Is Musk vTwitter a $44BN data governance failure?

Will they or won’t they? A love match has turned dysfunctional. Musk has discovered the body unclothed is not attractive. Jilted Twitter refuses to walk away from the altar, and is attempting to force the nuptials. Why?

Musk’s stated in April his top priority upon purchasing Twitter would be to “…eliminate spam and scam bots and the bot armies that are on Twitter. They’re making the product much worse…”. Hooray! Musk wants to:
1. Ensure that users are real people, authenticated (but not necessarily identified)
2. Improve the quality of the customer data
3. Increase the trust in the numbers, which in turn increases the confidence in decision making
4. Stop algorithms being manipulated by bad Bot accounts
All of which serve to
5. Improve the overall experience and engagement of Twitter users

So, Musk’s top priority is to improve the product by focusing on sustainable data, clear data governance and strong data management.

Twitter has estimated less than 5% of accounts to be bots, and not all bad. With 291M active users that works out to 15M - for context, that’s more accounts than the population of Qatar, who are part financing the deal. Twitter has stated this 5% cannot be verified externally - and it has said this well before Musk’s offer.
Musk is now trying to walk away from the deal due to the unverified 5% - a data governance failure.

I agree with Twitter that the number cannot be verified externally but it can be verified. I am also sure the answer will not be found in the 9k sampled records ordered by the Judge to be handed to Musk*. If the answer is to be found, then these 9k records must be engaged accounts, with an identifiable pattern of behaviour distinct from humans, however this is a low sample number that doesn’t inspire confidence. The answer will also not be found in access to the daily live stream of Tweets (aka the “fire hose”) which Twitter offered Musk.

The true answer will be found only in… live verification on current data with a benchmark acceptability of good bots. Not a Botomoter, but the introduction of a process across a statistically robust volume of engaged accounts to ascertain they are human owned. This is not just a ‘reCAPTCHA’ - this could be a combination of image, callback, multiple accounts check, an intelligent algorithm on the Tweets/accounts themselves. This should be the Judge’s order to Twitter. Or perhaps, Twitters order to themselves.

Musk knew about the Bot issue before the offer, and it’s not new in the industry (Instagram has an estimated 95M). Will any number be good enough for Musk? history has shown that what Musk wants, Musk gets. So, let this be a lesson instead on the importance of investing in data governance - you could have billions riding on it.

*Whether the judge can authorise 9k accounts to be passed to Musk without permission from the account holders themselves is another question

All views my own.

Sources:
1. The Verge: Elon Musk’s Twitter Plans are a Huge Can of Worms
2. The Guardian: Judge orders Twitter to turn over to Elon Musk data from 2021 users audit
3. NYTimes: What Are Spam Bots and Why They’re an Issue in Elon Musk’s Twitter Deal
4. Ars Technica: Botometer creator says Musk’s Twitter spam estimate “doesn’t mean anything”
5. The Information: Instagram’s Growing Bot Problem