As someone who worked with Twitter’s APIs a lot, and the metric shitton of metadata that is sent with every tweet, him thinking it’s 100 bytes is hilarious. I know Twitter typically does assessments of a client’s infrastructure before providing them access to the Firehose, to ensure they are actually able to keep up with the stream. It’s that much data.

He’ll likely have to sample the data anyway before he can do any meaningful analysis, so the whole exercise is pointless.