According to your example the non-human agent would gain more by self-modifying to an agent who always writes down 9. Is this not something that UDT agents can do?

You are paying for the classes, i.e. the attention and time of your teachers. Make sure to get your money's worth: if you don't understand something speak up, or contact the teacher after the class. If your class has teaching assistants contact them (for example by email) if you get stuck on the homework/exercises or don't understand something from the lecture. All of these people are literally being paid to answer these questions, be aware that this is a resource you have at your disposal at all times. A common failure mode is thinking: "It's embarrassing to speak up in front of the whole group and/or I don't want to waste their time, I'll just figure it out on my own later" - if you have this thought contact the teacher/assistant after the course about the part that wasn't clear.

I know not that much about astronomy, but I found this thread on Reddit and frankly I agree that the original paper does not spend enough time (the last two paragraphs of section 4.2) discussing the twin star hypothesis. Could anybody with more experience/knowledge explain why this is not plausible in this case? Is the main objection to this hypothesis that it is not clear how such a system would form without a disk of matter present?

Or is does the trick lie in the "stronger than a naive implementation of depth-limited search", and is there some reason why we expect depth-limited search to have sophisticated implementations, but do not expect this for probablistic search?

Something like that I think. The paper suggests that optimizations applied to depth-based search techniques in more sophisticated engines are already effectively like an approximation of probability-based search.

I am confused by section 5 in the paper about probabilistic generation of the search tree - the paper states:

Testing showed that a naive implementation of probability-limited search is slightly (26 +- 12 rating points) stronger than a naive implementation of depth-limited search.

But the creators of the most popular engines literally spend hours a day trying to increase the rating of their engine, and 26 rating points is massive. Is this probabilistic search simply that unknown and good? Or is does the trick lie in the "stronger than a naive implementation of depth-limited search", and is there some reason why we expect depth-limited search to have sophisticated implementations, but do not expect this for probablistic search?

Straight out of the box, the new machine plays at the same level as the best conventional chess engines, many of which have been fine-tuned over many years. On a human level, it is equivalent to FIDE International Master status, placing it within the top 2.2 percent of tournament chess players. But even with this disadvantage, it is competitive. “Giraffe is able to play at the level of an FIDE International Master on a modern mainstream PC,” says Lai. By comparison, the top engines play at super-Grandmaster level.

That's a pretty hard contradiction right there. The latter quote is probably the correct one. Modern chess engines beat any human player these days, even running on relatively modest hardware. That's assuming full length games. At blitz games computers are much better still, compared to humans, because humans are much more error prone.

So if this neural net is playing at master level it's still much, much weaker than the best computers. From master to grandmaster is a big leap, from grandmaster to world top is another big leap, and the best computers are above even that.

Still interesting of course.

Although it's not better than existing solutions, it's a cool example of how good results can be achieved in a relatively automatic way - by contrast, the evaluation functions of the best chess engines have been carefully engineered and fine-tuned over many years, at least sometimes with assistance from people who are themselves master-level chess players. On the other hand this neural network approach took a relatively short time and could have been applied by someone with little chess skill.

edit: Reading the actual paper, it does sound like a certain amount of iteration, and expertise on the author's part, was still required.

edit2: BTW, the paper is very clear and well written. I'd recommend giving it a read if you're interested in the subject matter.

I was always under the impression that the particular case of watching a movie was a strong example of belief as attire - forming an opinion (and discussing it at length) about particular bits of a movie everybody watched is a good way to show that you paid attention to details and are capable of complicated analysis (disclaimer: of course there are genuine movie critics, as well as movies that are actually bad and movies that are actually good. I just think the above explains the most common scenario).

I think your general point is correct, though, but I wouldn't say the phenomenon is caused primarily by sneering. People discount opinions they disagree with in general, which includes sincere criticism, and this effect is further strengthened if they get the impression that the differing opinion is offered with insulting as a primary goal.

Sorry; I don't know why your comment got downvoted so much. It seems reasonable to me.

RQM may not end in an I, but it is still an interptetation.

What the I in MWI means is that it is an interpretation, not a theory, and therefore neither offers new mathematical apparatus, nor testable predictions.

and finally we reject the idea that these observer-dependent representations can be combined to one global representation.

Not exactly, RQM objects to observer independent state. You can have global state, providing it is from the perspective of a Test Observer, and you can presumably stitch multiple maps into such a picture.

Or perhaps you mean that if you could write state in a manifestly basis-free way, you would no longer need to insist on an observer? I'm not sure. A lot of people are concerned about the apparent disappearance of the world in RQM. There seems to be a realistic and a non realistic version of RQM. Rovellis version was not realistic, but some have added an ontology of relations.

In other words, where should we begin searching for maps of a territory containing observers that make accurate maps with QM that cannot be combined to a global map?

its more of a should not than a cannot.

2) What experiment could we do to distinguish between RQM and for example MWI?

Well, we can't distinguish between MWI and CI, either.