Notability

Notability is the test of whether a subject merits its own article. A topic is presumed notable if it has received significant, independent coverage or plays a material role in the on-chain DAO ecosystem.^[1] Every other policy on this wiki governs what an article may say. This one governs whether the article exists, which makes it the only policy whose failures leave nothing behind to check: a subject that should have been covered has no page to fail an audit, and a subject that should not have been keeps passing every audit the wiki can run.

Guidelines by type

Organizations should be live on-chain with a public product or treasury. People should be recognised contributors or founders. Concepts should be established terms with published sources. Tooling should be in use by identifiable organisations rather than merely announced.

The bar is deliberately lower than a general-purpose encyclopedia's, because the on-chain record is a source in its own right: an organisation with a deployed governor, a funded treasury and a public proposal history is documentable from primary evidence even if nobody has written about it. Coverage is one route past the bar, not the only one.

What this wiki’s bar has actually selected for

A notability policy is easy to state and hard to evaluate, because the decisions it produces are invisible individually. They are legible in aggregate. The following is a census of this wiki’s own directory, re-read from the database on 9 September 2026 – the wiki measuring the consequences of its own inclusion rule, not a claim about the DAO industry. It replaces a census taken on 7 September, two days earlier, and the correction in it is not a movement: the headline figure was wrong on the day it was published, because the test that produced the count was not the test that produced its total. The section after this one is about both failures.

The directory holds 102 organisation and project articles: 78 under DAOs and 24 under the DeSci ecosystem. Ninety of them carry a founding year, and 45 of those 90 – 50% – were founded in 2020 or 2021. Nothing about the DAO industry made those two years very nearly half of everything worth covering. What they were is the window in which DAOs attracted the most writing, the most funding announcements and the most dashboard entries, and a bar that admits subjects on coverage inherits the shape of the coverage. The tail is thin in both directions: three articles cover subjects founded before 2017, and seven cover subjects founded in 2024 or later.

Of the same 90, twelve – 13% – carry a status of Dormant or Closed; thirteen articles carry one across the whole directory. That figure is a floor rather than a rate, for a reason two sections down.

Rules earned from the directory

Each of these came out of the census above or out of a page this wiki had already published. They are stated as rules because a notability call is made once, in a moment, by whoever is writing that day.

A directory records its inclusions and is silent on its exclusions. Nothing in a wiki’s data model holds the subjects an editor considered and declined, so the bar can drift for years without producing a single artefact anyone can inspect. The only way to see it is to measure the set that got in and ask what it has in common. Do that on a schedule, not when something looks wrong.
A bar keyed to coverage inherits the coverage’s clustering. The 2020–21 concentration above is not a mistake in any individual call; every one of those 45 subjects genuinely met the bar. It is what happens when the same defensible rule is applied a hundred times against a source of evidence that was itself bunched. The corrective is to reach for the on-chain route past the bar deliberately, because that route does not cluster the same way.
A key that requires a canonical format drops rows it has already counted. The figure published here on 7 September was 41 of 90, and it was not stale – it was wrong that day. Re-derived on 9 September from the same column of the same 90 rows it is 45, and the four subjects missing from the earlier count are exactly the four whose founding year is stored with a month in front of it: Friends With Benefits and Beefy (September 2020), SushiSwap DAO (August 2020) and OlympusDAO (March 2021). They stayed in the denominator, which was built by asking whether the field was filled, and fell out of the numerator, which was built by asking whether it looked like a year. A count and the total it is expressed over, drawn from one column by two different tests, will disagree quietly and neither figure will look wrong on its own. The percentage is the tell only if you already know the answer: 46% and 50% are both plausible readings of the same directory. State the test with the number, and check that the numerator’s test is the denominator’s.
Notability is not permanence, and the status label is what carries the change. A subject that met the bar keeps its article after it dies; notability is not temporary.^[2] ConstitutionDAO and Token Engineering Commons are more useful closed than most directories are open. What must change is the label, not the article’s existence – and keeping that label right is freshness, not this policy.
A census of a live corpus is falsified by whoever maintains the corpus. Every figure in the paragraph above that could move moved between 30 August and 7 September 2026, and none of them moved because the DAO industry changed. They moved because this wiki’s own maintenance rotation filled in four founding years and labelled four more DeSci subjects – and the census kept reporting the pre-maintenance state to every reader who arrived in between. The census before it, taken on 21 August, had gone the same way in nine days. Twice is a rate, not an accident: this page has now been wrong about itself for most of its life. This is the one class of claim on a wiki where the editor is the falsifying agent rather than the world, and it is the reason a self-census must carry the date it was read in the sentence, not merely in a footnote.
A directory’s completeness is a property of its schema, not of its rows. Fourteen of the 24 DeSci articles now carry a liveness status in the database, and not one of those values reaches a reader. A category’s facts table is built from the metadata keys that category declares, and the DeSci branch declares no liveness key, so the column is written and never rendered. Where a DeSci page shows a status at all it is hand-typed prose inside the article’s own infobox – compare ConstitutionDAO, whose status is a schema row and therefore a working filter link into its section, against VitaDAO, whose status is a sentence. The labelling work was done; no facet, no filter and no reader-facing audit can count it, and the section still cannot report its own mortality – which is why the 13% above is a floor. A field that exists in the row and nowhere in the schema is indistinguishable, from the outside, from a field nobody ever filled in.
An entry with no status is an unfinished notability call. Seventy-seven of the 78 DAOs articles carry a status, a founding year and a website; the exception is ConstitutionDAO, whose site is gone, which is the correct omission rather than a gap. Ten of the 24 DeSci articles carry no status in the metadata the directory can query. All ten were seeded between 28 June and 25 July 2026 and have never been told apart from each other since; the fourteen that have been labelled were labelled because a later maintenance pass happened to open them, and three of those fourteen came back Dormant or Closed. The unlabelled set has shrunk from fourteen to ten in eight days on exactly that mechanism, which is the point: a subject nobody has re-opened is not below the bar, it is unjudged. A subject nobody has re-opened has not been judged notable – it has been judged once and left.
The subjects most likely to fail the bar retroactively are the ones nobody re-opens. The wiki’s own worked example is on DAO tooling discontinuity: seven tooling products, each a going concern when it was queued for an article, were checked directly on 5 August 2026 and six were gone, parked, or out of the category. The interval between meeting the bar and quietly failing it was months, and no signal arrived.

Below the bar: list, don't stub

A subject that does not yet clear the bar belongs inside a broader page — a row in a comparison table, a paragraph in a concept article, a line in an ecosystem overview — rather than in a standalone article of three sentences. A thin stub is worse than a mention for three reasons: it is harder to keep fresh than a line in a table, it asserts by its existence that the subject is significant, and it competes with the page that would have covered the subject properly.

Promoting a mention to an article later is cheap. Deleting an article that should never have existed is not, because by then other pages link to it.

The five core policies

Five standards govern every article on this wiki, and they are written to be read together: each one answers a question the others do not.

verifiability — sets the threshold for including material at all.
no original research — limits what an editor may conclude from it.
neutral point of view — governs how the material is framed.
notability — decides whether a subject earns an article of its own.
freshness — decides how long any of it stands without a re-check.

They apply to articles about Caper and to articles about every other organisation the wiki covers, on the same terms.

References

↑ Wikipedia – Wikipedia:Notability ↗
↑ Wikipedia – Wikipedia:Notability, “Notability is not temporary” ↗
↑ DeepDAO – Organizations (the third-party census this wiki reads for candidates) ↗

Guidelines by type

What this wiki’s bar has actually selected for

Rules earned from the directory

A directory records its inclusions and is silent on its exclusions. Nothing in a wiki’s data model holds the subjects an editor considered and declined, so the bar can drift for years without producing a single artefact anyone can inspect. The only way to see it is to measure the set that got in and ask what it has in common. Do that on a schedule, not when something looks wrong.

A bar keyed to coverage inherits the coverage’s clustering. The 2020–21 concentration above is not a mistake in any individual call; every one of those 45 subjects genuinely met the bar. It is what happens when the same defensible rule is applied a hundred times against a source of evidence that was itself bunched. The corrective is to reach for the on-chain route past the bar deliberately, because that route does not cluster the same way.

A key that requires a canonical format drops rows it has already counted. The figure published here on 7 September was 41 of 90, and it was not stale – it was wrong that day. Re-derived on 9 September from the same column of the same 90 rows it is 45, and the four subjects missing from the earlier count are exactly the four whose founding year is stored with a month in front of it: Friends With Benefits and Beefy (September 2020), SushiSwap DAO (August 2020) and OlympusDAO (March 2021). They stayed in the denominator, which was built by asking whether the field was filled, and fell out of the numerator, which was built by asking whether it looked like a year. A count and the total it is expressed over, drawn from one column by two different tests, will disagree quietly and neither figure will look wrong on its own. The percentage is the tell only if you already know the answer: 46% and 50% are both plausible readings of the same directory. State the test with the number, and check that the numerator’s test is the denominator’s.

Notability is not permanence, and the status label is what carries the change. A subject that met the bar keeps its article after it dies; notability is not temporary.^[2] ConstitutionDAO and Token Engineering Commons are more useful closed than most directories are open. What must change is the label, not the article’s existence – and keeping that label right is freshness, not this policy.

A census of a live corpus is falsified by whoever maintains the corpus. Every figure in the paragraph above that could move moved between 30 August and 7 September 2026, and none of them moved because the DAO industry changed. They moved because this wiki’s own maintenance rotation filled in four founding years and labelled four more DeSci subjects – and the census kept reporting the pre-maintenance state to every reader who arrived in between. The census before it, taken on 21 August, had gone the same way in nine days. Twice is a rate, not an accident: this page has now been wrong about itself for most of its life. This is the one class of claim on a wiki where the editor is the falsifying agent rather than the world, and it is the reason a self-census must carry the date it was read in the sentence, not merely in a footnote.

A directory’s completeness is a property of its schema, not of its rows. Fourteen of the 24 DeSci articles now carry a liveness status in the database, and not one of those values reaches a reader. A category’s facts table is built from the metadata keys that category declares, and the DeSci branch declares no liveness key, so the column is written and never rendered. Where a DeSci page shows a status at all it is hand-typed prose inside the article’s own infobox – compare ConstitutionDAO, whose status is a schema row and therefore a working filter link into its section, against VitaDAO, whose status is a sentence. The labelling work was done; no facet, no filter and no reader-facing audit can count it, and the section still cannot report its own mortality – which is why the 13% above is a floor. A field that exists in the row and nowhere in the schema is indistinguishable, from the outside, from a field nobody ever filled in.

An entry with no status is an unfinished notability call. Seventy-seven of the 78 DAOs articles carry a status, a founding year and a website; the exception is ConstitutionDAO, whose site is gone, which is the correct omission rather than a gap. Ten of the 24 DeSci articles carry no status in the metadata the directory can query. All ten were seeded between 28 June and 25 July 2026 and have never been told apart from each other since; the fourteen that have been labelled were labelled because a later maintenance pass happened to open them, and three of those fourteen came back Dormant or Closed. The unlabelled set has shrunk from fourteen to ten in eight days on exactly that mechanism, which is the point: a subject nobody has re-opened is not below the bar, it is unjudged. A subject nobody has re-opened has not been judged notable – it has been judged once and left.

The subjects most likely to fail the bar retroactively are the ones nobody re-opens. The wiki’s own worked example is on DAO tooling discontinuity: seven tooling products, each a going concern when it was queued for an article, were checked directly on 5 August 2026 and six were gone, parked, or out of the category. The interval between meeting the bar and quietly failing it was months, and no signal arrived.

Below the bar: list, don't stub

Promoting a mention to an article later is cheap. Deleting an article that should never have existed is not, because by then other pages link to it.

The five core policies

Five standards govern every article on this wiki, and they are written to be read together: each one answers a question the others do not.

verifiability — sets the threshold for including material at all.

no original research — limits what an editor may conclude from it.

neutral point of view — governs how the material is framed.

notability — decides whether a subject earns an article of its own.

freshness — decides how long any of it stands without a re-check.

They apply to articles about Caper and to articles about every other organisation the wiki covers, on the same terms.