# The Experimental Apparatus

Pre­vi­ous: The prob­lem of other minds (Greater­wrong)

Science re­ally likes to iso­late vari­ables. From first grade, drop­ping a ball to mea­sure grav­ity ex­per­i­ment. We talk about the wind re­sis­tance on the ball, and how it might af­fect a mea­sure­ment. So we re­peat the ex­per­i­ment for re­li­a­bil­ity. As­sum­ing if we have slightly differ­ent wind, and we av­er­age out the noise we can prob­a­bly work out the an­swer on av­er­age by get­ting re­peata­bil­ity.

The “as­tute” of us will think about who drops the ball, how the ball drops and how we can drop the ball in an im­par­tial way so as to not af­fect the ex­per­i­ment our­selves when we run it.

This pro­cess pre­sumes there is some kind of setup that can iso­late us from the equa­tion when we drop the ball. And we can prob­a­bly drop the ball in such a way that it isn’t too im­por­tant for the re­sults of the ex­per­i­ment.

The “way too as­tute for their own good” will start to think about lo­cal den­sity effects of hav­ing hu­mans around and mov­ing and how hu­mans will still in­fluence the ex­per­i­ment with our teeny grav­i­ta­tional pulls even if it’s ir­rele­vant for the most part.

It’s eas­ier for in­ert ex­per­i­ments to iso­late a vari­able—It’s just a ball and just grav­ity. We take this prin­ci­ple from the ball and grav­ity and we ap­ply it to Psy­chol­ogy. If we can just stand back and watch peo­ple. We can do psy­chol­ogy re­search with­out dis­turb­ing them.

Well ac­tu­ally we know that we have trou­ble with re­peata­bil­ity of ex­per­i­ments. It’s re­mark­ably hard to do an ex­per­i­ment the same ev­ery time. And we strug­gle with it and we wres­tle with it. And we iso­late enough vari­ables that even­tu­ally we be­lieve we can pre­dict “how of­ten peo­ple cheat given cer­tain con­di­tions” (Dan Ariely stud­ies hu­man bi­ases and how cer­tain con­di­tions in­fluence peo­ple to cheat, he’s great fun to read but that’s not so im­por­tant right now).

Some­how there’s this trou­ble with iso­lat­ing vari­ables. All of sci­ence strug­gles with it. All of model mak­ing strug­gles with it. Ein­stein strug­gled with it. Any time we try to learn we strug­gle with it. We won­der—Was that the grav­ity on the ball that I’m try­ing to iso­late and mea­sure, or was that just the wind on the ball...

• “Pre­vi­ous” link is to bear­lamp post, not to LW ver­sion of it (which seems like it should be the tar­get).

I have to say it doesn’t seem like there’s a lot of con­tent here. Yes, do­ing sci­ence on peo­ple is hard be­cause peo­ple are im­por­tantly un­like perfectly smooth hard billiard balls, and ac­tu­ally even do­ing sci­ence on perfectly smooth hard billiard balls can be pretty tricky. This … isn’t ex­actly news. I guess this is part of a se­quence of posts that will in due time say some more sur­pris­ing things; it might be worth ei­ther mov­ing faster or (if that’s a bad idea for the read­er­ship of your blog) mak­ing some adap­ta­tions (e.g., batch­ing up a few bear­lamp posts and con­dens­ing them for LW).

Nit­pick: The ti­tle cur­rently has the non-word “Apara­tus” in it; it should be “Ap­para­tus”.

• Nit­pick fixed. Pre­vi­ous links be­ing finicky but also fixed.

Con­tent: foun­da­tion of fur­ther de­vel­op­ments. Con­cern that if I move faster I will lose peo­ple.