# denimalpaca comments on A Problem for the Simulation Hypothesis

• Let me be a lit­tle more clear. Let’s as­sume that we’re in a simu­la­tion, and that the par­ent uni­verse host­ing ours is the top level (for what­ever rea­son, this is just to avoid tur­tles all the way down). We know that we can har­ness the en­ergy of the sun, be­cause not only do plants uti­lize that en­ergy to me­tab­o­lize, but we also can har­ness that en­ergy and use it as elec­tric­ity; en­ergy can trans­fer.

Some ma­chine that we’re be­ing simu­lated on must take into ac­count these kinds of in­ter­ac­tions and make them hap­pen in some way. The ma­chine must rep­re­sent the sun in some way, per­haps as 0s and 1s. This en­cod­ing takes en­ergy, and if we were to sim­ply en­code all the en­ergy of the sun, the po­ten­tial en­ergy of the sun must ex­ist some­where in that ma­chine. Even if the sun’s in­for­ma­tion is com­pressed, it would still have to be de­com­pressed when used (or else we have a “lossy” sun, not good if you don’t want your simu­la­tions to figure out they’re in a simu­la­tion) - and com­press­ing/​de­com­press­ing takes en­ergy.

We know that even in a perfect simu­la­tion, the sun must have the same amount of en­ergy as out­side the simu­la­tion, oth­er­wise it is not a perfect simu­la­tion. So if a blue pho­ton has twice as much en­ergy as a red pho­ton, then that fact is what causes twice as much en­ergy to be en­coded in a simu­lated blue pho­ton. This en­ergy en­cod­ing is nec­es­sary if/​when the blue pho­ton in­ter­acts with some­thing.

Said an­other way: If, in our simu­la­tion, we en­code the en­ergy of phys­i­cal things with the small­est num­ber of bits pos­si­ble to de­scribe that thing, and blue pho­tons have twice as much en­ergy as red pho­tons, then it should take X bits to de­scribe the en­ergy of the red pho­ton and 2*X bits to de­scribe the blue pho­ton.

As to ex­tra en­ergy, as a prac­ti­cal (en­g­ineer­ing) mat­ter alone it would take more en­ergy to simu­late a thing even af­ter the en­cod­ing for the thing is done: in our uni­verse, there are no perfect en­ergy trans­fers, some is in­evitably lost as heat, so it would take ex­tra en­ergy to over­come this loss. Se­condly, if the simu­la­tion had any meta-data, that would take ex­tra in­for­ma­tion and hence ex­tra en­ergy.

• I still don’t un­der­stand. (Less tact­fully, I think what you’re say­ing is sim­ply wrong; but I may be miss­ing some­thing.)

Sup­pose we have one simu­lated pho­ton with 1000 units of en­ergy and an­other with 2000 units of en­ergy. Here is the bi­nary rep­re­sen­ta­tion of the num­ber 1000: 1111101000. And here is the bi­nary rep­re­sen­ta­tion of the num­ber 2000: 11111010000. The sec­ond num­ber is longer—by one bit—and there­fore may take a lit­tle more en­ergy to do things with; but it’s only 10% big­ger than the first num­ber.

Now, if we imag­ine that even­tu­ally each of those pho­tons gets turned into lots of lit­tle blobs car­ry­ing one unit of en­ergy each, or in some other way has a bunch of in­ter­ac­tions whose num­ber is pro­por­tional to its en­ergy, then in­deed you end up with an amount of simu­la­tion effort pro­por­tional to the en­ergy. But it’s not clear to me that that must be so. And if most in­ter­ac­tions in­side the simu­la­tion in­volve the ex­change of a quan­tity of en­ergy that’s larger than the amount of en­ergy re­quired to simu­late one in­ter­ac­tion—which seems kinda un­likely, which is one rea­son why I am sym­pa­thetic to your ar­gu­ment over­all, but again I see no ob­vi­ous way to rule it out—then even if simu­la­tion effort is pro­por­tional to en­ergy the rele­vant con­stant of pro­por­tion­al­ity could be smaller than 1.

• I tried to see if any­one else had pre­vi­ously made my ar­gu­ment (but bet­ter); in­stead I found these ar­gu­ments:

http://​​ra­tio­nalwiki.org/​​wiki/​​Si­mu­lated_re­al­ity#Feasibility

I think the fea­si­bil­ity ar­gu­ment de­scribed here bet­ter en­cap­su­lates what I’m try­ing to get at, and I’ll defer to this ar­gu­ment un­til I can bet­ter (more math­e­mat­i­cally) state mine.

“Yet the num­ber of in­ter­ac­tions re­quired to make such a “perfect” simu­la­tion are vast, and in some cases re­quire an in­finite num­ber of func­tions op­er­at­ing on each other to de­scribe. Per­haps the only way to solve this would be to as­sume “simu­la­tion” is an anal­ogy for how the uni­verse (op­er­at­ing un­der the laws of quan­tum me­chan­ics) acts like a quan­tum com­puter—and there­fore it can “calcu­late” it­self. But then, that doesn’t re­ally say the same thing as “we ex­ist in some­one else’s simu­la­tion”.” (from the link).

This con­clu­sion about the uni­verse “simu­lat­ing it­self” is re­ally what I’m try­ing to get at. That it would take the same amount of en­ergy to simu­late the uni­verse as there is en­ergy in the uni­verse, so that a “self-simu­lat­ing uni­verse” is the most likely con­clu­sion, which is of course just a base uni­verse.