# Why is multi worlds not a good explanation for abiogenesis

I’m not a ex­pert in the multi world the­ory. So this ques­tion could very well be ex­tremely stupid. How­ever, given the as­sump­tion that there are nearly in­finite amount of wor­lds that are slightly differ­ent than each other, nearly ev­ery pos­si­ble event would hap­pen. This in­cludes the for­ma­tion of life. Now what are the odds that we would be wit­ness­ing that world, as far as I can tell 100 per­cent.

Now I’m not clear ex­actly how of­ten quan­tum events lead to a slightly differ­ent world but even at the rate of 1 quan­tum event a year in the en­tire uni­verse. should lead to a near in­finite ex­plo­sion of com­pletely differ­ent uni­verses.

Now I’m not claiming that this is the ex­pla­na­tion for abio­ge­n­e­sis or that abio­ge­n­e­sis is proof of multi wor­lds be­cause that would be multi wor­lds of the gap fal­lacy how­ever I’m not clear why I have never even seen this ex­pla­na­tion even once for abio­ge­n­e­sis.

I also sus­pect that math­e­mat­i­cally many wor­lds would usu­ally be the wrong ex­pla­na­tion for nearly ev­ery­thing be­cause it runs into se­ri­ous odds prob­lems and in 99.99999 per­cent of cases there is a bet­ter ex­pla­na­tion. how­ever it should at least be con­sid­ered

COULD SOMEONE EXPLAIN TO ME EXACTLY WHERE I WENT WRONG

• In many wor­lds, ev­ery­thing hap­pens, but not ev­ery­thing hap­pens with equal “prob­a­bil­ity”. Less mirac­u­lous paths to­wards life are more likely than more mirac­u­lous paths to­wards life. Thus, even if the life sees it­self with prob­a­bil­ity 100%, it most likely sees it­self evolved the least mirac­u­lous way.

So, at the end, we are in the same situ­a­tion as we were be­fore con­sid­er­ing many wor­lds: look­ing for the most likely way life could have evolved, be­cause that is most likely our his­tory.

(In other words, many wor­lds do in­tro­duce mir­a­cles, but they still fa­vor the solu­tions that didn’t use them.)

• that’s what i thought but I was won­der­ing why this is not used as a counter to the­is­tic proof from abiogenesis

• Many wor­lds is a model that cur­rently has no testable pre­dic­tions, no micro/​macro con­nec­tion, con­tra­dicts gen­eral rel­a­tivity (which as­sumes there is a sin­gle space­time), and in gen­eral proves too much: nearly any­thing can be a con­se­quence of in­finitely many wor­lds. Ad­di­tion­ally, the ob­serv­able uni­verse has only 10^122 qubits, which limits the num­ber of pos­si­ble states, in­clud­ing pos­si­ble wor­lds.

So, your best bet is to avoid in­vok­ing many wor­lds for ex­plain­ing any­thing. You can cer­tainly use pos­si­ble wor­lds, as log­i­cal coun­ter­fac­tu­als based on the lack of knowl­edge of the “real” world, to con­sider which de­ci­sion would be best, for ex­am­ple. But not many wor­lds. Those cur­rently have zero pre­dic­tive power and 100% ex­plana­tory power, which is equiv­a­lent to “God did it”.

• I know al­most no physics so this might be a stupid ques­tion, but aren’t “makes no testable pre­dic­tions” and “con­tra­dicts gen­eral rel­a­tivity” con­tra­dic­tory? Wouldn’t con­tra­dict­ing an­other the­ory im­ply some kind of a pre­dic­tion?

• That’s a good point! I was definitely un­clear, and even sloppy in my claims.

My first state­ment, “makes no testable pre­dic­tions” refers to “pure” quan­tum physics, speci­fi­cally quan­tum me­chan­ics and quan­tum field the­ory on a fixed space­time back­ground, where what the mat­ter is do­ing does not af­fect, in the first ap­prox­i­ma­tion, what hap­pens to the space­time it­self (which is the sub­ject of gen­eral rel­a­tivity). We know it is not a good as­sump­tion in gen­eral, be­cause it leads to con­tra­dic­tions like the var­i­ous black hole evap­o­ra­tion para­doxes. But it works within cer­tain limits. Within those limits, many wor­lds add ab­solutely noth­ing new and pre­dictable.

Sadly, the ex­trap­o­la­tion of QM to the realm where grav­ity is still weak but already mat­ters, is still an un­charted ter­ri­tory, over 90 years later. The gen­er­ally ac­cepted claim (but only a claim) is that the uni­tary evolu­tion part of the QM scales up into the macro­scopic world in some way, and the mea­sure­ment pos­tu­late emerges from this up­scal­ing even­tu­ally. Many wor­lds make a more spe­cific claim, that we live in many wor­lds, de­co­her­ing (split­ting) all the time, and that there is noth­ing new hap­pen­ing be­yond the ba­sic de­co­her­ence. How­ever, there is still the grav­i­ta­tional foot­print of those wor­lds, un­less you figure out how they split the space­time it­self, as well. In that sense, many wor­lds make a claim in the do­main where QM has never been ob­served that is in­com­pat­i­ble with the the­ory that rules that do­main. it is still just an on­tolog­i­cal claim though, with­out any pre­dic­tive power in it.

Not sure if this makes sense.

• while i can’t ac­tu­ally un­der­stand what your say­ing be­cause I don’t un­der­stand physics well enough. As far as I know its not con­tro­ver­sial to use the multi world model in the less wrong fo­rums and that most peo­ple I re­spect use it fully. Is what your writ­ing rele­vant to my ques­tion or to the en­tire less­wrong that be­lieve that the many wor­lds ex­pla­na­tion is correct

• its not con­tro­ver­sial to use the multi world model in the less wrong fo­rums and that most peo­ple I re­spect use it fully

the key word is “in the less­wrong fo­rums”. This is be­cause Eliezer Yud­kowsky, the founder and the main con­trib­u­tor for a long time, pro­moted both MWI and Bayesi­anism as cor­ner­stones of ra­tio­nal­ity. Nei­ther is nec­es­sary for ei­ther epistemic or in­stru­men­tal ra­tio­nal­ity, but they are use­ful rea­son­ing de­vices. No one re­ally “uses” those di­rectly to make de­ci­sions in life, even though most peo­ple pre­tend to. In ac­tu­al­ity, they use those to jus­tify the de­ci­sions already made, con­sciously or sub­con­sciously. The rea­son is that the Bayes the­o­rem re­lies on eval­u­a­tion of prob­a­bil­ities, some­thing hu­mans are not very good at. At least not un­til you spend as much time as Eliezer, Scott and some oth­ers on self-cal­ibra­tion. And MWI is gen­er­ally used as a fancy name for “imag­ine pos­si­ble out­comes and as­sign prob­a­bil­ities to them”, which has noth­ing to do with physics what­so­ever, when it is not mi­sused for dis­cussing quan­tum suicide/​im­mor­tal­ity, or, well, to jus­tify an­throp­ics.

• nearly any­thing can be a con­se­quence of in­finitely many worlds

This feels like com­plain­ing that if you flip a coin mil­lion times, all out­comes are pos­si­ble.

• Why do you call it com­plain­ing?

• I stipu­late that nearly any­thing can be a con­se­quence of MWI, but not with equal prob­a­bil­ity. If I see a thou­sand quan­tum coin­flips in a row all come up heads, I don’t think “well, un­der MWI any­thing is pos­si­ble, so I haven’t learned any­thing”. So I’m not sure in what sense you think it proves too much.

(I think this is roughly what Villiam was get­ting at, though I can’t speak for him.)

• I stipu­late that nearly any­thing can be a con­se­quence of MWI, but not with equal prob­a­bil­ity.

Note that MWI pos­tu­lates uni­tary evolu­tion of the wave func­tion, and in uni­tary evolu­tion there are no prob­a­bil­ities, ev­ery­thing is com­pletely de­ter­minis­tic, no ex­cep­tions. None. Let it sink in:

NO PROBABILITIES. PURE DETERMINISM OF THE WAVE FUNCTION EVOLUTION

There have been nu­mer­ous at­tempts to sad­dle this uni­tary evolu­tion with some­thing ex­tra that would give us the em­piri­cally ob­served prob­a­bil­ities. Everett sug­gested some in his PhD, many oth­ers did, with marginal suc­cess. The only state­ment nearly ev­ery­one is on board with is that, if we were to look for a way to as­sign prob­a­bil­ities, the Born rule is the only sen­si­ble one. In that sense, the Born rule is not an ar­bi­trary one, but a unique way to map wave func­tion to prob­a­bil­ity. The need to get prob­a­bil­ities from the uni­tary evolu­tion of the wave func­tion is not built into the MWI, but is grafted on it by the need to con­nect this the­ory with ob­ser­va­tions, ex­actly like the Born rule in the Copen­hagen in­ter­pre­ta­tion was.

That said, we might be on the cusp of some­thing su­per mega ex­tra in­ter­est­ing ob­served in the next few years, much more so than the re­cent black hole dough­nut seen by the EHT: Mea­sur­ing grav­i­ta­tional field from Schrod­inger cat-like ob­jects. There are no definite pre­dic­tions on what we will see in this case, be­cause QM and gen­eral rel­a­tivity cur­rently do not mix, and this is what makes it so ex­cit­ing. I have men­tioned it in a blog post dis­cussing how MWI emerges from uni­tary evolu­tion:

https://​​ed­ge­of­grav­ity.word­press.com/​​2019/​​01/​​19/​​en­tan­gle­ment-many-wor­lds-and-gen­eral-rel­a­tivity/​​

There is some dis­cus­sion of this is­sue on­line, and I have mused about it on my blog some time ago:

https://​​ed­ge­of­grav­ity.word­press.com/​​2019/​​02/​​25/​​schrod­ingers-cat­trac­tion/​​

• I’m still not en­tirely clear what you mean by “MWI proves too much”.

If I try to trans­late this into sim­pler terms, I get some­thing like: MWI only matches our ob­ser­va­tions if we ap­ply the Born rule, but it doesn’t mo­ti­vate the Born rule. So there are many sets of ob­ser­va­tions that would be com­pat­i­ble with MWI, which means P(data | MWI) is low and that in turn means we can’t up­date very much on P(MWI | data).

Is that ap­prox­i­mately what you’re get­ting at?

(That would be a non­stan­dard us­age of the phrase, es­pe­cially given that you linked to the wikipe­dia ar­ti­cle when us­ing it. But it kind of fits the name, and I can’t think of a way for the stan­dard us­age to fit.)

• It seems like we are talk­ing about some­thing similar. If you in­ter­pret MWI as “any­thing can hap­pen with some prob­a­bil­ity, and, given that we are here ob­serv­ing it, the pos­te­rior prob­a­bil­ity is ob­vi­ously high enough”, then you can use it to ex­plain any­thing. I agree that my us­age was not quite stan­dard, but it fits some­what, be­cause you can use MWI to jus­tify any con­clu­sion, in­clud­ing an ab­surd one.

• The nub of the ar­gu­ment is that ev­ery time we look in our sock drawer, we see all our socks to be black.

Many wor­lds says that our socks are always black.

The Copen­hagen in­ter­pre­ta­tion says that us ob­serv­ing the socks causes them to be black. The rest of the time the socks are pink with green spots.

Both the­o­ries make iden­ti­cal pre­dic­tions. Many wor­lds is much sim­pler to fully spec­ify with equa­tions, and has el­e­gant math­e­mat­i­cal prop­er­ties. The Copen­hagen in­ter­pre­ta­tion has spe­cial case rules that only kick in when ob­serv­ing some­thing. Ac­cord­ing to this the­ory, there is a fun­da­men­tal phys­i­cal differ­ence be­tween a com­plex col­lec­tion of atoms, and an “ob­server” and some­where in the de­vel­op­ment of life, crea­tures flipped from one to the other.

The Copen­hagen in­ter­pre­ta­tion doesn’t make it clear if a cat is a very com­plex ar­range­ment of molecules, that could in the­ory be un­der­stood as a quan­tum pro­cess that doesn’t in­volve the col­lapse of wave func­tions, or if cats are ob­servers and so col­lapses wave func­tions.

• Many wor­lds says that our socks are always black.

Nope. What dis­t­in­guishes wor­lds, if their con­tents are the same?

The Copen­hagen in­ter­pre­ta­tion says that us ob­serv­ing the socks causes them to be black.

Nope. That would be con­scious­ness-causes-col­lapse. Which is a differ­ent the­ory.

• Ac­cord­ing to this the­ory, there is a fun­da­men­tal phys­i­cal differ­ence be­tween a com­plex col­lec­tion of atoms, and an “ob­server” and some­where in the de­vel­op­ment of life, crea­tures flipped from one to the other.

You seem to re­fer to some straw­man ver­sion of the Copen­hagen in­ter­pre­ta­tion that no physi­cist sub­scribes to. Be­ing brain­washed by Eliezer’s writ­ings can do that. He is very elo­quent and per­sua­sive. Con­sider read­ing other sources. Scott Aaron­son’s blog is a good start. Wikipe­dia has a bunch of use­ful links, too.

• Con­sider a the­ory to be a col­lec­tion of for­mal math­e­mat­i­cal state­ments about how ideal­ized ob­jects be­have. For ex­am­ple, Con­ways game of life is a the­ory in the sense of a com­pletely self con­tained set of rules.

If you have mul­ti­ple the­o­ries that pro­duce similar re­sults, its helpful to have a bridg­ing law. If your the­o­ries were New­to­nian me­chan­ics, and gen­eral rel­a­tivity, a bridg­ing law would say which num­bers in rel­a­tivity matched up with which num­bers in New­to­nian me­chan­ics. This al­lows you to trans­late a rel­a­tivis­tic prob­lem into a New­to­nian one, solve that, and trans­late the an­swer back into the rel­a­tivis­tic frame­work. This pro­duces some er­rors, but of­ten makes the maths eas­ier.

Quan­tum many wor­lds is a sim­ple the­ory. It could be simu­lated on a hy­per­com­puter with less than a page of code. There is also a the­ory where you take the code for quan­tum many wor­lds, and add “ob­servers” and “wave­func­tion col­lapse” with ex­tra func­tions within your code. This can be done, but it is many pages of ar­bi­trary hacks. Call this the­ory B. If you think this is a straw­man of many wor­lds, de­scribe how you could get a hy­per­com­puter out­side the uni­verse to simu­late many wor­lds with a short com­puter pro­gram.

The bridg­ing be­tween Quan­tum many wor­lds and hu­man clas­si­cal in­tu­itions is quite difficult and sub­tle. Faced with a simu­la­tion of quan­tum many wor­lds, it would take a lot of un­der­stand­ing of quan­tum physics to make ev­ery­day changes, like cre­at­ing or mov­ing macro­scopic ob­jects.

The­ory B how­ever is sub­stan­tially eas­ier to bridge to our clas­si­cal in­tu­itions. The­ory B looks like a chunk of quan­tum many wor­lds, plus a chunk of clas­si­cal in­tu­ition, plus a bridg­ing rule be­tween the two.

The any de­scrip­tion of the Copen­hagen in­ter­pre­ta­tion of quan­tum me­chan­ics seems to in­volve refer­ences to the clas­si­cal re­sults of a mea­sure­ment, or a clas­si­cal ob­server. Most ver­sions would al­low a su­per­po­si­tion of an atom be­ing in two differ­ent places, but not a su­per­po­si­tion of two differ­ent pres­i­dents win­ning an elec­tion.

If you don’t be­lieve atoms can be in su­per­po­si­tion, you are ig­nor­ing lots of ex­per­i­ments, if you do be­lieve that you can get a su­per­po­si­tion of two differ­ent peo­ple be­ing pres­i­dent, that you your­self could be in a su­per­po­si­tion of do­ing two differ­ent things right now, then you be­lieve many wor­lds by an­other name. Other­wise, you need to draw some sort of ar­bi­trary cut­off. Its al­most like you are bridg­ing be­tween a the­ory that al­lows su­per­po­si­tions, and an in­tu­ition that doesn’t.

• “Now I’m not clear ex­actly how of­ten quan­tum events lead to a slightly differ­ent world”

The an­swer is Very Very of­ten. If you have a piece of glass and shine a pho­ton at it, such that it has an equal chance of bounc­ing and go­ing through, the two pos­si­bil­ities be­come sep­a­rate wor­lds. Shine a mil­lion pho­tons at it and you split into wor­lds, one for each com­bi­na­tion of pho­tons go­ing through and bounc­ing. Note that in most of the wor­lds, the pat­tern of bounces looks ran­dom, so this is a good source of ran­dom num­bers. Pho­tons bounc­ing of glass are just an easy ex­am­ple, al­most any phys­i­cal pro­cess splits the uni­verse very fast.

• The prob­lem is that, at some level, pure Boltz­mann brains be­come more prob­a­ble than quan­tum abio­ge­n­e­sis. That is, if abio­ge­n­e­sis is very im­prob­a­ble, than ap­pear­ing of the whole con­scious mind from the ocean full of amino acids be­come more prob­a­ble, and such mind un­likely will ob­serve the whole evolu­tion­ary his­tory.

How­ever, as we ob­serve abio­ge­n­e­sis, it means that this event is more prob­a­ble than Boltz­mann brain for­ma­tion, and such prob­a­bil­ities could be calcu­lated. Some cos­mol­o­gists did it.

• It seems weird that given our laws of na­ture it would be more prob­a­ble that boltz­mann brains would form be­cause boltz­mann brains are more com­pli­cated than rna as far as I could tell

I thought the whole prob­lem with boltz­mann brains was in the fine tun­ing ar­gu­ment and the mul­ti­verse

• Could you also link me to a good ex­pla­na­tion of the odds of boltz­mann brains

Thank you

• thanks for the re­ply