Nobody is on call. Everything else is on the record.
The offer on the rest of this site is “clone it and run it”, which means running a stranger’s script on your own machine. This is the stranger.
Two people have ever committed to this
- Taha Mahmoodi
- writes the code, merges the pull requests, edits the catalogue,
and reads the email. There is nobody to escalate to.
github.com/Taha-Mahmoodi - One other person
- has commits in the public repositories. Their contributions are listed on each repository’s contributor page, which is public and linkable, and that is the whole of what we will say about someone who has not agreed to be described on a company website.
That is the list. Across every public repository in the organisation, no third person has ever contributed, and nobody from outside the organisation has contributed at all. check this claim →
What our own GitHub profile said instead
It said “built by our team and contributors who earn it.” That has never happened. Nobody outside the organisation has ever contributed to any of these repositories.
The organisation lists eleven members and a developers team of ten.
Eight of the ten have never opened a pull request. Our
CONTRIBUTING.md names six roles and says each is backed by
a GitHub team; three of those teams exist. We wrote the table we wanted
before it was true, and it is still not true.
None of it is on this site. All of it is on /claims,
struck, with the date we checked it.
check these claims →
Six things this site will not do
- No star counts.
- Ours are small and a competitor’s are large, and neither number is a fact about whether the software works.
- No customer logos.
- We have no customers. A logo bar is a wall of other companies’ trademarks and most of them on the web rest on an invoice.
- No screenshots of things that do not run.
- Nothing we have built is deployed at a URL you can click. A product shot would be a picture of a local development machine.
- No photographs of people.
- There are none on this page. We do not have a photograph of the two of us working, and we will not generate one. A generated face attached to a real name is a lie about a person.
- No number we cannot reproduce.
- Every count on this site comes from a command a stranger can run. When we could not agree on a counting rule, the number came off instead of getting a footnote.
- No cookie banner.
- There is nothing to consent to. This site loads no analytics, sets no cookie, and pulls nothing from another company’s server. The banner is missing because the tracking is, and that is the cheapest way to comply with anything. The one exception is the form on /hosted: submitting it sends your address somewhere, and /legal/privacy names where.
We publish the corrections, including the ones that cost us
Our research document on a competitor had a stronger claim available and our own research did not support it. The document says so, in its own text, in the paragraph where the stronger version would have gone. We wrote that before there was a website to be embarrassed on.
Four things went wrong while we built this site. All four are published:
- We copied a number that had rotted.
- Our research said a competitor had about 17,600 stars. It has 26,482. The document carrying the wrong number also carries the instruction to re-verify anything load-bearing before acting on it, which is the instruction we did not follow. check this claim →
- The number that undercut us was our own.
- One README claims 5,983 lines across eleven files. They total 5,493. It is one of the few numbers our own checking script does not guard, which is how it survived. check this claim →
- Our profile claimed a product was running that is not.
- It said one of the eight was shipped and running inside another. Searching the whole of that repository finds the name in three markdown documents and nowhere in the source. There is no deployment anywhere in the organisation. check this claim →
- Our profile claimed a gate that is not enforced.
- It said a pull request does not merge without a passing check. The check runs. The branch it merges into is not protected, so nothing is required, and the allowlist exempts the only person who merges anything. check this claim →
Every one of those is a row on /claims, struck, dated,
with the command that disproved it. The page was built for this and it
lists what did not survive, because a claims page that only lists the
survivors is an advertisement.
One address. One person reads it
There is no queue, no rota, no target response time, and no contract that promises one. If you write, Taha reads it. If he is asleep, it waits.
What replaces the promise is the record. The CI logs are public and timestamped by GitHub rather than by us. The issues are public. The commit that fixed a thing is public and dated, and so is the gap between the two.
If that is not enough support for what you are about to do with this, then it is not enough, and you should not do it. That is also the honest reason there is nothing hosted: holding other people’s notes is an on-call obligation and we have not taken one on.
Read the commit history — public, timestamped by GitHub, not by us
The proof of method is the collection — receipts at /claims.