Dragons in the Algorithm
Adventures in Programming
by Michael Chermside

Faster than the Speed of Falsification

Lots of science writers are getting very excited about the news that an experiment measured neutrinos going faster than light. But as we all know, they are overreacting.

Source: http://arxiv.org/abs/1109.4897Look, this really shouldn't surprise anyone who has ever done any actual science. Karl Popper had a really simple and idealistic concept of how science works: you make a theory, use it to make a prediction, run some experiments to verify the prediction, then if the theory keeps getting it right then there's probably something to it.

Beautiful theory, but in the REAL world any experimenter spends 90% of their time trying to get the experiment to give the right results. It is actually quite challenging to figure out where the boundary is between getting your apparatus to work properly and just fiddling with it until you get the result you expected. A good experimentalist understands that boundary and spends a lot of time worrying about possible sources of errors.

That is why I think the paper that started all this furor was a GOOD paper to publish. Not because I seriously believe that neutrinos go faster than light... and I doubt that the authors of the paper actually think that either. But because it is a very surprising result, and something unexpected is going wrong here. Are our measurements of position from GPS less accurate than we think? How about the time measurements? These are the things that the paper focuses on, because they are the most important part. Not to publish this just to avoid looking foolish -- that would be a real loss (and biased). So the media will just have to have their field day.

Posted Mon 26 September 2011 by mcherm in Science