Please commit more blatant academic fraud

So cool. I wish there were more steps in this direction.

> Should these be mandatory?

As an outsider, I would say no. However that should have a detrimental effect on the ability to get published if you are from an academic institution. I can totally see a gray area when it comes to non-academic publishers.

> There's also the question of what code should be made public

Shouldn't it be as much as possible? The way I see it is that the easier to replicate a paper, the more people can focus on its true utility. That is, by abstracting ourselves away over the details of the paper, what does this really contribute? From my point of view, publishing as much as possible maximizes the real-world impact of the research.

Yes, code releasing is a big talking point and most major groups release code. It doesn't solve the subtler problems,like claiming 1% improvements as a solid contribution, unequal hyperparameter tuning, cherry picking datasets, overclaiming the novelty, salami publishing, lacking ablations, not comparing against better methods etc etc. Releasing code is a step but not a panacea. At least it partially protects against fully made up numbers which is not nothing, but it's just scratching the surface. In many cases the code is not so easy to use and the original authors may not be willing to help you with reproducing it, ignoring emails and GitHub issues. If you can't reproduce it, they can always claim something. In one case I was told they rewrote the codebase in a new framework since they generated the results in the paper and there may be small discrepancies due to that. Apparently that was enough change that they were no longer the best in the benchmark comparison. But nobody cares at this point, the paper was published, they are on to their next project. People also have the attitude that it doesn't matter anymore, the field has moved on already, a retraction doesn't change much just hurts the author and the group. The vast majority of papers isn't used anywhere actually in practice, the field is moving so fast that any method can only remain on top for a couple of months.

Also, I don't think AI really is worse at this. It's rather that we are more open about it because we have less to lose and professors don't have an iron grip, due to the existence of industry. In some of academia your whole future depends on whose ass you kiss, it's better in CS. It's a bit like sexism accusations directed at nerds, as if they particularly bad.

The molochian cancer is everywhere. The less you see it may just indicate its more effectively hidden.

Registering before getting data is common, so they can make you agree to some terms. It's not suspicious in itself.

There are no hard and fast rules of thumb. You can look at the group, the people involved, the country, the writing style, the quality and thoroughness of ablation, is there code available etc. But nothing is 100%.

Psycopaths are attracted to power and are willing to do anything for it so they are more likely to get it. Power comes with money. Money allow psychopaths to have more children. The children are more likely to be psychopaths too so they will also end up in positions of power with more money. As automation increases, the system can afford to support an increasing number of psychopaths and they quickly take up all available positions of power via ruthless means. Psychopaths end up occupying all the important political and judicial positions. Altruistic people who follow the rules can't afford to compete in the market place, they have fewer children, they are discriminated against by psychopaths who recognize that they are different from themselves. Altruists become extinct as they are slowly replaced by machines which make altruism a redundant character trait; not necessary for a 'functioning' society. Now that we have machines to automate everything, the economy will no longer rely on 'exploitees' to do the hard value-adding work.

Then a whole new class of ultra-psychopaths (e.g. serial killers) will become more common and proceed to wipe out the regular psycopaths through violent means. Humans will get progressively worse and end up driving themselves to extinction. There's not going to be any robot uprising; humans will make themselves extinct before that happens.