Maybe something simpler with some subset of the functionality that some subset of users care about. Your point still stands about taking a long time to suck up the data. Also, now you'll probably get thrown in jail for doing 1% of the shady evil shit FB have done (and continue to do!).
I'm aware of that. The question is: how much of this functionality is actually used and appreciated by the users?
> Maybe something simpler with some subset of the functionality that some subset of users care about.
I'd venture to say "the majority of users care about". I did some amateurish research a while ago asking people how they are using FB. Most just scroll the feed, chat via FB, click "like" and occasionally comment; some participate in FB groups. There is a ton of functions rarely used, and some are deliberately broken (for example, FB deliberately removed past birthday notifications to create a sense of urgency to log on more often, and you need a workaround to catch up if you log in less often).
If you actually start using HumHub and compare it to FB, the former is a real pleasure to use. Will it ever have more users? Practically speaking, it's impossible. In any case, my point is not how fast one can create a usable FB clone, but that the code itself is no longer something that need to be jealously guarded.
But they got that user data, though having the code. The code is still the essential bit in the first place. If someone else had had the Facebook code before they did, they could have gotten the data.
It's a good clean data set to train ML models that is the current proprietary advantage set.
It's because arguably in the past (80-90s) the highest value was with the code, and it was it that was jealously guarded. Now the code doesn't have that much value and there are clones of almost everything, and the weight shifted towards the piles of user data. In other words, if I'm motivated enough, I can create a Facebook clone (with less useless functionality but arguably better user experience) in 2-3 months, but I'll never be able to gather a fraction of the user data they managed to get.