Games of Pattern, part 3

[globetrvllr2]

Statisticians use models to approximate reality. They look for patterns in the exceptions or *error terms.*

Exceptions' patterns tell us about the models' problems.

Proper models should not create patterns in the exceptions.

[globetrvllr2]

Thanks for the great series.

Statistics look for patterns in exceptions or *error terms* to see if its model is wrong. The error terms should be randomly distributed around zero.

Statistical models are missing variables if the exceptions repeatedly cycle.

They need to weigh the variables if the errors get larger as the variables get larger. (Heteroskedasticity)

[globetrvllr2]

My main point is that you do need to look for patterns in the exceptions to understand how well your model prevents them.

But your series is great. It relates game theory to theories of knowledge and politics. Incredible! Thanks.

[denito9474]

Thanks, that was great. Do you play much go?

[Mariborchan]

I propose a drinking game. Rewind this video and drink a shot of vodka every time you hear except~. If you survive YOU are the exception and therefore there is no more need for an exception representation.

[AgentGhost]

I'll try it. lulz

[randyhelzerman]

Um.....unless I'm mistaken though, that last 20 seconds or so sounded a wee bit....um....Kantian? Perhaps not though, your point was not about the necessary structure of experience, rather the language we use to represent that experience...

[randyhelzerman]

Insanely great. By far the best series of vids on YouTube.

[bubonicnate]

nice demonstration of the political ideas in the last vid to general formations of illusionary knowledge - and I see how this all ties into our discussion the last coupel nights on stickam. And beyond the problems you stated with generalizing these "exceptions" - any one of them could be an overwhelming set of data that completely over shadows the importance of all the other data in importance or for use in understanding.

[bubonicnate]

I would also like to re-itterate soemthing you and I both agreed on - and that is it may very well be these exceptions exist that allows us to define the pattern in which they are included.

[oldiousnei]

In addition, I would add that the one of the most powerful tools at our disposal is our sensitivity to the degrees (thresholds) of generalization. Our ability to calibrate this "instrument" is directly proportional to the effectiveness of the representation. Great vids Ben.

[scoutie111]

Ben..Just finished watching the series. Well done! Would love to discuss with you in Stickam. Note that the degree to which there is 'exception,' i.e., variation from the pattern and no pattern in the variation, within the field of educational research reeks havoc in the field. Also, I'm wondering if the problem of predictability in the GO game is similar to the problem of 'exception.' (Am I seeking a pattern here?? ;-)

[Boucrate]

Excellent!

[Z4lUS]

exceptional vid