Who maintains the index, how entries are written, and how mirrors are checked.
IAPSO2017 is a software index. It exists because finding a working copy of a specific version of a specific program should not take an afternoon of clicking through pages that promise a file and deliver a survey. Every entry here is a page describing one release, with the version string, the archive size, what changed, what it needs to run, and four mirrors that have all been checked.
A small group, none of whom do this full time. Some of us maintain the checker, some write entries, one person handles the takedown mailbox and the contact queue. There is no company behind the site, no office, and no funding beyond what covers the mirrors. That has been the arrangement for years and it works because the workload is spread thin rather than concentrated.
An entry starts either from a request or from a checker noticing a newer build of something already listed. Someone pulls it, installs it on a clean machine, confirms the version string, notes the actual archive size rather than the one the source claimed, and writes the page. The overview, features, requirements and installation notes are written by whoever tested it, which is why they read differently from entry to entry and why they occasionally contain an opinion.
Nothing is scraped. Scraping would let the library be enormous and would make it useless, because a scraped index is a list of links that were correct once. The whole value here is that the page and the file agree with each other.
Every entry is served from four independent mirrors. A checker pulls from each of them on a rolling schedule, confirms the transfer completes, confirms the archive opens, and confirms the internal version matches the page. Two consecutive failures pull a mirror out of rotation and raise it in the queue for whoever is on duty. Entries do not sit on this site pointing at routes that stopped working eighteen months ago.
Because every page is written by a person who installed the thing first. The catalog grows by a handful of entries a week and most of that growth comes from the requests page. If you want something that is not here, ask for it rather than assuming it was considered and rejected.
The contact page reaches the general queue. Takedown notices go through the DMCA page instead, because they are handled by a different person and a different process. Requests for new titles belong on the requests page so they end up in the queue rather than in an inbox where they get lost.