Back to archive

Thread

4 tweets

1
And yeah, it is not *terribly* much of data, but I guess if you want to do any more long time analysis, my guess is it would be more in the 10s of TB of data. Put it in S3 buckets, let a big Spark cluster run against it, it‘s not impossible, especially if you have the money.
2
@srchvrs I mean back in TWIMPACT times we were doing trend analysis in real-time on a single machine, but that‘s another story :)
3
@srchvrs IMHO the futility of the whole exercise lies in that it will be both hard to agree on what a bot is, and Twitter already said they also only consider specific kinds of users. Unless Elon‘s team agrees with that and replicates all of that it will be like oranges with apples. twitter.com/paraga/status/…
4
@srchvrs And if they replicate I‘d expect they get the numbers that Twitter reported (or there is a bug or in fact Twitter has been lying). So it will probably come down to either side saying that their definition is the right one, etc. Ah well, what a waste of resources.