What Are Wikipedia's Perennial Sources? We Counted All 497.

What the list actually is

Wikipedia's perennial sources list is a register of publications whose reliability editors have debated repeatedly, with each entry carrying a status and a summary of the discussions that produced it. It is one of the most consequential documents in publishing that almost nobody in publishing reads.

Its own framing is modest. The page states it is “not a policy or guideline” and describes itself as a non-exhaustive compilation summarising consensus from the reliable sources noticeboard. It is a convenience index of settled arguments, not a licensing regime.

Entries are not added casually. Inclusion requires “an uninterrupted request for comment on the source's reliability, or two or more significant discussions” at the noticeboard. A source appears here because editors have fought about it more than once and reached a position worth recording.

The list matters beyond Wikipedia because it is one of the few places where a large, public, adversarial community has recorded durable judgements about named outlets, with reasons attached. Nothing comparable exists in the marketing literature.

What we measured, and how

The list is too large to render as a single page, so its table is transcluded from eight subpages. We retrieved the raw wikitext of all eight on 18 September 2026 and parsed the status assigned to every entry.

We counted in two independent ways and required them to agree before publishing anything. The first counted the status template attached to each row; the second counted the CSS status class on the row itself. Both returned 497 entries with an identical distribution.

There is a ninth page, an overflow area holding 35 further rows. We excluded it from every figure here, because the page states that those rows “will not be visible on the main page” until template limits are resolved. Including them would report a list readers cannot see. Counting them would have raised the total to 532 without materially changing the proportions.

Blacklisted entries carry a base status as well as the blacklist flag, so we report them inside the generally unreliable and deprecated totals rather than as a separate category, and state the blacklisted count separately below.

What the statuses mean

Generally reliable means the source “is reliable in most cases on subject matters in its areas of expertise”, with “a reputation for fact-checking, accuracy, and error-correction”. It is a statement about typical use, not a blanket endorsement.

No consensus means the source is “marginally reliable”, where editors “may not have been able to agree on whether the source is appropriate, or may have agreed that it is only reliable in certain circumstances”. This is the category most often misread as neutral; in practice it means every citation is arguable.

Generally unreliable means the source “is questionable in most cases” and “should normally not be used”. Deprecated is stronger still: consensus exists to deprecate the source and “use of the source is generally prohibited”.

Blacklisted is the only status with a technical mechanism behind it. For those sources, “edits that attempt to add this source are automatically prevented on a technical level”, so the judgement is enforced by software rather than by editors.

Breitbart News shows what the strongest end of the scale looks like in practice. Its entry records that “due to persistent abuse, Breitbart News is on the Wikipedia spam blacklist, and links must be whitelisted before they can be used”, following a 2018 request for comment showing “a very clear consensus” to deprecate it. The entry still preserves a narrow use: it “can still be used as a primary source when attributing opinions, viewpoints, and commentary”.

The distribution is worse than most people assume

Taken together, 164 entries are generally unreliable and 66 are deprecated. That is 230 entries, against 156 rated generally reliable. Among the publications Wikipedia has repeatedly had to adjudicate, the unreliable substantially outnumber the reliable.

This is not a random sample of the press and should not be read as one. A source reaches this list by being contentious, so the population is selected for dispute. Thousands of ordinary local newspapers and trade journals are cited on Wikipedia every day without ever appearing here.

What the distribution does show is the shape of the argument when reliability is genuinely contested: outright rejection is more common than endorsement, and a further 110 entries sit unresolved. Only about a third of contested sources are cleared for routine use.

The deprecated category also demonstrates that these judgements have teeth. The Daily Mail was the first source deprecated, in a 2017 request for comment reaffirmed in 2019, and the entry states it “should not be used for determining notability, nor should it be used as a source in articles”.

The entry count also understates how many distinct judgements the list contains, because outlets are not treated as atomic. Twenty-one publication families carry more than one entry, split by byline type, section or ownership era, and in several cases the entries for a single masthead span three different statuses. The list is closer to a register of editorial conditions than a register of publications.

Where the sources a PR budget produces actually sit

The entries relevant to purchased media are unambiguous, and they are not flattering. PR Newswire is rated generally unreliable, with the stated reason that “press releases published on the site are not subject to editorial oversight”.

Crunchbase is deprecated, following a 2019 request for comment, on the basis that “a significant proportion of Crunchbase's data is user-generated content”. Company and executive profiles on Bloomberg are rated no consensus and are “generally considered to be based on company press releases”, with explicit consensus that “these profiles should not be used to establish notability”.

LinkedIn and Medium are both generally unreliable as self-published platforms, usable only for uncontroversial self-description and “never…for third-party claims related to living persons”. TechCrunch sits at no consensus, with editors noting articles “often have too little independent analysis” to satisfy the depth requirement for corporate coverage.

The pattern across all of these is consistent: the determining factor is editorial oversight and independence from the subject, not the size or recognisability of the brand carrying the content.

It is worth stressing that this is not snobbery about popular outlets, because the list repeatedly rules the other way. Yahoo News and USA Today are both rated generally reliable. Quora, Reddit and IMDb are all generally unreliable. What separates them is not audience size or perceived prestige but whether an identifiable editorial process stands behind the content and whether the source is independent of what it describes.

What this means for anyone buying coverage

If encyclopaedic standing is among your reasons for buying placement, this list is the specification you are actually being measured against, and it is public. Checking an outlet's status before commissioning costs nothing and takes a minute.

The corollary is that outlet prestige and source status are different variables. Several household-name publications appear here at no consensus, and several obscure specialist outlets are generally reliable. A masthead is not a rating.

The list also rewards a more precise question than the one usually asked. Not is this outlet good, but which section, which byline type, and which period — because as our companion report on Forbes shows, Wikipedia frequently rates different parts of the same publication differently.

Two caveats from the list itself are worth carrying alongside any status you look up. The page stresses that “consensus can change” and that “context matters tremendously when determining how to use this list”. A status is a summary of arguments that happened, not a rule applied automatically, and a source rated generally reliable can still be the wrong citation for a particular claim while one rated no consensus may be perfectly adequate for an uncontroversial fact about itself.

Media Morpheus sells media and podcast placement. The status of the wire category in particular cuts against a service we offer, and it is reported here as found rather than omitted.

Methodology and limitations

On 18 September 2026 we retrieved the raw wikitext of the eight subpages that compose Wikipedia:Reliable sources/Perennial sources and parsed every classified entry. We counted status two independent ways — from the status template on each row and from the row's CSS status class — and both methods returned 497 entries with the same distribution, which is the figure reported. A ninth overflow subpage holds 35 additional rows; the project states these are not displayed on the live list pending resolution of software template limits, so we excluded them from all percentages and note the alternative total of 532 for transparency. Blacklisted entries carry both a base status and a blacklist flag and are counted within their base status. Limitations: this list is selected for controversy and is not a sample of the press as a whole, so the proportions describe contested sources only. Statuses change as new discussions conclude, and the figures are a snapshot of one date. The list is also explicitly not a policy or guideline, and individual citations are decided in context by editors.

Conclusion

Wikipedia maintains the most detailed public register of source reliability that exists, and on 18 September 2026 it classified 497 contested publications. Under a third are rated generally reliable. More entries are rated generally unreliable than reliable, 66 are deprecated outright, and 25 are enforced by software that prevents them being added at all. For anyone whose media strategy assumes that recognisable outlets are interchangeable with credible sources, the list is a direct and freely available correction. It also shows where wire releases, user-generated profiles and self-published platforms actually stand when a large adversarial community has to decide whether to trust them.

Frequently Asked Questions

Sources

1. official Wikipedia. Reliable sources/Perennial sources: the list itself, its status legend and definitions, and its statement that it is not a policy or guideline.
2. vendor research Media Morpheus, September 2026. Original parse of the raw wikitext of all eight list subpages on 18 September 2026, counting entry status by two independent methods. Method, reconciliation and the treatment of the overflow page are stated above.
3. official Wikipedia. General notability guideline, for the treatment of press releases and other sources not independent of the subject.