Re:Fake Physics (2017)
by whatshisface on Saturday November 24, 2018 @12:53PM
(#18523135)
Rejecting strings on aesthetic grounds is no better than accepting it on aesthetic grounds.
[ Parent
| on HN ]
The problem is not asthetics the problem is overfitting the data. Add enough knobs and you can approximate anything, but your simply encoding data not finding anything new.
Re:Fake Physics (2017)
by whatshisface on Saturday November 24, 2018 @03:09PM
(#18523761)
There is an infinite amount of complexity in a randomly selected real number. A famous example of packing lots of behavior in to a small number of parameters is [0]. A physical constant that will change everything drastically if it is changed even a little bit carries just as much information as one hundred constants to which the universe is less sensitive. That's the argument behind why the fine-tuning in the standard model is an indication that it is not sufficient.
[0]https://fermatslibrary.com/s/drawing-an-elephant-with-four-c...
That argument falls flat in two ways, first we are dealing with 12-13 digits so taking about infinity is silly.
Second, the standard model is simply the most ifnomation dense way of describing ~all known observations. Nobody thinks it’s truth just accurate. So, saying it overfits the data is fine, that’s how it’s constructed. Supplanting it needs to reduce the amount of information required to describe the model or similar amounts while being more elegant to call it progress.