When Uniswap launched UNI in September 2020, 21.27% of the supply went to team members and future employees and 18.04% to investors, vesting over four years. That is the usual way a DAO pays its founders: a slice of the supply set aside before anyone else can buy, released on a schedule fixed at launch. The schedule pays out the same whether the protocol thrives or stalls.
A caper's founder starts with no tokens. Nothing is set aside for them or anyone else before the first buy. They are paid a share of every trade as it happens, buys and sells alike, and the larger part of it in XRD. A caper nobody trades pays its founder nothing, so the founder's income depends on the caper still being worth trading long after launch.
caper.network/wiki/foundatio…
Caper, week to 11 October.
Governance. A proposal now carries one fee, which its proposer pays on creating it and which scales with the radix:native value of its largest option, above a minimum. Triggering a passed proposal is free. Both changes went live on Stokenet, the Radix test network, in the 6 October redeploy. The site settles a closed vote itself, so the TALLY VOTES button is gone, and anyone can press EXECUTE on a passed proposal, as the contract already allowed. The site copies each tally, trigger and resolve verdict off the on-chain record, so it cannot miss one. A caper's page shows your exit share in XRD, and a settled debate counts as concluded rather than live.
In the contracts, not yet deployed: a trade will name the caper it buys or sells directly, and each caper will publish the state of its own curve, which cuts the network fee on a buy, a sell or a swap.
Site. The price chart is the one radix.wiki uses: it opens on candles with a volume pane, switches to an area line, runs from 24 hours to all time by day, week or month, and sits in a panel beside the Trollbox, the caper's chat room. A caper's About moves to a tab beside Market. An idle caper position appears once, in Holdings. Every element shows a focus ring for keyboard users, and a missing page returns a real 404. Caper's one-line description is now "perpetual fundraising for organizations on Radix".
Analytics. Caper counts its own visitors and Plausible is gone: no cookie, no stored address, and each visitor is hashed under a salt deleted daily. The private stats page charts every figure by day with its change, narrows to any page or source it lists, and reaches back a year through the Plausible history.
Wiki. The editor edits every block it shows, keeps Save in view, saves on Cmd-S, warns before you leave unsaved changes, and opens the link field on the current link. Clicking through a text block no longer marks it edited. Gitcoin DAO's page records the Governor upgrade it executed on 5 October, and Arbitrum DAO's records the wind-down of Arbitrum Gaming Ventures.
Under the hood. wiki-formant, the shared package behind caper.network and radix.wiki, is updated through the week. Colour literals become design tokens, the security policy's list of embeddable sites comes from the sanitiser's list, and the wiki and essays are cached for a day and refreshed on edit. Production builds skip Turbopack's build cache, which had shipped stale styles. The video studio films the governance flow: ballot, trigger, resolve and exit.
caper.network/wiki/governanc…
Aave's token holders voted 572,726 to 0 on 10 October to form the Aave Foundation, a Cayman company with no members that will hold the Aave trademark, domains and code. Aave Labs, which builds the protocol, gets no seat on it. The Aave DAO page now covers the terms.
caper.network/wiki/daos/lend…
The Proposals page now covers the 11 October redeploy, which changed how a trade names its caper and left voting, the market window and the proposal fee as they were. caper.network/wiki/governanc…
Gitcoin's token holders have replaced the contract that runs their on-chain votes. The new Governor, executed on 5 October with 2.71m GTC for and none against, adds a guardian that can cancel a malicious proposal but never change or force one, and a quorum the DAO can adjust by vote. Gitcoin moved after Tally, the front end it voted through, announced its shutdown in March. caper.network/wiki/daos/publ…
Arbitrum DAO page: the wind-down of Arbitrum Gaming Ventures, the DAO's gaming venture fund. Token holders voted 191m $ARB to 8k to stop new investments and send about 143.7m ARB back to the treasury. caper.network/wiki/daos/netw…
AxonDAO, a decentralized-science network built around member-owned health data, also runs its own GPU cluster: NVIDIA B200s deployed in March 2026 and rented to enterprise customers for AI and life-sciences research.
It publishes the cluster's earnings on a public dashboard. By 8 October its GPUs had grossed $157k.
That gives it something most DAOs lack: a customer paying for a service, rather than a treasury spent down from a token sale.
caper.network/wiki/desci/eco…
Most token votes weigh a ballot by what the voter held at one earlier block. Compound's Governor reads it at the block the vote opened; a Snapshot space reads it at the block the proposal names. Nothing the voter does after that block changes the count, so a holder can sell their whole position and still vote with it, or vote first and sell a minute later.
A caper reads the voter's balance twice: when they cast the ballot, and again when the ballots are counted. The smaller of the two counts. A member who votes with a large position and sells half of it before the count is counted at half, and one who sells all of it is counted at nothing. A voter who holds none of the caper's tokens when they cast cannot vote at all.
caper.network/wiki/governanc…
In July 2024 a group of holders bought $COMP on the open market and used it to pass Compound Proposal 289. It would have moved about 499,000 COMP of the DAO's reserves, some $24m, into a yield vault the group would influence. It passed 682,000 votes to 634,000 and was withdrawn after a public backlash. Compound weighs a vote by the COMP behind it, so tokens bought that month counted the same as tokens held since launch.
A caper weighs a holding together with a record of taking part. Members add to that record only by casting ballots, and nobody can buy it, borrow it or receive it from another member. A holder who buys a large position the week before a vote brings one ballot's worth of record: the one they cast. Members who have voted on every proposal for a year bring the record of all of those votes.
Stake still counts: of two members with the same record, the larger holder weighs more.
caper.network/wiki/governanc…
Caper Network retweeted
Governance is often treated as overhead until something breaks.
If we want it to receive serious attention and investment, we need to become better at explaining what risk it prevents, what value it protects, and what better governance actually changes.
luma.com/sh98otew
Caper Network retweeted
Exactly, our payouts are onchain. Transparent and resilient.
Check out fund.bread.coop
Find out how we generate funds and distribute them.
Plenty of DAO tokens trade for less than the treasury holds per token. A holder who wants out can only sell to another holder, because the treasury never buys. The discount stays, and it draws buyers who purchase at the discount and then campaign for a wind-down to collect the difference.
A caper's treasury pays the leaving member directly. Anyone who has voted in a caper at least once can leave in one transaction: they hand back their tokens and receive two payments, their share of the treasury and the price of those tokens sold back to the caper's own market. Nobody approves it and nothing is scheduled.
The exit is not all or nothing. A member can redeem part of a position and keep the rest.
caper.network/wiki/foundatio…
Caper, week to 4 October.
Contracts. Each trade now pays a single skim rate, the highest on the stretch of the curve it covers, so a large trade pays more than the same trade split into smaller ones. The skim is the share of every buy and sell that goes to the traded caper's treasury, its founder and the Caper Commons; it now falls away more slowly as a caper's supply sells, and the earliest buy is still the cheapest. Both changes are live on Stokenet, the Radix test network.
In the code, not yet deployed: the core contract, which cannot change once Caper launches on mainnet, will refuse any trade that leaves a curve's reserve short of what the curve owes its holders, so neither a bug nor an upgrade passed by vote could drain it; an upgrade passed by vote will lapse if nobody activates it within one execution window; trades on different capers will no longer write to one shared component; and the core gets the last changes planned before mainnet freezes it, including room for longer market windows in governance.
Governance. An upgrade passed by vote now executes in two transactions. The proposal page offers ACTIVATE UPGRADE once the timelock allows it and shows the time left until then.
Site. A founder can reinstate a line they removed from their caper's Trollbox room, its chat; a line deleted by its own author stays down.
Wiki. Every knowledge-base article has a revision history at its address plus /history, and each revision opens to the words it removed and added, block by block. Caper pages' histories show the same. Wide tables scroll inside their own column, and the markdown copy of each page, the version agents read, keeps the line breaks of blocks nested inside other blocks.
Under the hood. A load test now runs after every redeploy and resumes from the step that failed. The database copy of on-chain trades gets a tracked schema history, ledger positions and larger ids. Publishing the core now requires passing tests, a cost preview against the live network, a database backup and a check that the app's constants match the contracts. The deployment guide gains a mainnet launch checklist and a ledger of deployments. The fee model, an internal page, gains a backtest that prices what a founder and a caper's treasury would have earned on $REDDICKS's trading history on Ociswap. Image uploads authenticate against caper's own storage.
caper.network/wiki/markets/b…
Most DAOs hold two kinds of vote with the same tool. A signalling vote asks what members think; a binding vote decides what the treasury does. On the voting page the two look identical, and which one a result was is often settled afterwards, by the people who would have to carry it out.
A caper settles that before anyone votes. Every option on a proposal is typed when the proposal is created: either an action the treasury will perform, such as a payment to a named account, or a debate position that performs nothing. The type is stored on-chain with the option and cannot be changed once the proposal exists.
A debate position can win the count, and the contract ends it there. A winning action goes to the market instead: holders get a window in which selling counts as an objection, and if the price holds, anyone can carry the action out on-chain. Nobody can decide afterwards that a vote was only advice, or that a debate was a mandate.
caper.network/wiki/governanc…
Pump.fun popularised the bonding-curve launch. A token's price rises up a curve as buyers arrive, and once enough has been bought the token graduates: its liquidity moves to a DEX pool and trading carries on there. The curve is a launch phase, and after it the price comes from a pool with terms of its own.
A launch built that way is fair for as long as it lasts. The handover to the open market is the part its rules stop covering.
A caper's curve does not graduate. The contract that quoted the first buy quotes every later buy and sell as well, and what buyers pay stays in its reserve until a holder sells tokens back. There is no listing day, no pool to seed and no second market. Someone who buys in the second year trades on the same curve as someone who bought in the first hour.
caper.network/wiki/markets/b…
Many DAOs are handed their treasury some time after they exist. A foundation or company raises and holds the funds, the members vote the DAO into being, and the money moves across on a later date, once a legal entity is incorporated and signers are appointed. Until then, members plan a budget held in an account they do not control, and they learn the figure when it arrives.
A caper has no handover. Its treasury is created in the same transaction as the caper, as an account the caper's own contract controls, and the first trade in its token pays into it. No foundation, founder or signer holds the funds for the members in the meantime.
Anyone can read the balance on the ledger at any time. Funds go out only through a proposal the members pass, or to a member who exits with their share.
caper.network/wiki/markets/r…
Caper, week to 27 September.
Contracts. The separate trade fee is gone. Every buy and sell already paid the skim, the cut taken from each trade on top of the curve price, and the skim is now shared: part goes to the traded caper's own treasury and part to the founder, both in XRD, and a small part goes in tokens to the $CAPER treasury.
Leaving a caper works from a real wallet again. An earlier change had made every exit signed by a wallet key fail, which the test suites could not see because their accounts need no signature. An exit now also burns vote tokens only from the account it pays out to; vote tokens are what members earn by casting ballots.
A swap between two capers makes fewer calls between contracts, so it costs less to run. The contracts were redeployed from scratch on Stokenet, the Radix test network, on 21 and 25 September to ship these changes.
Site. The capers directory shows each caper's own share card on its tile, the same image a link to it shows on X. A proposal the ballot rejected no longer shows market figures beside its result, since the market never judged it. The live feed opens shorter, with a Load more button. Corners across the app now take one radius, drawn from the logo's wing, and the logo, favicon, colours, share cards and wiki diagrams are all generated from one stylesheet.
Security. Pages that change per visitor now run only the scripts the server issued for that request, rate limits key on the visitor's real address, and wallet sign-in responses are never cached.
Wiki. Page titles carry the site name again, and each page gives search engines its own description instead of one shared line. Section indexes count the pages in their sub-sections. The side menu marks the current page at every depth on first load. A new check walks every rendered page for missing or duplicate titles and skipped heading levels. Pages that described the old fee were corrected and the skim diagram redrawn, and a page on Mango DAO was added.
Under the hood. CI fails on unused code, files and dependencies, and on inline style combinations that belong in the stylesheet; the maintenance scripts get a typecheck; blog dates are pinned to UTC like the rest of the site; duplicate number formatters became one; and the internal fee model now answers one question: what a founder and a caper's treasury earn a month at a given trading volume.
caper.network/wiki/markets/b…