Comment by xnn on Simulate and Defer To More Rational Selves · 2014-09-09T09:11:48.757Z · LW · GW

I wish this were posted in main.

Comment by xnn on Fighting Biases and Bad Habits like Boggarts · 2014-08-23T10:42:24.482Z · LW · GW

I wish this (and more like this) was posted in main.

Comment by xnn on Ethical frameworks are isomorphic · 2014-08-14T09:21:31.434Z · LW · GW

(ducks before accusations of misusing "isomorphic")

Old joke:

Q: Are these two objects isomorphic?

A: The first one is, but the second one isn't.

Comment by xnn on Open thread, 11-17 August 2014 · 2014-08-12T21:23:34.414Z · LW · GW

See Coupon collector's problem, particularly "tail estimates".

Comment by xnn on Separating university education from grading · 2014-07-04T08:20:36.447Z · LW · GW

There are absolutely differences in the perceived value of the same degree from different institutions. I've also heard of deparments that have received complaints from businesses that the quality of graduates with a certain degree has dropped due to grade inflation. That has led to a painful period of lower results as they try to correct the problem.

Comment by xnn on Separating university education from grading · 2014-07-04T08:17:43.163Z · LW · GW

If the lecturer happens to be a postdoc (which is not unheard of in a final year course, for instance) then they are also likely to do the marking. I agree that exam marking is unlikely to be farmed out to postdocs who are otherwise not involved with the course.

Comment by xnn on Separating university education from grading · 2014-07-04T08:15:22.159Z · LW · GW

The draft examination papers are sent to the external examiners, who make comments as they deem appropriate. These might, for example, be "this question is too hard" or "this question is not clear". The final approved exams are then taken by the students and marked by the lecturers who set them. The external examiners then return during the moderation process, helping to decide where the boundaries for various classifications should be set.

Comment by xnn on [News] Turing Test passed · 2014-06-09T09:12:56.029Z · LW · GW

Side note for those who might not have come across him before: Kevin Warwick is a Professor of Cybernetics who, among other things, has communicated with his wife through electrical cables attached directly to nerves in their forearms.

Comment by xnn on Open Thread February 25 - March 3 · 2014-02-25T21:38:53.653Z · LW · GW

The other videos I've sampled from that channel have also been good.

Comment by xnn on Effective public college tuition vs. private college tuition · 2014-02-14T23:41:49.577Z · LW · GW

The UK university system is of course very different from that of the US, but there's a similar effect. Tuition fees are roughly constant, but Oxford and Cambridge have generous bursaries for students from lower income families and subsidised housing, so are generally cheaper than the less prestigious (and less wealthy) alternatives.

Comment by xnn on Cambridge (England) lecture: Existential Risk: Surviving the 21st Century, 26th February · 2014-02-14T20:46:54.076Z · LW · GW

You should be able to get it listed in the University and College weekly emails. (Perhaps you already have pet undergraduates to do this for you.) Subject societies will often advertise talks likely to be of interest to their members too. There's also What's On, and the old fashioned approach of taking posters round the departments.

Comment by xnn on Open Thread, December 2-8, 2013 · 2014-01-13T22:54:53.785Z · LW · GW

UK banks lost a test case a few years ago that led to a lot of people getting back however many years of overdraft charges, plus interest. The same thing happened a bit later with "payment protection insurance", intended to cover loan repayments if you lost your job, but with so many exclusions as to be almost worthless.

The end result was something like a forced savings policy. Cue people who avoided the initial trap wondering where their free money is.

You have to wonder sometimes.

Comment by xnn on Naturalistic trust among AIs: The parable of the thesis advisor's theorem · 2013-12-15T20:21:50.284Z · LW · GW

"therefore, any serious AI needs a formal system with more oomph than PA"

The problem with that is that the same argument goes through in exactly the same way with any stronger system replacing PA. You might first try something like adding a rule "if PA proves that PA proves S, then S". This solves your original problem, but introduces new ones: there are now new statements that your system can prove that it can prove, but that it can't prove. Eliezer discusses this system, under the name PA+1, in You Provably Can't Trust Yourself .