| | |

Intuitively Probabilistic Programmer [wannabe]

thinking and drooling

You know what, I get the feeling that I’m somehow “destined” to be a “probabilistic guy”

(it has a spiritual touch)

A few minutes ago I was thinking that “IT” is simply about reducing ambiguity. which is basically increasing specificity. Problem is, the world is inherently uncertain.
And IT usually doesn’t cope well with this that changes a lot (hence the need for BPMs and such). IT is very good for things that are strictly in order, and certain, and known upfront.

The more we get to details, the more the uncertainty (and hence the ‘probability factor’) increases.

This is how the real world seems to be modeled, according to some scientists. On a macro level, the universe is calm and orderly (i.e. planets and such), and we have Newton law. On a very micro level though, beyond atoms, things get very ecstatic and we have quantum theory. Which is simply the technical term for ‘probabilistic uncertainty of matter or whatever’.

Then I was thinking, why am I so attached to this ‘probability’ thingy?

IT world currently don’t really have a “probabilistic” programming model. Ok, we have imperative. And it’d be a full chore to write probabilistic functions using if-then and loops and such. We have functional programming, which is rare-r and a bit more inline with ‘probabilistic’, but still, functional programming still expects a well-defined function to transform input to output.

What I think, is there should be a probabilistic programming model, which should perform well in some areas of the real world problems. I’m calling it… “Intuitive programming”. (“Intentional programming” is a term already snatched, and it’s different anyway)

What tools do we have to do probabilistic programming? I’m not sure. Never familiar with it.

My final bachelor paper was about Bayesian Theorem. See the match? I certainly didn’t “specify” Bayes as my goal when I was in college. It was just something that happens to be one of my only options when I needed to choose a final paper topic. (as a background, I never thought of probabilistic statistical theme either, I was just interested in folksonomy and del.icio.us at that time… I wonder what made me into Bayes) The result was very good for me, but I know behind the scenes that my paper wasn’t all that good. It’s lousy…

It’s probabilistic! “There’s an 60% chance that my paper was ‘good enough’ to deserve an A… and it did!” 😛

Too many choices. Too many probabilities. Has got to be intuitive.

It doesn’t mean it’s 60% good. It just means (assumptively) that even the paper was 90% bad, the probabilistic chance was good enough. Thanks Bayes! 😉


Similar Posts