GigLookup is a directory of gig and independent work across all 50 states and Washington, DC — 363 listings, checked one at a time against each company's own pages, with the date and the source on every card.
The short version: We read each company's own pages ourselves — not aggregators, not articles. Every listing shows how to apply, when we last checked, and what we couldn't confirm. Nobody pays to appear, to rank higher, or to be described more kindly. Anything a company charges you before you earn shows on the card in red, above the earnings talk. And if something's missing, that means we haven't researched it — not that it doesn't exist. The long version of all of this follows.
It was built by a contractor who needed this information himself and could not find it anywhere in one place — the listings that turn out to be employee jobs, the ones that quietly charge you before you earn anything, the ones whose sign-up page died two years ago. This site is the list he wanted.
It means one specific thing, and not more than that: somebody fetched the company's own page and read it. Not a press release, not an aggregator, not a jobs-site summary — the company's own words about who may apply.
Every listing says exactly one of these, and the wording is enforced rather than chosen:
The word yet appears on none of them. It used to appear on records we had already checked, which promised a door no evidence supported.
If something is not here, that means we have not researched it yet — not that it does not exist. This is a list of what has been checked, and it is nowhere near a list of everything that is out there. A gap is a gap in our work. Pages for areas we have not covered say so in those words rather than showing you an empty result and letting you conclude there is nothing near you.
Not every area has had the same amount of research, and the results page for each area says which it got. Three levels, assigned from our own research records — never from how good the market looks:
Where our records are ambiguous between two levels, the area gets the LOWER label — an unsure label must claim less, not more.
Behind each label sit counted research passes ("sweeps") over a versioned list of work lanes, and each area's record names its passes and their dates. That bookkeeping used to print on every area page; it lives here and in the research records now, and the area pages keep the three things a reader uses — the level, this explanation, and the last-researched date.
483 claims are marked as unproven. They came from an older research method, and we have now re-read every source that can be read — the number stood above nine hundred before that work. What remains is not a backlog; it is a boundary, and here is what it is made of.
283 of these claims sit on records that hire through job boards or whose pages have gone dark — there is no page of theirs to re-read. The other 200 sit on records with live pages that still cannot pay their claims: pages that show their content only inside a mobile app, or show nothing to an automated reader; sites that block automated checks (each attempt dated and retried — and those blocks do sometimes lift: one did this week); and 59 pay-basis claims we keep on purpose.
That last group deserves plain words: we could make this number smaller today by deleting claims that are probably true but unconfirmed — a listing's pay basis, say, that its own pages simply never state. We don't, because that would trade what the card tells you for a better-looking statistic.
What an unproven claim means for you: it's a detail we haven't re-read from the source. If that detail matters to your decision, check that specific thing on the company's page before you act.
This number now moves when the world or our rules change, not by grinding: a source publishing something it doesn't publish today (one waitlist page changed this week and its claim resolved), a wall coming down, or a research method we haven't ruled on yet. Publishing the number — and what it is made of — is deliberate: a limitation nobody can see is one nobody ever fixes.
A further 106 claims about metro-level operating status are carried in their own declared class: derived from our own coverage research, not from the company's words. Companies do not publish their metro lists, so these can never be confirmed from a provider's page — counting them as owed re-reading would overstate a debt that cannot be paid that way, and presenting them as provider-sourced would be worse. They say what they are, everywhere they appear.