value::map())
+        .set(&birds, "blackbird", automerge::Value::map())
         .unwrap()
         .unwrap();
-    doc1.set(blackbird, "feathers", "black").unwrap().unwrap();
+    doc1.set(&blackbird, "feathers", "black").unwrap().unwrap();
 
     doc2.merge(&mut doc1);
 
-    doc1.del(birds, "blackbird").unwrap();
+    doc1.del(&birds, "blackbird").unwrap();
 
-    translate_obj_id(&doc1, &doc2, blackbird);
-    doc2.set(blackbird, "beak", "orange").unwrap();
+    doc2.set(&blackbird, "beak", "orange").unwrap();
 
     doc1.merge(&mut doc2);
 
@@ -746,14 +736,14 @@ fn does_not_interleave_sequence_insertions_at_same_position() {
     let mut doc2 = new_doc_with_actor(actor2);
 
     let wisdom = doc1
-        .set(automerge::ROOT, "wisdom", automerge::Value::list())
+        .set(ObjId::Root, "wisdom", automerge::Value::list())
         .unwrap()
         .unwrap();
     doc2.merge(&mut doc1);
 
     let doc1elems = doc1
         .splice(
-            wisdom,
+            &wisdom,
             0,
             0,
             vec![
@@ -766,10 +756,9 @@ fn does_not_interleave_sequence_insertions_at_same_position() {
         )
         .unwrap();
 
-    let wisdom_in_doc2 = translate_obj_id(&doc1, &doc2, wisdom);
     let doc2elems = doc2
         .splice(
-            wisdom_in_doc2,
+            &wisdom,
             0,
             0,
             vec![
@@ -788,16 +777,16 @@ fn does_not_interleave_sequence_insertions_at_same_position() {
         &doc1,
         map! {
             "wisdom" => {wisdom => list![
-                {doc1elems[0].native() => "to"},
-                {doc1elems[1].native() => "be"},
-                {doc1elems[2].native() => "is"},
-                {doc1elems[3].native() => "to"},
-                {doc1elems[4].native() => "do"},
-                {doc2elems[0].translate(&doc2) => "to"},
-                {doc2elems[1].translate(&doc2) => "do"},
-                {doc2elems[2].translate(&doc2) => "is"},
-                {doc2elems[3].translate(&doc2) => "to"},
-                {doc2elems[4].translate(&doc2) => "be"},
+                {doc1elems[0] => "to"},
+                {doc1elems[1] => "be"},
+                {doc1elems[2] => "is"},
+                {doc1elems[3] => "to"},
+                {doc1elems[4] => "do"},
+                {doc2elems[0] => "to"},
+                {doc2elems[1] => "do"},
+                {doc2elems[2] => "is"},
+                {doc2elems[3] => "to"},
+                {doc2elems[4] => "be"},
             ]}
         }
     );
@@ -811,20 +800,19 @@ fn mutliple_insertions_at_same_list_position_with_insertion_by_greater_actor_id(
     let mut doc2 = new_doc_with_actor(actor2);
 
     let list = doc1
-        .set(automerge::ROOT, "list", automerge::Value::list())
+        .set(ObjId::Root, "list", automerge::Value::list())
         .unwrap()
         .unwrap();
-    let two = doc1.insert(list, 0, "two").unwrap();
+    let two = doc1.insert(&list, 0, "two").unwrap();
     doc2.merge(&mut doc1);
 
-    let list_in_doc2 = translate_obj_id(&doc1, &doc2, list);
-    let one = doc2.insert(list_in_doc2, 0, "one").unwrap();
+    let one = doc2.insert(&list, 0, "one").unwrap();
     assert_doc!(
         &doc2,
         map! {
-            "list" => { list.translate(&doc1) => list![
-                { one.native() => "one" },
-                { two.translate(&doc1) => "two" },
+            "list" => { list => list![
+                { one => "one" },
+                { two => "two" },
             ]}
         }
     );
@@ -838,20 +826,19 @@ fn mutliple_insertions_at_same_list_position_with_insertion_by_lesser_actor_id()
     let mut doc2 = new_doc_with_actor(actor2);
 
     let list = doc1
-        .set(automerge::ROOT, "list", automerge::Value::list())
+        .set(ObjId::Root, "list", automerge::Value::list())
         .unwrap()
         .unwrap();
-    let two = doc1.insert(list, 0, "two").unwrap();
+    let two = doc1.insert(&list, 0, "two").unwrap();
     doc2.merge(&mut doc1);
 
-    let list_in_doc2 = translate_obj_id(&doc1, &doc2, list);
-    let one = doc2.insert(list_in_doc2, 0, "one").unwrap();
+    let one = doc2.insert(&list, 0, "one").unwrap();
     assert_doc!(
         &doc2,
         map! {
-            "list" => { list.translate(&doc1) => list![
-                { one.native() => "one" },
-                { two.translate(&doc1) => "two" },
+            "list" => { list => list![
+                { one => "one" },
+                { two => "two" },
             ]}
         }
     );
@@ -863,31 +850,68 @@ fn insertion_consistent_with_causality() {
     let mut doc2 = new_doc();
 
     let list = doc1
-        .set(automerge::ROOT, "list", automerge::Value::list())
+        .set(ObjId::Root, "list", automerge::Value::list())
         .unwrap()
         .unwrap();
-    let four = doc1.insert(list, 0, "four").unwrap();
+    let four = doc1.insert(&list, 0, "four").unwrap();
     doc2.merge(&mut doc1);
-    let list_in_doc2 = translate_obj_id(&doc1, &doc2, list);
-    let three = doc2.insert(list_in_doc2, 0, "three").unwrap();
+    let three = doc2.insert(&list, 0, "three").unwrap();
     doc1.merge(&mut doc2);
-    let two = doc1.insert(list, 0, "two").unwrap();
+    let two = doc1.insert(&list, 0, "two").unwrap();
     doc2.merge(&mut doc1);
-    let one = doc2.insert(list_in_doc2, 0, "one").unwrap();
+    let one = doc2.insert(&list, 0, "one").unwrap();
 
     assert_doc!(
         &doc2,
         map! {
-            "list" => {list.translate(&doc1) => list![
-                {one.native() => "one"},
-                {two.translate(&doc1) => "two"},
-                {three.native() => "three" },
-                {four.translate(&doc1) => "four"},
+            "list" => {list => list![
+                {one => "one"},
+                {two => "two"},
+                {three => "three" },
+                {four => "four"},
             ]}
         }
     );
 }
 
+#[test]
+fn should_handle_arbitrary_depth_nesting() {
+    let mut doc1 = new_doc();
+    let a = doc1.set(ObjId::Root, "a", automerge::Value::map()).unwrap().unwrap(); 
+    let b = doc1.set(&a, "b", automerge::Value::map()).unwrap().unwrap();
+    let c = doc1.set(&b, "c", automerge::Value::map()).unwrap().unwrap();
+    let d = doc1.set(&c, "d", automerge::Value::map()).unwrap().unwrap();
+    let e = doc1.set(&d, "e", automerge::Value::map()).unwrap().unwrap();
+    let f = doc1.set(&e, "f", automerge::Value::map()).unwrap().unwrap();
+    let g = doc1.set(&f, "g", automerge::Value::map()).unwrap().unwrap();
+    let h = doc1.set(&g, "h", "h").unwrap().unwrap();
+    let j = doc1.set(&f, "i", "j").unwrap().unwrap();
+
+    assert_doc!(
+        &doc1,
+        map!{
+            "a" => {a => map!{
+                "b" => {b => map!{
+                    "c" => {c => map!{
+                        "d" => {d => map!{
+                            "e" => {e => map!{
+                                "f" => {f => map!{
+                                    "g" => {g => map!{
+                                        "h" => {h => "h"}
+                                    }},
+                                    "i" => {j => "j"},
+                                }}
+                            }}
+                        }}
+                    }}
+                }}
+            }}
+        }
+    );
+
+    Automerge::load(&doc1.save().unwrap()).unwrap();
+}
+
 #[test]
 fn save_and_restore_empty() {
     let mut doc = new_doc();
@@ -900,26 +924,25 @@ fn save_and_restore_empty() {
 fn save_restore_complex() {
     let mut doc1 = new_doc();
     let todos = doc1
-        .set(automerge::ROOT, "todos", automerge::Value::list())
+        .set(ObjId::Root, "todos", automerge::Value::list())
         .unwrap()
         .unwrap();
 
-    let first_todo = doc1.insert(todos, 0, automerge::Value::map()).unwrap();
-    doc1.set(first_todo, "title", "water plants")
+    let first_todo = doc1.insert(todos.clone(), 0, automerge::Value::map()).unwrap();
+    doc1.set(&first_todo, "title", "water plants")
         .unwrap()
         .unwrap();
-    let first_done = doc1.set(first_todo, "done", false).unwrap().unwrap();
+    let first_done = doc1.set(first_todo.clone(), "done", false).unwrap().unwrap();
 
     let mut doc2 = new_doc();
     doc2.merge(&mut doc1);
-    let first_todo_in_doc2 = translate_obj_id(&doc1, &doc2, first_todo);
     let weed_title = doc2
-        .set(first_todo_in_doc2, "title", "weed plants")
+        .set(first_todo.clone(), "title", "weed plants")
         .unwrap()
         .unwrap();
 
     let kill_title = doc1
-        .set(first_todo, "title", "kill plants")
+        .set(&first_todo, "title", "kill plants")
         .unwrap()
         .unwrap();
     doc1.merge(&mut doc2);
@@ -929,13 +952,13 @@ fn save_restore_complex() {
     assert_doc!(
         &reloaded,
         map! {
-            "todos" => {todos.translate(&doc1) => list![
-                {first_todo.translate(&doc1) => map!{
+            "todos" => {todos => list![
+                {first_todo => map!{
                     "title" => {
-                        weed_title.translate(&doc2) => "weed plants",
-                        kill_title.translate(&doc1) => "kill plants",
+                        weed_title => "weed plants",
+                        kill_title => "kill plants",
                     },
-                    "done" => {first_done.translate(&doc1) => false},
+                    "done" => {first_done => false},
                 }}
             ]}
         }
diff --git a/edit-trace/Cargo.toml b/edit-trace/Cargo.toml
index 68d47433..375c8995 100644
--- a/edit-trace/Cargo.toml
+++ b/edit-trace/Cargo.toml
@@ -5,6 +5,9 @@ edition = "2018"
 license = "MIT"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+[[bin]]
+name = "edit-trace"
+bench = false
 
 [dependencies]
 automerge = { path = "../automerge" }
diff --git a/edit-trace/benches/main.rs b/edit-trace/benches/main.rs
index fed72f1e..fb5540f7 100644
--- a/edit-trace/benches/main.rs
+++ b/edit-trace/benches/main.rs
@@ -1,13 +1,13 @@
-use automerge::{Automerge, Value, ROOT};
+use automerge::{Automerge, Value, ObjId};
 use criterion::{criterion_group, criterion_main, BenchmarkId, Criterion, Throughput};
 use std::fs;
 
 fn replay_trace(commands: Vec<(usize, usize, Vec<Value>)>) -> Automerge {
     let mut doc = Automerge::new();
 
-    let text = doc.set(ROOT, "text", Value::text()).unwrap().unwrap();
+    let text = doc.set(ObjId::Root, "text", Value::text()).unwrap().unwrap();
     for (pos, del, vals) in commands {
-        doc.splice(text, pos, del, vals).unwrap();
+        doc.splice(&text, pos, del, vals).unwrap();
     }
     doc.commit(None, None);
     doc
diff --git a/edit-trace/src/main.rs b/edit-trace/src/main.rs
index 94fde72c..85b437b7 100644
--- a/edit-trace/src/main.rs
+++ b/edit-trace/src/main.rs
@@ -1,4 +1,4 @@
-use automerge::{Automerge, AutomergeError, Value, ROOT};
+use automerge::{Automerge, AutomergeError, Value, ObjId};
 use std::fs;
 use std::time::Instant;
 
@@ -19,12 +19,12 @@ fn main() -> Result<(), AutomergeError> {
     let mut doc = Automerge::new();
 
     let now = Instant::now();
-    let text = doc.set(ROOT, "text", Value::text()).unwrap().unwrap();
+    let text = doc.set(ObjId::Root, "text", Value::text()).unwrap().unwrap();
     for (i, (pos, del, vals)) in commands.into_iter().enumerate() {
         if i % 1000 == 0 {
             println!("Processed {} edits in {} ms", i, now.elapsed().as_millis());
         }
-        doc.splice(text, pos, del, vals)?;
+        doc.splice(&text, pos, del, vals)?;
     }
     let _ = doc.save();
     println!("Done in {} ms", now.elapsed().as_millis());
diff --git a/scripts/ci/js_tests b/scripts/ci/js_tests
new file mode 100755
index 00000000..7d55db77
--- /dev/null
+++ b/scripts/ci/js_tests
@@ -0,0 +1,17 @@
+THIS_SCRIPT=$(dirname "$0");
+WASM_PROJECT=$THIS_SCRIPT/../../automerge-wasm;
+JS_PROJECT=$THIS_SCRIPT/../../automerge-wasm/automerge-js;
+
+# This will take care of running wasm-pack
+yarn --cwd $WASM_PROJECT build;
+# If the dependencies are already installed we delete automerge-wasm. This makes
+# this script usable for iterative development.
+if [ -d $JS_PROJECT/node_modules/automerge-wasm ]; then
+    rm -rf $JS_PROJECT/node_modules/automerge-wasm
+fi
+# --check-files forces yarn to check if the local dep has changed
+yarn --cwd $JS_PROJECT install --check-files;
+yarn --cwd $JS_PROJECT test;
+
+
+
diff --git a/scripts/ci/run b/scripts/ci/run
index c03f2991..42367e10 100755
--- a/scripts/ci/run
+++ b/scripts/ci/run
@@ -6,3 +6,4 @@ set -eou pipefail
 ./scripts/ci/build-test
 ./scripts/ci/docs
 ./scripts/ci/advisory
+./scripts/ci/js_tests
diff --git a/todo.adoc b/todo.adoc
new file mode 100644
index 00000000..8bafc3c4
--- /dev/null
+++ b/todo.adoc
@@ -0,0 +1,3 @@
+* Add an `ElementId::Head` 
+* Make all fields of Transaction private
+* Remove panics from From implementations in value.rs