Retraction: My information about Mastodon reaching 4 million registered users was based on the @mastodonusercount bot, but I am afraid upon closer inspection its data is incorrect.

The merit of that metric is debateable in the first place. What really matters is MAU (monthly active users), not total registered users, but most stats aggregators only collect/focus on the latter, and as such, that's what people see and keep track of and get excited about. And my take on it is, if people get excited about it, why not? It's still a milestone, you just have to understand the nature of the milestone.

That being said the nature of inaccuracy of that particular bot's data is doubly debateable. The number does align with what an alternative stats aggregator (fediverse.network) reports for Mastodon, however both include an infamous server that imported a large number of user records from its pre-Mastodon past. So while that server technically is running Mastodon software, the source of that number is different and should not be included in ours.

Then again you might argue statistics on anything fediverse-related will never be 100% accurate. Any stats aggregator relies on discovering servers to ask and those servers reporting truthful numbers back. There's always going to be deviations because one moment a server's here and the next moment it's having downtime. Then it's back again. With thousands of them. Of course some people have multiple accounts, and some accounts don't represent people.

It's a unique flavour of a problem but not a unique problem... How would you know if a centralized service's number were accurate? A large plot point of HBO's Silicon Valley was a startup artificially inflating their user numbers with fake activity to placate investors.

🤷‍♂️

@Gargron There are statistical methods dating back to World War 2 that can likely be applied. Mathematicians came up with much better estimates of things like tank production compared with spies trying to figure out the same thing.