diff --git a/README.md b/README.md
index 153d68c..428980d 100644
--- a/README.md
+++ b/README.md
@@ -26,6 +26,10 @@ Provider Records are used to find the [peerIDs](https://docs.libp2p.io/concepts/
 
 https://tabcat.github.io/zzzync/
 
+## Spec
+
+https://github.com/tabcat/zzzync/blob/master/spec.md
+
 ---
 
 This work is being funded as part of a [grant](https://github.com/tabcat/rough-opal) by [Protocol Labs](https://protocol.ai)
diff --git a/spec.md b/spec.md
new file mode 100644
index 0000000..fc53963
--- /dev/null
+++ b/spec.md
@@ -0,0 +1,122 @@
+
+# Zzzync Spec Document
+
+This document is meant to provide clarity on zzzync as a protocol.
+
+Protocol Version: `1.0.0-beta`
+
+---
+
+<br/>
+
+## DCID
+
+DCID are a definition made by Zzzync and are not recognized as a separate format by IPFS.
+They are of the same format as CIDs but are created differently.
+DCID are created by taking the CID of a manifest/setup document for some dynamic content, then prefixing `'/dcoi/'` decoded utf-8 bytes to the CID multihash, and then hashing into a different CID.
+
+> 'dcoi' is an acronym meaning *dynamic content over ipfs*
+
+DCID format:
+
+```
+<0x01 (CIDv1)><0x55 (multicode raw)><sha256(<'/dcoi/' decoded utf-8><manifest CID multihash>)>
+```
+
+---
+> **Q:** Why not just use `<'/dcoi/' decoded utf-8><manifest CID multihash>` as the routing key. Why convert it back to a CID?
+
+> **A:** The kad-dht api in javascript makes working with CIDs easier. As the protocol matures this may be changed.
+---
+
+<br/>
+
+## PeerId
+
+A [PeerId](https://github.com/libp2p/specs/blob/master/peer-ids/peer-ids.md) is a Libp2p definition.
+They are used to identify nodes on the network and are made of a cryptographic keypair.
+They are unique to a device and must not be shared.
+
+<br/>
+
+## Advertisers
+
+Advertisers are used to point from a DCID to PeerIds.
+Advertisers can use any system to do so.
+There may be multiple advertisers to use and defined under this protocol and they can be used together.
+
+Advertisers need only re-advertise when the system requires it, to keep the advertisements available.
+
+### DHT Advertiser
+
+The DHT advertiser uses the IPFS DHT's Provider Records to point from DCIDs to PeerIds.
+Each record points to a different PeerId and records stay on the network for a maximum of 48hours.
+
+The `ADD_PROVIDER` query is used to advertise that a PeerId is the provider of a DCID.
+The `GET_PROVIDERS` query is used to discover PeerIds that are providing a DCID.
+
+The [IPFS DHT spec](https://github.com/libp2p/specs/tree/master/kad-dht) provides further information.
+
+<br/>
+
+## Namers
+
+Namers are used to point from a PeerId to a CID.
+The CID is the latest version of some dynamic content.
+Namers can use any system that provides verifiable guarantees of a mutable PeerId -> CID mapping.
+There may be multiple namers to use and defined under this protocol
+
+Namers need only republish after a change has been made to a local replica, changing its CID.
+
+### IPNS Namer
+
+The IPNS namer uses the Interplanetary Name System to resolve PeerIds to CIDs.
+
+[IPNS spec](https://specs.ipfs.tech/ipns/ipns-record/)
+
+IPNS Records have a value field which is encoded as bytes and can contain something other than a CID.
+For Zzzync's purpose there should only ever be IPFS path here, encoded utf-8.
+An immutable IPFS path is utf-8 encoded string that includes a multibase encoded CID with an `/ipfs/` prefix:
+
+```
+/ipfs/<multibase encoded CID>
+```
+
+### W3Name Namer
+
+The W3Name namer uses the W3Name system to resolve PeerIds to CIDs.
+W3Name system was built to be a substitute for IPNS.
+
+The design and records used are very similar so the IPNS Namer section on value field format also applies here.
+
+PeerIds are made of cryptographic keys.
+The private key of the PeerId being used must be used to sign the W3Name records.
+
+<br/>
+
+## Replication Protocol
+
+### Advertisement
+
+After a change has been made to a local replica and the replica data has been uploaded to another machine:
+
+1. Use Namer to publish device unique PeerId -> CID of replica.
+2. Use Advertiser to advertise DCID -> PeerId
+
+### Discovery
+
+1. Use Advertiser to query PeerIds for DCID.
+2. Use Namer for each PeerId to resolve replica CIDs.
+
+If the replica data for the CID has been uploaded to another machine offline replication can be completed.
+
+<br/>
+
+## Replica Hosts
+
+For offline discovery to be possible the records created by the Advertisers/Namers must be available.
+For offline replication the referenced replica data remain available.
+Without both offline collaboration cannot occur in this context.
+
+This document does not specify how data should be hosted for this purpose.
+It only specifies how to advertise and discover the latest versions for some dynamic content.
diff --git a/src/dcid.ts b/src/dcid.ts
index f9ce9e4..241fc37 100644
--- a/src/dcid.ts
+++ b/src/dcid.ts
@@ -4,7 +4,7 @@ import { sha256 } from 'multiformats/hashes/sha2'
 import { concat } from 'uint8arrays/concat'
 import { fromString } from 'uint8arrays/from-string'
 
-// dynamic content on ipfs
+// dynamic content over ipfs
 const DCOI_KEY = fromString('/dcoi/')
 
 // similar to ipns routing key in js-ipns