frontier field notes · no. 07 ↩ back to the index
Read from the source - and the community arguing about it

The weight of history.

Client-side validation is RGB's superpower: no global ledger, complete privacy, assets that live between two parties. But it has a cost the marketing skips - every transfer carries its entire past. For a stablecoin doing millions of hops, that past has weight. This note measures it, reads the three ways out, and sits in on the exact thread where RGB's builders are debating the fix.

protocol: RGB v0.11.1 concern: consignment growth status: open · being argued now
RGB - a coin is
its layers of history
Hops made0
History it now carries~1 KB
Each hop stacks another layer of history - and ships the whole stack onward. Keep tapping.
§ 00

The law nobody repeals.

Three facts about client-side validation. The first two are why RGB exists. The third is the whole subject of this note - and they are the same fact, seen from two sides.

the gift

No global ledger

State lives between the parties, never broadcast. That is the privacy and the scalability pitch: the chain stores commitments, not amounts. No. 05 called this RGB's strongest claim.

the gift

You verify it yourself

The receiver replays the asset's history against the schema and trusts no server. Sovereignty by construction - the property No. 06 found genuinely intact even inside a browser tab.

the cost

History travels with the coin

To verify, you need the history - so a transfer ships a consignment: every state transition back to genesis. Privacy and this weight are one mechanism. You cannot keep one and drop the other for free.

§ 01

Weigh a coin's past.

A rough model, honest about being rough: consignment size scales with the transition history an allocation carries. Drag the history a coin has accumulated and watch what has to move on the next transfer. The 100 MB figure below is not invented - it's the number a market maker raised in the RGB chat this month.

Model: ~1 KB per transition, a deliberately conservative stand-in. Real consignments vary with contract schema, witness data and allocation structure. The point is the shape of the curve, not a spec sheet.

~100 MB
to send, and to validate, per outgoing transfer
§ 02

Three doors out.

Asked directly in the chat whether 0.11.1 can compact this history, a core contributor was candid: today, one blunt tool exists, and two better ones are research. Here they are, with the trade each one makes.

door one · shipping today

Burn & reissue

Destroy the allocation and mint it fresh. The new coin has a genesis and no ancestry - history reset to zero. It works right now, with no new protocol.

availabilityin v0.11.1, today
costan on-chain event; breaks continuity of the asset
fitblunt - a reset, not a compaction
trade · operational, not cryptographic
door two · near-term

Trusted checkpoints

A trusted party attests "history up to here is valid." Past a checkpoint, you stop re-validating. Achievable relatively soon - and, as the contributor noted, a trust trade-off blockchains already accept (Bitcoin Core ships one).

availabilitynear-term, if formalized
costreintroduces a trusted validator
fitpragmatic first milestone
trade · trust for practicality
door three · research

ZK compression

Prove the history is valid without shipping it - a succinct proof replaces the megabytes. Trustless, and the ideal end state. It also needs specific ZK expertise (the yellowpaper's zk-AluVM) and is "an entire research area still to be explored."

availabilityresearch · timeline unknown
costhard problem; specialist work
fittrustless - keeps the gift intact
trade · complexity for trustlessness

The tell: there is no active research team on the long-term fix yet - effort is on "lower-hanging fruit to speed up validation." High-volume flows are expected to live on L2 channels, where the transaction history doesn't grow. Which is true - and quietly reframes the pitch: RGB on-chain for issuance and settlement, Lightning for the millions of hops in between.

§ 03

The weight, by the numbers.

Straight from the thread and the docs - not a benchmark, a set of stated facts.

~100 MB
consignment a market maker projected after 100k transitions
1
tool that ships today - burn & reissue
2
researched fixes - checkpoints and ZK compression
0
teams actively building the long-term fix right now
genesis
how far back every transfer's history must reach
L2
where high-volume flow is meant to live - history stops growing
§ 04

The room where it's argued.

Field notes usually read the source. This one reads the conversation - the RGB community channel, late July 2026, lightly condensed. A market maker presses on the limit; a contributor answers plainly. Names shortened; wording close to the originals.

Market makerthe question
For a long-lived, high-volume fungible asset, a maker's ancestry could reach a 100 MB consignment. Does every outgoing transfer then require sending ~100 MB? Any way in 0.11.1 to compact or truncate this history?
Contributorthe answer
At the moment the only mechanism to truncate is a burn and reissuance event. In the future, two things can be implemented: checkpoints from trusted parties, so past the first checkpoint you can stop validating; and ZK consignment compression. That's feasible in theory, but an entire research area still to be explored.
Market makerthe push
Isn't this a real scalability concern for high-volume fungible assets? It feels like it should be addressed asap.
Contributorthe reframe
High volumes are expected to leverage L2s where the tx history doesn't grow. But yes, this is definitely a priority to optimize - right now the focus is lower-hanging fruit to speed up validation.
Market makerthe offer
Is there ongoing work we could contribute to or help accelerate - checkpointing, state compression, proof aggregation?
Contributorthe honest close
No ongoing research team is actively working on the long-term solutions just yet. A good first step would be to formalize a proposal. Checkpoints involve trust but are achievable soon; compression can be trustless but needs ZK expertise and will take a while. Starting with checkpoints makes sense - it's a trade blockchains already accept.

Source: RGB Community channel, 24-25 July 2026. Condensed and lightly edited for length; no claims added.

§ 05

What it means for the dollar.

USDT on RGB launches into exactly this. None of it is disqualifying - but here's how the weight of history actually lands on a Bitcoin-native stablecoin.

READ 01

The stablecoin is the hard case

Consignment growth bites hardest on high-velocity fungible assets - which is precisely what a payments stablecoin is. The flagship use case is also the stress test.

READ 02

Lightning is the answer, and the asterisk

Inside a channel, history stops growing - so the L2 reframe is real. The asterisk: opening and closing channels still touches on-chain allocations that carry ancestry.

READ 03

Checkpoints will arrive first

The pragmatic path reintroduces a trusted validator - a small, familiar compromise, but worth watching for anyone who bought RGB for its trustlessness.

READ 04

The proposal slot is open

A contributor explicitly invited someone to formalize a checkpoint proposal. This is a contribution surface hiding in plain sight - the kind these field notes exist to point at.

§ 06

What the footnotes say.

The parts that don't fit on a launch slide.

FN·1

"Infinitely scalable" needs a footnote

RGB's own framing calls client-side validation infinitely scalable because each user validates only their own assets, not global state. True - and it quietly relocates the cost from the network to the individual transfer. Per-user scaling and per-transfer weight are the same design seen from two ends. Even Spark's write-up says it plainly: every recipient must verify the full history, which grows with each transfer.

read the whole sentence
FN·2

The field lifehack: intermediary wallets

Asked how he'd airdrop a token to 20,000 users, one community member shrugged: set up intermediary wallets to reduce the tx history in each. Dumb but efficient, he said. He's right - and the fact that hand-managing ancestry is a real tactic tells you the weight is real, today, not in some theoretical future.

the workaround is the tell
FN·3

zk-AluVM already has a name

Door three isn't vapor: the RGB yellowpaper already specifies a zero-knowledge variant of AluVM - zk-AluVM - as the vehicle for compression. Named, scoped, and unbuilt. The distance between "specified" and "shipping" is the whole story of this frontier.

specified ≠ shipping
FN·4

The series, compounding

No. 05 called client-side validation RGB's strongest privacy claim; No. 06 confirmed it holds even in a browser tab. This note is the same property's bill. Privacy, sovereignty, and the weight of history are one mechanism - you've now seen all three faces of it. No. 05 · No. 06

continuity · one mechanism, three notes