A personal token is a tradeable token issued by an individual against themselves – a claim on, or access to, their future output, attention, or success. The practice is often called self-tokenization, and it overlaps with the broader categories of creator tokens and social tokens. The recurring promise is the same: let a person be capitalized the way a company is, so that people who believe in someone can back them early and share in what follows.

Antecedents

The idea predates blockchains. In 2008 the artist Mike Merrill incorporated himself and sold shares in his own person, letting shareholders vote on decisions in his life – an early, deliberately literal experiment in treating an individual as a joint-stock enterprise. A decade earlier, in 1997, the banker David Pullman securitized David Bowie's back-catalogue royalties into Bowie Bonds, a bond that let investors buy a stream of one person's future income. Neither was a token, but both established the underlying move: an individual issuing a claim on their own future and selling it to backers.

The 2020 wave

Cheap on-chain issuance turned a rare, bank-intermediated act into something anyone could do in an afternoon, and in 2020 a wave of personal and social tokens followed. Contemporary coverage was already flagging the risks. That June, the entrepreneur Alex Masmej sold $ALEX, raising roughly $20,000 and letting holders vote on his personal choices. Many such tokens were issued and traded through custodial platforms, the largest of which was Roll.

A parallel strand used bonding curves rather than a custodial order book. Zap, a protocol built around bonding curves, let anyone mint a personal token whose price rose and fell along a curve: backers bonded the ZAP token to mint the personal token and could un-bond to sell it back to the curve for ZAP. In 2020 a personal token issued in the name of Ben Gravis became one of the first of these to run on a curve rather than a custodial ledger. What the ledger records is the artefact, not the act: the contract is typed ERC-721 rather than a fungible personal token, and it has been silent since April 2021 – read again on 25 August 2026 at a total supply of two, both held by one address. The curve gave holders something the custodial tokens lacked – a standing counterparty to sell back into – but it attached no shared treasury and no governance to the token, and the curve remained the issuer's own arrangement.

Why most failed

The 2020 tokens are usually written off as a speculative bubble, but the more durable lesson is structural. Three things were typically missing:

• No treasury the holders owned. Buying a personal token generally funded a price, not a pooled balance sheet the backers had a claim on. The token was a bet on the issuer's trajectory, backed by nothing the holders controlled.
• No enforceable exit. Leaving meant selling into whatever bid existed. On custodial platforms that could evaporate; even the bonding-curve versions redeemed only against the issuer's own curve, which the issuer could stop maintaining.
• Custodial and key risk. Where a platform held the keys, its failure was the holders' failure. In early 2021 an attacker drained roughly $5.7 million from Roll's hot wallet, and personal tokens minted on it fell sharply in a single week.

The through-line is that holders held upside and a sell button, and nothing in between. When trust in the creator or the custodian wavered, there was no owned collateral to fall back on and no defined way to redeem a share, because there was no share.

What the record shows in 2026

The 2020 wave is usually described in the past tense but rarely checked. Three of its landmarks were re-verified first-hand for this page on 10 August 2026, and the results sharpen the argument above rather than soften it.

• Roll, the largest custodial issuer, is gone. The domain tryroll.com has no A record on the apex, www, or app at either Google's (8.8.8.8) or Cloudflare's (1.1.1.1) resolver. The Internet Archive's CDX index records its last successful capture on 9 November 2025 and its last capture attempt of any kind on 17 November 2025; every capture from 22 July 2025 onward returned a byte-identical homepage (the same CDX digest), so the site was frozen for months before the domain lapsed. The custody problem the hack exposed did not end with the hack – it ended with the platform, and with it any token page a holder might still have wanted to read.
• The bonding-curve personal token has twelve events in its entire history. The Ben Gravis contract has recorded exactly twelve token transfers since deployment, the first on 3 May 2020 and the last on 8 April 2021 at block 12,197,817 – read from Blockscout's keyless decoded-transfer API. Every one of the twelve is either a mint from the zero address or a burn back to it. Not one is a transfer between two holders. A handful of backers bonded, the same handful un-bonded, and the token has been silent for over five years: the curve worked exactly as designed and there was never a secondary life for it to protect.
• The pre-blockchain experiment outlived the tokens. kmikeym.com still resolves and still serves Mike Merrill's shareholder record, eighteen years after he incorporated himself. The oldest version of the idea – a person, a share register, and a standing obligation to the holders – is the one still running.

Taken together these are not three separate failures. Custodial issuance died with its custodian, curve-only issuance died of having nothing to hold, and the arrangement that survived is the one where the claim was defined and the issuer stayed answerable for it.

How a caper differs

A caper applies the same "individual as an institution" idea but supplies the two things the 2020 wave omitted. Buys along the bonding curve fill a shared treasury that every holder has an enforceable claim on, rather than vanishing into a price. And every holder has an exit right: they can burn their vote and redeem a proportional share of that treasury – the same proportion that sets their weight in a vote. A backer is therefore never in the 2020 position of holding a claim on nothing; at any moment they can name their portion of the collateral and leave with it. This is what separates a personal token that is a bet on someone's mood from one that behaves like a small institution.

Further reading

• The Personal Caper – from Bowie Bonds to self-incorporation, and what a personal token needs to be credible.
• The Exit Right – why a redeemable, vote-weighted claim on a treasury is the load-bearing primitive.
• How DAOs Fail and Curation Markets – adjacent failure modes and the bonding-curve lineage.

The personal caper

The 2020 wave gave a person issuance and a market price but nothing they owned behind it, so when trust broke the claim broke with it. A personal caper closes that gap with the two primitives that wave lacked: a treasury the token actually holds, and a binding exit that redeems a real share of it — the same weight, (t·v) / (V·T), that decides a vote.

How the vote weight and the exit actually work

The two claims above – that a holder's exit is enforceable and that it redeems the same proportion as their vote – are worth stating precisely, because the precision is what distinguishes them from the 2020 promises. Both come from one function in the protocol's logic package, and both are stated here against the live contract source rather than a pinned commit. That distinction is load-bearing: the ballot-era upgrade of 16 August 2026 changed where one of the formula's terms is read from without changing a character of the formula itself, so a verification note pinned to an earlier revision would still read as current.

A member's weight is

w = (t · v) / (V · T)

where t is the member's governance-token amount, v their vote tokens, V the caper's total vote supply, and T its circulating token supply. One expression, but the two paths feed it from different places: a ballot reads t as the stake escrowed in the caper's ballot box and v as the voter's whole vote-token balance, while exit() reads both off the buckets the member hands in – so an exit that surrenders part of a position redeems that part's share. Three properties follow, and each one answers something the 2020 wave left open.

Votes are soulbound; the weight is not

Vote tokens are minted directly to the voter's account each time they vote and cannot be transferred – they are earned by participating and there is no market in them. But holdings enter the formula as a multiplicative factor, so a larger position is strictly more weight at the same participation. It is worth being blunt about this, because it is the usual place these designs are oversold: a caper is not a one-person-one-vote system with a soulbound badge. Participation and stake both count, and neither substitutes for the other – a holder who never votes has v = 0 and therefore no weight at all, and a voter who holds nothing has t = 0 and likewise none.

Selling before settlement cuts your own vote

A ballot records the holdings behind it, but settlement does not trust that record. It reads each voter's live balance and clamps the weight to min(held at vote, balance now). So a holder who votes and then sells into the rise has their own ballot reduced to whatever they still hold when the proposal settles. The 2020 personal token's holder could extract on sentiment and leave the same afternoon with their say intact, because there was no settlement to be present for. Here the vote and the position have to survive together.

An exit is one transaction with two legs

Exiting requires both buckets – vote tokens and governance tokens, both non-zero – and pays out twice. That first condition is a gate rather than a formality: a holder who has never voted holds no vote tokens and so cannot exit at all. They can still sell their position back along the curve like any other holder, but the claim on pooled collateral belongs to the members who took part in deciding what the collateral was for. The treasury share is computed with the same w above, before the vote tokens are burned, so the proportion that decided a member's votes is the proportion of collateral they leave with. In the same transaction the governance tokens are sold back along the bonding curve at its current price, less the trade fee. A leaver therefore takes their share of the treasury and the curve's standing bid, and the caper's own vote supply shrinks by exactly the weight that left. Nothing about that requires the issuer's cooperation, which is the whole of the difference from 2020.