[pdx]

I agree. Just buy them. If they spend $100K a year on buying fake accounts and then banning those accounts, that will be more effective than messing about paying a team of 10 engineers to do pattern analysis. If they spent $2M a year on buying and then banning fake accounts, they would squash it completely.

They're not doing it because they like the big follower counts themselves. They're in on the con.

[stouset]

How does paying the botters market rate raise their cost of entry?

[kibwen]

The price of a single bot is amortized across thousands upon thousands of follows. If it costs a penny to have one bot issue one follow, and a single bot performs 5,000 follows, then that bot has earned $50. If Twitter were to buy follows for the market rate (one penny) then ban those accounts, it would drive the market rate for bots up to (using our hypothetical example) $50 per follow. (This is an oversimplified example, of course, but reducing the average number of follows that a given bot can perform would indeed have the effect of raising the market rate, though we can continue to argue over the magnitude; it's self-evident that if generating bots didn't have fixed cost overheads, then we'd see fewer bots with incredulously inhuman numbers of followed accounts.)

[jaggederest]

This seems liable to cause unintended side effects...

https://en.wikipedia.org/wiki/Cobra_effect

[pdx]

I disagree. The cobra effect occurred because the English provided a market for cobras where there was no market before. There's already a large market for followers and twitter adding itself to that market in such a limited way does not add significant new demand.

[AlexCoventry]

The Reagan administration tried this in the Iran-Contra affair. They only succeeded in incentivizing more terrorism and abduction of US citizens.