Saturday, October 22, 2016

Paper written by iOS autocomplete accepted for conference

Nonsense paper written by iOS autocomplete accepted for conference

Some comments I just made in an email about this :

Ouch! "Unreasonable" recurrent neural networks will be doing all our jobs soon. (https://karpathy.github.io/2015/05/21/rnn-effectiveness/)

What happens when we have automatic tools to summarize news items, journal articles, and perhaps provide early filters (routing-queries) in "what's interesting" from the journals. Perhaps lazy reviewers will just run the filter over a paper to see if its worth reading.

And then if we use the same models to generate papers ???

Just like automatic high-speed trading algorithms go off and autonomously create bubbles and crashes in the market, could we see runaway feedback between automated science paper generators and automated science paper accepters create entirely new subfields of science / pseudoscience?

I suppose, given that this is a physics conference, it's a great response to the Sokol Affair.

OTOH, sadly, it seems like this is a junk conference put together by "open access" publishers. The terrible thing here is that the term "open access" which should be a term of approval, is now getting discredited as vanity publishing.

I don't quite suspect a conspiracy by academic publishers ... but ... it is awfully convenient.  :-/

Saturday, October 08, 2016

Servers down

It seems all my servers are down ... giving DNS errors. Some problem with the name provider I think ...
Will keep you all posted.