RE: Steem must improve and the key metric for that improvement is "utility"
Thanks @dana-edwards, for your always thoughtful replies (that distinguishes you from many others, by the way). Agree with you, in addition one reason why people can´t be rational is, that they usually don´t have 100% information to judge rationally,most miss the full picture / all stakeholders views, etc.
I am not that optimistic if whole communities act together more rationally. Sure, outliers and morons are averaged away. But if people do not know the big picture, communities don´t know it necessarily as well. Besides there often form parties in communities, which stalemate themselves by striving in opposite directions in worst case. Often opinions are biased due to emotions or manipulations (there are whole books about mass manipulation). At the end a lame compromise is pursued due to conflicting interests or mere beliefs.
Manually you are correct because human brains cannot scale. I do not advocate we do things manually. Once we apply machine intelligence to the Steem data already generated we can make magic happen. This could indeed be an SMT and represent the data analytics or even predictive analytics functionality of Steem.
Right, I saw your post about this. I truely hope this will work out, but so far machine learning is rather a big, "sexy" promise and remains to be seen, if it can hold what people expect. It is only so good as the assumptions and goals are that are behind the algorithms and the interpretations (similar to Statistics, just neutral numbers, but people do interprete them to their needs). Probably the profit maximization can happen only, if real AIs step in? But who will then profit most? Still the humans?
The more data you have to feed it the better the learner gets because it works on statistics. More examples, better more efficient learning. So I think Steem data can be useful for improving Steem.