My mother has threatened promised to call tonight.
So I was making a mental list of what I can tell her about when she asks how I spent the weekend. The problem is that I don't think she will be able to identify, let alone identify with, any of the things I could mention:
- I blogged. (My mother doesn't know I have a blog. I doubt she even knows what a blog is.)
- I collected some posts for the Carnival of GRADual Progress. (See note to #1.)
- I read an article called "Parataxe, Hypotaxe et Corrélation dans la Phrase Latine" [Parataxis, Hypotaxis and Correlation in the Latin Phrase] (Let's just pretend that was a helpful translation, okay?)
- Geekman and I installed a different distribution of Linux on my university computer. (My mother is uncomfortable with the idea of using Firefox as a web browser because she "likes Windows better".)
- I memorised the pronominal system of Middle Egyptian. (Do I even need to comment on this one?)
- I designed a computer program to randomly suggest combinations of the contents of our fridge, in the hope of generating ideas for exciting new sandwiches. I wanted to design an evolutionary algorithm in order to evolve the perfect combination, but got stuck at the point where I needed to create a metric to measure goodness of sandwich.
15 Comments:
It's such a challenge to put one's day-to-day doings in general terms, isn't it?
I tend to get that in a sort of half-half way. I can tell my mother if I did lots of baking/cooking, and she'll be all "wow, that's great, I have to try that recipe", and I can tell my dad about my uni work and the like. But I can't put it the other way around, or there is a deathly silence on the other end of the phone, leading to a "what is...?" question.
Any relatives further removed though? I stick to similar stories as the "I went for a walk" line.
Ohmygod. You need to tell us more about #6. That's freakin' hilarious.
Miss M., at least your father can handle stories about your uni work. Mine is pretty much on the same wave-length as my mother.
As for the sandwich generator, I'm not sure there's more to tell, Dr Brazen Hussy. You input a list of what's in your fridge, the program creates a few random combinations of ingredients and compares them, ranks them, and then the best ones get to cross-breed by swapping ingredients, which creates a new generation that gets ranked and bred from, etc, until you have the perfect sandwich. (Occasionally ingredients get to mutate too, i.e. mayonnaise might be randomly replaced by thousand island dressing.) Only I can't decide what criteria the program should rank the sandwiches on.
The only obvious thing I could think of is "sogginess" as a negative thing. Which would have to be defined by each ingredient having a sogginess value associated with it, and over a certain sogginess threshold, the less soggy sandwich wins. But I couldn't really come up with any other metrics.
Maybe the user would just have to manually evaluate each pair and select the winner to "breed" for the next generation.
Oh my! If only I could even start to write a program similar to #6! Quite an interesting feat and thing to do. Yeah, I think I would tell my mom I went for a walk as well. :)
#6 sounds awesome. I always have trouble thinking of things to tell my mum on the phone, too.
Tell her about MI3 and katie holmes. :)
Your program continues to intrigue me. I often speak of the ever-elusive "deliciousness factor" and adjustments to dishes that raise said deliciousness factor. But indeed, how would you program a computer to recognize it? Interesting.
Yes I think the user has to sample it -- that's the external environmental pressure element of natural selection. Of course the user could simply simulate this by imagination for the most horrible combinations.
I think it's really important that the computer doesn't do the evaluation. Otherwise you are permanently limited by your pre-defined criteria. The wonderful thing about human evaluation is that the criteria could be anything (e.g. easy to hold while driving to work). That's like real life -- natural selection could be due to anything.
Something to bear in mind is that there's no one great sandwich -- there are many. Even one person may have multiple favourites. However, you could have multiple people evaluate each sandwich. Eash person would represent an ecological niche -- for example the ecological niche of olive lovers or vegans....
Yeah, Geekman came up with the idea of using the internet as a sort of 'Mechanical Turk' for the evaluation stage.
You could have a recipe site and the combinations created by your program get posted there, and people out in the world vote for their favourites.
Do we get to vote as many times as we like and potentially banish any sub-standard combinations to the black halls of oblivion? I guess we'd need the sandwich equivalent of being voted out of the house...
There's sure to be some freak who loves the weirdest combination, though, miss m.
Styley, you need to set up that recipe site!
hotsandwichornotsandwich.com?
Ironically I was planning to make a recipe site, and of course all self-respecting recipe sites need an experimental department. I wonder if I can steal this.... I see a daily sandwich RSS feed.
I think sexual reproduction is the way to go (it's not my answer to everything but it's often worth a look). Here, the higher rated sandwiches have an increased chance of both survival and attracting a highly rated sandwich mate. Winning sandwiches could mate with other winning sandwiches and exchange their fillings. Gosh I don't feel hungry at all now.
Awesome, Matthew. You should definitely steal this. I can't be bothered actually implementing it myself.
And I was definitely meaning sexual reproduction.
If you make a whole heap of money off this, I want a payout :)
If I ever make money with any of my spare time projects, Hell will freeze over (which will be col as I now have skates so I'm frozen-hell-ready).
Talk to me! (You know you want to!)