namshub

And all the while machine learning engineers explicitly work to make them more addictive. At least tobacco companies didn't have a monopoly on information and they still held on to plausible deniability for as long as possible. This a serious problem that also affects our capacity to solve the problem.

They just fit a sigmoid to it, that isn't a biological limit.

Selective ignorance is a dangerous strategy... Far better to evaluate the facts as fully as possible and find the actual reasons for and against. However a straightforward argument for eugenics is probably not very accurate or broad in its consideration. In that sense I guess it could be inaccurate or true but incomplete and/or misleading on its own.

Pop in an 'on average' and I would agree with you there. Bearing in mind that the intra-group variance is probably comparable to the difference in the means of the groups.

It's not just the extraction of data, it's also the optimisation methods:

- Make it a legal requirement to display on the website what optimisation techniques are used and what metrics are optimised.

- Create recommendation systems/search engines/web browsers as a public good, with utilities that are more humane and as such reduce the effects of addictive and harmful software in the medium term.

- Legislate on psychoactive software with the same fiduciary responsibility and oversight as applied to doctors and surgeons.