Half a croissant, on a plate, with a sign in front of it saying '50c'
h a l f b a k e r y
non-lame halfbakery tagline

idea: add, search, annotate, link, view, overview, recent, by name, random

meta: news, help, about, links, report a problem

account: browse anonymously, or get an account and write.



Punishable operating system

Punish a computer until it changes its ways
  [vote for,

This is probably very naive and may be elsewhere on here but i haven't found it. My lack of understanding of computers makes it difficult for me to express this clearly or sensibly.

The operating system constantly monitors what happens on the computer, in terms of processes, memory used, speed, drivers employed and so forth. In many cases, there are at least two options for doing each thing which perform the same task as differently as possible. Each one starts off with a probability rating associated with it, initially one in two if there are two such tasks. Before a task is run, the operating system generates a pseudorandom number which decides which way it will run the task concerned.

Now, suppose the computer does something "naughty" like run slowly, losing data, making a silly noise or displaying something weirdly, or it does something "good" like running particularly smoothly, exiting an application quickly, playing media non-choppily and the like. There are two keys on the keyboard labelled "PUNISH" and "REWARD". Pressing the former reduces the probability rating for the specific ways of doing the tasks concerned and pressing the latter increases them. Failure to complain also increases the probability marginally. This way, tasks which seem to interfere with the running of the computer are weeded out, leaving those which work well, and bugs are sorted out without conscious intervention.

I'm sure this is bollocks, but could someone explain why please?

nineteenthly, Dec 31 2011


       Is this a genetic algorithm, where the "fitness" of each mutation is defined largely by reference to the current mood of the user?   

       If so, wouldn't it require a very patient user to spend a long time responding to randomly sub-optimal behaviour before the algorithm even reached a break-even point?
pertinax, Dec 31 2011

       [Quest], the problem isn't always RAM even when it is. I have no idea if this happens, but i get the impression that launching programs in different orders makes a difference to performance, and i wonder if this is to do with something like a program claiming memory belongs to it and then not relinquishing that claim when it exits, or memory locations not being consecutive but being treated as if they are, so that the RAM gets fragmented in the same way as backing storage would. As i say though, i have no idea if this is true or if it ever happened.   

       [Bigsleep], i realise it's not very specific but i think the usual situation is that users just have no or little idea why there's a problem, even if they are well-informed about computers, just because so much stuff is hidden, so it's hard to see how it could be made specific.
nineteenthly, Dec 31 2011

       Those are the first things which come to mind. I just wish i understood what the problem was. It seems to go beyond all reason. However, that's too general for here. As it happens, you remind me of another idea i had.
nineteenthly, Dec 31 2011

       [19thly], I think you need to buy more processing power/RAM. Newer iterations of software are written on the basis of ever greater resources being available, in a cruel parody of Moore's Law.   

       Any machine with less than 4GB of RAM is really behind the game, if it's using recent software. 16GB of RAM and upward, coupled with a dual or quad core 3GHz+ processor provides a far more pleasant computing experience for not a lot more capital outlay.
infidel, Jan 01 2012


back: main index

business  computer  culture  fashion  food  halfbakery  home  other  product  public  science  sport  vehicle