PhD advice

Ravin Balakrishnan of our department gave a talk on "How to succeed in Grad School". I think he's a very good presenter and certainly knows what he's talking about.
For my memory, I am noting his most notable points:

1. Your supervisor is not there to supervise you, but rather to give you advice. Therefore he should be called "advisor". Every PhD student can only be responsible himself for what he is doing. (Ravin instantly convinced me of this and I am now using the term "advisor" exclusively.)
   
2. Publish only in first-rate conferences. Second-rate is just not worth it.
3. The "related work" section in your paper is to pay respect to those people who are most likely reviewing the paper. This is not a dirty trick, since there is too much citable related work out there anyway; you are just making a smart selection.
4. Choose a risky research area. If it does not have possibility of failure, it is more likely to be development than research.

I have tried to evaluate my own research project according to its risk and I find that there is actually a lot of risk because I am claiming that my method will be able to do something which many people say is just not possible. (Namely, I am claiming that the grind-work of formal proof which others say has to be done automatically can in fact be done by humans and not be utterly boring. I am claiming that this actually trains humans and helps them also formally prove other things which computers can not (yet) automatically prove.)
But I have to admit, that I would probably not have the courage to pursue this risky road, were there not a safety line: no matter how well my little proof tool will fare in research, I think that it has relatively safe chances to be useful as a teaching tool. That's why I can sleep well at night. And still dream about and hope for some modest success as a researcher.