Me: i am going to kill senator tom cotton
CIA Agent: ohhohohoho. we got him this time.
Me: in minecraft
CIA Agent: GOD DAMMIT. SHIT. FUCK. *punches a hole in the drywall* FUUUCK
SCI-HUB NEEDS YOUR HELP!
Do you have a few dozen gigabytes of free diskspace and a machine that can seed it via #bittorrent?
You can help save an unbelievable trough of scientific knowledge from disappearing behind the elites' paywalls!
Spread the word and seed till you bleed – for the betterment of humanity!
ironically while software companies bite their fingernails over how to ensure they are getting paid for their products the vast, VAST majority of their development work relies on unpaid maintenance of "free" software and this problem would evaporate if you simply had, say, a nationalized or otherwise publicly funded entity to perform these critical maintenance tasks
it's true that classical widget-based economics doesn't apply very well to software, which is expensive to develop but practically zero-cost to copy. in a sane society you would just take this as a sign that should be the domain of the commons instead of erecting 10000 component rube goldberg machines of laws and software protections to enforce the profit motive
If anyone's curious, basically the problem is that you have a graph with two distinguished vertices s, and t. Each edge has some probability to be deleted (independent of each other), and afterwards you can ask if there's still a path from s to t. The original problem was just to simulate a single trial for a fixed graph (easy, linear time), but I wanted to see if I could just compute the probability beforehand. The inefficient solution uses a deletion-contraction recurrence, and so its runtime is exponential in the number of edges. Also interestingly, the result is a polynomial in the edge probabilities, almost like the Tutte polynomial but not quite because of the two distinguished vertices.
Hm so today a programming problem for an interview got me thinking "how could I cheat and use fancy math to pre-compute the result" basically. I spent the day thinking about it, came up with a very inefficient solution, looked it up, and it's a special case of a problem called s-t reliability, which is #P-hard in general.
dark mathemagician | 23 | they/them
Working on a math PhD, mainly algebraic geometry. I also do things with computers sometimes.