Denis Drescher(Denis Drescher)

• Denis Drescher 19 May 2020 12:11 UTC

  11 points

  on: [Stats4EA] Ex­pec­ta­tions are not Outcomes

  I’ve found Chris­tian Tarsney’s “Ex­ceed­ing Ex­pec­ta­tions” in­sight­ful when it comes to rec­og­niz­ing and maybe cop­ing with the limits of ex­pected value.

  The prin­ci­ple that ra­tio­nal agents should max­i­mize ex­pected util­ity or choice­wor­thi­ness is in­tu­itively plau­si­ble in many or­di­nary cases of de­ci­sion-mak­ing un­der un­cer­tainty. But it is less plau­si­ble in cases of ex­treme, low-prob­a­bil­ity risk (like Pas­cal’s Mug­ging), and in­tol­er­ably para­dox­i­cal in cases like the St. Peters­burg and Pasadena games. In this pa­per I show that, un­der cer­tain con­di­tions, stochas­tic dom­i­nance rea­son­ing can cap­ture most of the plau­si­ble im­pli­ca­tions of ex­pec­ta­tional rea­son­ing while avoid­ing most of its pit­falls. Speci­fi­cally, given suffi­cient back­ground un­cer­tainty about the choice­wor­thi­ness of one’s op­tions, many ex­pec­ta­tion-max­i­miz­ing gam­bles that do not stochas­ti­cally dom­i­nate their al­ter­na­tives “in a vac­uum” be­come stochas­ti­cally dom­i­nant in virtue of that back­ground un­cer­tainty. But, even un­der these con­di­tions, stochas­tic dom­i­nance will not re­quire agents to ac­cept op­tions whose ex­pec­ta­tional su­pe­ri­or­ity de­pends on suffi­ciently small prob­a­bil­ities of ex­treme pay­offs. The sort of back­ground un­cer­tainty on which these re­sults de­pend looks un­avoid­able for any agent who mea­sures the choice­wor­thi­ness of her op­tions in part by the to­tal amount of value in the re­sult­ing world. At least for such agents, then, stochas­tic dom­i­nance offers a plau­si­ble gen­eral prin­ci­ple of choice un­der un­cer­tainty that can ex­plain more of the ap­par­ent ra­tio­nal con­straints on such choices than has pre­vi­ously been rec­og­nized.

  See also the post/​se­quence by Daniel Koko­ta­jlo, “Tiny Prob­a­bil­ities of Vast Utilities”. I’m link­ing to the post that was most valuable to me, but by de­fault it might make sense to start with the first one in the se­quence. ^^

• Denis Drescher 16 May 2020 10:50 UTC

  3 points

  in reply to: EdoArad’s comment on: Model­ers and Indexers

  Yeah, to­tally agree! The Birds and Frogs dis­tinc­tion sounds very similar! I’ve pock­eted the origi­nal ar­ti­cle for later read­ing.

  And I also feel that the Adap­tors–In­no­va­tors one is “may be slightly cor­re­lated but is a differ­ent thing.” :-)

• Denis Drescher 16 May 2020 10:36 UTC

  3 points

  in reply to: Lukas_Gloor’s comment on: Model­ers and Indexers

  Yes! I’ve been think­ing about you a lot while I was writ­ing that post be­cause you your­self strike me as a po­ten­tial coun­terex­am­ple to the use­ful­ness of the dis­tinc­tion. I’ve seen you do ex­actly what you de­scribe and gen­er­ally dis­play com­fort in situ­a­tions that in­dex­ers would nor­mally be com­fortable in, while at the same time you ev­i­dently have quite similar pri­ori­ties to me. So ei­ther you break the model or you’re just re­ally good at both! :-)

• Denis Drescher 16 May 2020 10:31 UTC

  3 points

  in reply to: RomeoStevens’s comment on: Model­ers and Indexers

  Thank you!

  Yeah, that feels fit­ting to me too. I found these two posts on the term:

  https://​​www.less­wrong.com/​​posts/​​xqAnKW46FqzPLnGmH/​​causal-re­al­ity-vs-so­cial-re­al­ity https://​​www.less­wrong.com/​​posts/​​j2mcSRx­hjRy­hyLJEs/​​what-is-so­cial-reality

  A lot of so­cial things ap­pear ar­bi­trary when deep down they must be de­ter­minis­tic. But bridg­ing that gap is per­haps both com­pu­ta­tion­ally in­fea­si­ble and doesn’t lend it­self to par­tic­u­larly pow­er­ful ab­strac­tions (ex­cept for in­ten­tion­al­ity). At the same time, though, the sub­ject is more in­ex­tri­ca­bly in­te­grated with the en­vi­ron­ment, so that it makes more sense to model the en­vi­ron­ment as fal­ling into in­ten­tional units (agents) who are re­ac­tive. And then maybe cer­tain bar­gain­ing pro­ce­dures emerged (be­cause they were adap­tive) that are now in­te­grated into our psy­che as cus­toms and moral in­tu­itions.

  For these bar­gain­ing pro­ce­dures, I imag­ine, it’ll be im­por­tant to ab­stract use­fully from spe­cific situ­a­tions to more gen­eral games. Then you can clas­sify a new situ­a­tion as one that ei­ther re­quires go­ing through the bar­gain­ing pro­ce­dure again or is a near-repli­ca­tion of a situ­a­tion whose bar­gain­ing out­come you already have stored. That would re­quire ex­actly the in­dexer types of abil­ities – ab­stract­ing from situ­a­tions to archetypes and stor­ing the archetypes.

  (E.g., if you sell books, there’s a stored bar­gain­ing solu­tion for that where you de­clare a price, and if it’s right, hand over the book and get the money for it, and oth­er­wise keep the book and don’t get the money. But if you were the first to cre­ate a search en­g­ine that in­dexes the full-text of books, there were no stored bar­gain­ing solu­tions for that and you had to go through the bar­gain­ing pro­ce­dures.)

  It also seems to me that there are peo­ple who, when in doubt, tend more to­ward run­ning through the bar­gain­ing pro­ce­dure, while oth­ers in­stead tend more to­ward ob­serv­ing and learn­ing es­tab­lished bar­gain­ing solu­tions very well and maybe widen­ing their refer­ences classes for games. I as­so­ci­ate the first a bit with en­trepreneurs, low agree­able­ness, Aus­tralia, and the Pavlov strat­egy, and the sec­ond with me, agree­able friends of mine, Ger­many/​Switzer­land, and tit for tat.

Model­ers and Indexers

• Denis Drescher 1 May 2020 13:35 UTC

  6 points

  on: Bored at home? Con­tribute to the EA Wiki!

  I love the idea of wikis for EA knowl­edge, but is there an at­tempt un­der­way yet to con­soli­date all the ex­ist­ing wikis, be­yond the Wikia one? Maybe you can co­or­di­nate some data im­port with the other peo­ple who are run­ning EA wikis.

  When the Pri­or­ity Wiki was launched, (I but much more so) John Maxwell com­piled some of the ex­ist­ing wikis here.

  I think for one of these wikis to take off, it’ll prob­a­bly need to be­come the clear Schel­ling point for wiki ac­tivity – maybe an in­te­gra­tion with the con­cepts plat­form or the fo­rum and a con­soli­da­tion of all the other wikis as a ba­sis.

  I imag­ine there’d also need to be a way for ac­tive wiki au­thors to gain rep­u­ta­tion points, e.g., in this fo­rum, so wiki edit­ing can have added benefits for CV build­ing. Less Wrong also has fo­rum and wiki, and the fo­rum is a very similar soft­ware, so maybe they already have plans for such a sys­tem.

• Denis Drescher 1 May 2020 10:26 UTC

  3 points

  in reply to: Ahrenbach’s comment on: Does Utili­tar­ian Longter­mism Im­ply Directed Pansper­mia?

  Oh yeah, I was also talk­ing about it only from util­i­tar­ian per­spec­tives. (Ex­cept for one aside, “Others again re­fuse it on de­on­tolog­i­cal or lex­i­cal grounds that I also em­pathize with.”) Just util­i­tar­i­anism doesn’t make a pre­scrip­tion as to the ex­change rate of in­ten­sity/​en­ergy ex­pen­di­ture/​… of in­di­vi­d­ual pos­i­tive ex­pe­riences to in­di­vi­d­ual nega­tive ex­pe­riences.

  It seems that rea­son­able peo­ple think the out­come of B might ac­tu­ally be worse than A, based on your re­sponse.

  Yes, I hope they do. :-)

  Sorry for re­spond­ing so briefly! I’m fal­ling be­hind on some read­ing.

• Denis Drescher 25 Apr 2020 20:21 UTC

  3 points

  in reply to: Ahrenbach’s comment on: Does Utili­tar­ian Longter­mism Im­ply Directed Pansper­mia?

  I think I’m not well placed to an­swer that at this point and would rather defer that to some­one who has thought about this more than I have from the van­tage points of many eth­i­cal the­o­ries rather than just from my (or their) own. (I try, but this is­sue has never been a pri­or­ity for me.) Then again this is a good ex­er­cise for me in moral per­spec­tive-tak­ing or what it’s called. ^^

  It seems C > B > A, with the differ­ence be­tween A and B greater than the differ­ence be­tween B and C.

  In the pre­vi­ous re­ply I tried to give broadly ap­pli­ca­ble rea­sons to be care­ful about it, but those were mostly just from Precipice. The rea­son is that if I ask my­self, e.g., how long I would be will­ing to en­dure ex­treme tor­ture to gain ten years of ul­ti­mate bliss (ap­par­ently a pop­u­lar thought ex­per­i­ment), I might be ready to in­vest a few sec­onds if any, for a trade­off ra­tio of 1e7 or 1e8 to 1. So from my van­tage point, the r-strate­gist style “pro­cre­ation” is very dis­valuable. It seems like it may well be dis­valuable in ex­pec­ta­tion, but ei­ther way, it seems like an enor­mous cost to bear for a highly un­cer­tain pay­off. I’m much more com­fortable with care­ful, K-strate­gist “pro­cre­ation” on a species level. (Mag­nus Vind­ing has a great book com­ing out soon that cov­ers this prob­lem in de­tail.)

  But as­sum­ing the ag­nos­tic po­si­tion again, for prac­tice, I sup­pose A and C are clear cut: C is over­whelm­ingly good (as­sum­ing the Long Reflec­tion works out well and we suc­cess­fully max­i­mize what we re­ally ter­mi­nally care about, but I sup­pose that’s your as­sump­tion) and A is sort of clear be­cause we know roughly (though not very viscer­ally) how much dis­value our an­ces­tors have paid for­ward over the past mil­lions of years so that we can hope­fully even­tu­ally cre­ate a utopia.

  But B is wide open. It may go much more nega­tive than A even con­sid­er­ing all our past gen­er­a­tions – suffer­ing risks, dystopian-to­tal­i­tar­ian lock-ins, per­ma­nent pre­his­toric lock-ins, etc. The less cer­tain it is, the more of this dis­value we’d have to pay for­ward to get one utopia out of it. And it may also go pos­i­tive of course, al­most like C, just with lower prob­a­bil­ity and a de­lay.

  Peo­ple have prob­a­bly thought about how to spread self-repli­cat­ing probes to other planets so that they pro­duce ev­ery­thing a species will need at the des­ti­na­tion to re­build a flour­ish­ing civ­i­liza­tion. Maybe there’ll be some DNA but also com­put­ers with all sorts of knowl­edge, and child-rear­ing robots, etc. ^^ But a civ­i­liza­tion needs so many in­ter­lock­ing parts to func­tion well – all sorts of gov­ern­ment-like in­sti­tu­tions, trust, trade, re­sources, … – that it seems to me like the vast ma­jor­ity of these civ­i­liza­tions ei­ther won’t get off the ground in the first place and re­main locked in a prob­a­bly dis­valuable Stone Age type of state, or will per­ma­nently fall short of the utopia we’re hop­ing for even­tu­ally.

  I sup­pose a way for­ward may to con­sider the great­est un­cer­tain­ties about the pro­ject – prob­a­bil­ities and mag­ni­tudes at the places where things can go most badly net nega­tive or most awe­somely net pos­i­tive.

  Maybe one could look into Great Filters (they may ex­ist less nec­es­sar­ily than I had pre­vi­ously thought), be­cause if we are now past the (or a) Great Filter, and the Great Filter is some­thing about civ­i­liza­tion rather than some­thing about evolu­tion, we should prob­a­bly as­sign a very low prob­a­bil­ity to a civ­i­liza­tion like ours emerg­ing un­der very differ­ent con­di­tions through the prob­a­bly very nar­row pansper­mia bot­tle­neck. I sup­pose this could be tested on some re­mote is­lands? (Ethics com­mit­tees may ob­ject to that, but then these ob­jec­tions also and even more ap­ply to untested pansper­mia, so they should be taken very se­ri­ously. Then again they may not have read Bostrom or Ord. Or Pearce, Gloor, To­masik, or Vind­ing for that mat­ter.)

  Oh, here’s an idea: The Drake Equa­tion has the pa­ram­e­ter f_i for the prob­a­bil­ity that ex­ist­ing life de­vel­ops (prob­a­bly roughly hu­man-level?) in­tel­li­gence, f_c that in­tel­li­gent life be­comes de­tectable, and L for the longevity of the civ­i­liza­tion. The prob­a­bil­ity that in­tel­li­gent life cre­ates a civ­i­liza­tion with similar val­ues and po­ten­tial is prob­a­bly a bit less than f_c (these civ­i­liza­tions could have any moral val­ues) but more than the product of the two fs. The pa­per above has a table that says “f_i: log-uniform from 0.001 to 1” and “f_c: log-uniform from 0.01 to 1.” So I sup­pose we have some 2–5 or­ders of mag­ni­tude un­cer­tainty from this source.

  The longevity of a civ­i­liza­tion is “L: log-uniform from 100 to 10,000,000,000” in the pa­per. An ad­vanced civ­i­liza­tion that ex­ists for 10–100k years may be likely to have passed the Precipice… Not sure at all about this be­cause of the risk of lock-ins. And I’d have to put this dis­tri­bu­tion into Guessti­mate to get a range of prob­a­bil­ities out of this. But it seems like a ma­jor source of un­cer­tainty too.

  The eth­i­cal trade­off ques­tion above feels al­most okay to me with a 1e8 to 1 trade­off but oth­ers are okay with a 1e3 or 1e4 to 1 trade­off. Others again re­fuse it on de­on­tolog­i­cal or lex­i­cal grounds that I also em­pathize with. It feels like there are eas­ily five or­ders of mag­ni­tude un­cer­tainty here, so maybe this is the big­ger ques­tion. (I’m think­ing more in terms of an op­ti­mal com­pro­mise util­ity func­tion than in moral re­al­ist terms, but I sup­pose that doesn’t change much in this case.)

  In the best case within B, there’s also the ques­tion whether it’ll be a de­lay com­pared to C of thou­sands or of tens of thou­sands of years, and how much that would shrink the cos­mic en­dow­ment.

  I don’t trust my­self to be prop­erly morally im­par­tial about this af­ter such a cur­sory in­ves­ti­ga­tion, but that said, I would sup­pose that most moral sys­tems would put a great bur­den of proof on the in­ter­ven­tion be­cause it can be so ex­tremely good and so ex­tremely bad. But tack­ling these three to four sources of un­cer­tainty and maybe oth­ers can per­haps shed more light on how de­sir­able it re­ally is.

  I em­pathize with the no­tion that some things can’t wait un­til the Long Reflec­tion, at least as part in a greater port­fo­lio, be­cause it seems to me that suffer­ing risks (s-risks) are a great risk (in ex­pec­ta­tion) even or es­pe­cially now in the span un­til the Long Reflec­tion. They can per­haps be ad­dressed through differ­ent and more tractable av­enues than other longterm risks and by re­searchers with differ­ent com­par­a­tive ad­van­tages.

  A ne­glected case above is where weapon X de­stroys life on earth, earth en­gages in di­rected pansper­mia, but there was already life in the uni­verse un­be­knownst to earth. I think we agree that B is su­pe­rior to this case, and there­fore the differ­ence be­tween B and A is greater. The ques­tion is does the differ­ence be­tween this case and C sur­pass that be­tween A and B. Call it D. Is D so much worse than C that a preferred loss is from B to A? I don’t think so.

  Hmm, I don’t quite fol­low… Does the above change the rel­a­tive or­der of prefer­ence for you, and if so, to which or­der?

  So I guess the im­plied po­si­tion would be that we should pre­pare a biotic hedge in case things get es­pe­cially dire, and in­vest more in SETI type searches. If we know that life ex­ists el­se­where in the uni­verse, we do not need to de­ploy the biotic hedge?

  There are all these risks from draw­ing the at­ten­tion of hos­tile civ­i­liza­tions. I haven’t thought about what the risk and benefits are there. It feels like that came up in Precipice too, but I could be mix­ing some­thing up.

• Denis Drescher 25 Apr 2020 11:40 UTC

  9 points

  on: Does Utili­tar­ian Longter­mism Im­ply Directed Pansper­mia?

  Such an effort would likely be ir­re­versible or at least very slow and costly to re­verse. It comes at an im­mense cost of op­tion value.

  Directed pansper­mia also bears greater av­er­age-case risks than a con­trol­led ex­pan­sion of our civ­i­liza­tion be­cause we’ll have less con­trol over the func­tion­ing and the welfare stan­dards of the civ­i­liza­tion (if any) and thus the welfare of the in­di­vi­d­u­als at the des­ti­na­tion.

  Toby Ord’s Precipice more or less touches on this:

  Other bold ac­tions could pose similar risks, for in­stance spread­ing out be­yond our So­lar Sys­tem into a fed­er­a­tion of in­de­pen­dent wor­lds, each drift­ing in its own cul­tural di­rec­tion.

  This is not to re­ject such changes to the hu­man con­di­tion—they may well be es­sen­tial to re­al­iz­ing hu­man­ity’s full po­ten­tial. What I am say­ing is that these are the kind of bold changes that would need to come af­ter the Long Reflec­tion. Or at least af­ter enough re­flec­tion to fully un­der­stand the con­se­quences of that par­tic­u­lar change. We need to take our time, and choose our path with great care. For once we have ex­is­ten­tial se­cu­rity we are al­most as­sured suc­cess if we take things slowly and care­fully: the game is ours to lose; there are only un­forced er­rors.

  Ab­sent the break­throughs the Long Reflec­tion will hope­fully, we can’t even be sure that the moral value of a species spread out across many so­lar sys­tems will be pos­i­tive even if its ex­pected ag­gre­gate welfare is pos­i­tive. They may not be will­ing to trade suffer­ing and hap­piness 1:1.

  I could imag­ine benefits in sce­nar­ios where Earth gets locked into some small-scale, sta­ble, but un­de­sir­able state. Then there’d still be a chance that an­other civ­i­liza­tion emerges el­se­where and ex­pands to re­claim the space around our so­lar sys­tem. (If they be­come causally dis­con­nected from us be­fore they reach that level of ca­pa­bil­ity, they’re prob­a­bly not so differ­ent from any in­de­pen­dently evolved life el­se­where in the uni­verse.) But that would come at a great cost.

  The ap­proach seems similar to that of r-strate­gist species that have hun­dreds of offspring of which on av­er­age only two sur­vive. Th­ese are thought to be among the ma­jor sources of dis­value in na­ture. In the case of di­rected pansper­mia we could also be sure of the high de­gree of phe­nom­e­nal con­scious­ness of the quasi-offspring so that the ex­pected dis­value would be even greater than in the case of the r-strate­gist species where many of the offspring die while they’re still eggs.

  In most other sce­nar­ios, risks are ei­ther on a plane­tary scale or re­main so long as there aren’t any clusters that are so far apart as to be causally iso­lated. So in those sce­nar­ios, an ex­pan­sion be­yond our so­lar sys­tem would buy min­i­mal risk re­duc­tion. That can be achieved at a much lesser cost in terms of ex­pected dis­value.

  So I’d be more com­fortable to defer such grave and near-ir­re­versible de­ci­sions to fu­ture gen­er­a­tions that have de­liber­ated all their as­pects and im­pli­ca­tions thor­oughly and even-hand­edly for a long time and have reached a widely shared con­sen­sus.

• Denis Drescher 23 Apr 2020 15:44 UTC

  3 points

  on: De­nis Drescher’s Shortform

  [“If you value fu­ture peo­ple, why do you con­sider near term effects?” by Alex HT: Per­sonal take­aways.]

  I find it dis­con­cert­ing that there are a lot of very smart peo­ple in the EA com­mu­nity who fo­cus more on near-term effects than I cur­rently find rea­son­able.

  “If you value fu­ture peo­ple, why do you con­sider near term effects?” by Alex HT makes the case that a lot of rea­sons to fo­cus on near-term effects fall short of be­ing per­sua­sive. The case is based cen­trally on com­plex clue­less­ness. It closes with a se­ries of pos­si­ble ob­jec­tions and why they are not per­sua­sive. (Alex also cites the amaz­ing ar­ti­cle “Growth and the case against ran­domista de­vel­op­ment.”)

  The ar­ti­cle in­vites a dis­cus­sion, and Michael St. Jules re­sponded by ex­plain­ing the shape of a util­ity func­tion (bounded above and be­low) that would lead to a near term fo­cus and why it is a sen­si­ble util­ity func­tion to have. This seems to be a com­mon rea­son to pre­fer near-term in­ter­ven­tions, judg­ing by the num­ber of up­votes.

  There are also hints in the dis­cus­sion of whether there may be a rea­son to fo­cus on near-term effects as a Schel­ling point in co­or­di­na­tion prob­lem with fu­ture gen­er­a­tions. But that point is not fully de­vel­oped, and I don’t think I could steel­man it.

  I’ve heard smart peo­ple ar­gue for the mer­its of bounded util­ity func­tions be­fore. They have a num­ber of mer­its – avoid­ing Pas­cal’s mug­ging, the St. Peters­burg game, and more. (Are there maybe even some benefits for deal­ing with in­finite ethics?) But they’re also awfully un­in­tu­itive to me.

  Be­sides, I wouldn’t know how to se­lect the right pa­ram­e­ters for it. With some pa­ram­e­ters, it’ll be nearly lin­ear in a third-de­gree-polyno­mial in­crease in ag­gre­gate pos­i­tive or nega­tive valence over the com­ing mil­len­nium, and that may be enough to pre­fer cur­rent longter­mist over cur­rent near-ter­mist ap­proaches.

  Re­lated: https://​​globalpri­ori­tiesin­sti­tute.org/​​chris­tian-tarsney-the-epistemic-challenge-to-longter­mism/​​

De­nis Drescher’s Shortform

• Denis Drescher 5 Apr 2020 11:42 UTC

  2 points

  on: (How) Could an AI be­come an in­de­pen­dent eco­nomic agent?

  I’ve read some doc­u­ments where the de­vel­op­ers of a cryp­tocur­rency were wor­ried that it might be­come pos­si­ble to re­store a lot of lost crypto that no one cur­rently has ac­cess too – pre­sum­ably be­cause it might lead to an in­fla­tion? I don’t re­mem­ber where I read it or what the con­crete con­cerns are. Maybe some­one with more blockchain knowl­edge can fill in the de­tails.

• Denis Drescher 5 Feb 2020 19:06 UTC

  4 points

  on: Good Done Right conference

  Great sleuthing! I re­mem­ber my sur­prise in late 2014, months af­ter the con­fer­ence, when I found out that some­thing with such a roaster of speak­ers had hap­pened, and I hadn’t no­ticed it. Luck­ily I also found that in­valuable Sound­cloud ac­count. :-D

• Denis Drescher 12 Jan 2020 14:06 UTC

  2 points

  on: Co­or­di­nat­ing Com­mit­ments Through an On­line Service

  I love this idea! Of course you’d want to test the de­mand for it cheaply. Maybe there is already a Kick­starter-like plat­form where you need to meet a min­i­mum num­ber of con­trib­u­tors rather than just a min­i­mum to­tal con­tri­bu­tion (or a max­i­mum con­tri­bu­tion per con­trib­u­tor). Then you could just use that plat­form for a test run. If not among Kick­starter-like plat­forms, then maybe among pe­ti­tion plat­forms? Or you could re­pur­pose a mere Mailchimp newslet­ter signup for this! You could style it to look like a solemn sign­ing of a con­di­tional pledge.

  If there is such a plat­form, you could see if you can get a char­ity like An­i­mal Equal­ity on board with the ex­per­i­ment, one that has a sub­stan­tial on­line au­di­ence. (They’ll also be happy about the newslet­ter signups, though that should re­quire a sep­a­rate opt-in.)

  Fi­nally, you could run quick anony­mous sur­veys of the par­ti­ci­pants: What did they do, and what would they have done with­out the cam­paign. Per­haps one af­ter a month and one af­ter a year or so. (It would also be in­ter­est­ing again to fol­low up af­ter sev­eral years be­cause of ve­gan re­ci­di­vism, which usu­ally sets in af­ter around 7 years afaik.)

  Maybe you can even do all of that with­out any cod­ing.

• Denis Drescher 18 Sep 2019 23:00 UTC

  8 points

  on: What things do you wish you dis­cov­ered ear­lier?

  My lat­est break­through: Text to speech read­ers set to a high read­ing rate. (Plus cheap and good-qual­ity Mpow Blue­tooth head­phones.)

  Now I effortlessly plow through my read­ing list at up to 400 words per minute dur­ing com­mutes. My micro­morts/​km are lower too since I don’t need to keep my eyes glued to the Kin­dle.

  I started out at lower rates and then grad­u­ally in­creased them over the course of a few weeks as the lower rates started to feel slow.

  I mostly use @Voice Aloud Reader and Pocket for An­droid.

  PS: Thanks for your recom­men­da­tions, es­pe­cially Ot­ter!

• Denis Drescher 20 Aug 2019 18:47 UTC

  12 points

  in reply to: William_MacAskill’s comment on: Ask Me Any­thing!

  ∸

• Denis Drescher 20 Aug 2019 18:28 UTC

  10 points

  in reply to: Denis Drescher’s comment on: Ask Me Any­thing!

  Do I un­der­stand you cor­rectly that you’re rel­a­tively less wor­ried about ex­is­ten­tial risks be­cause you think they are less likely to be ex­is­ten­tial (that civ­i­liza­tion will re­bound) and not be­cause you think that the typ­i­cal global catas­tro­phes that we imag­ine are less likely?

• Denis Drescher 20 Aug 2019 18:25 UTC

  12 points

  in reply to: William_MacAskill’s comment on: Ask Me Any­thing!

  I’d like to vote for more de­tail on:

  I find (non-ex­tinc­tion) tra­jec­tory change more com­pel­ling as a way of in­fluenc­ing the long-run fu­ture than I used to.

  Un­less the change in im­por­tance is fully ex­plained by the rel­a­tive repri­ori­ti­za­tion af­ter up­dat­ing down­ward on ex­is­ten­tial risks.

• Denis Drescher 14 Aug 2019 16:03 UTC

  3 points

  on: Seek­ing feed­back on Cause Pri­ori­ti­za­tion Platform

  Cool idea!

  A few things I just no­ticed:

  1. It would be nice if the re­order­ing hap­pened only on reload and not im­me­di­ately. I some­times ticked an “im­por­tance” box fully in­tent to also tick an­other one, but then the block jumped away, and I’d have to go look­ing for it again.

  2. I sec­ond Saulius’s feed­back: I also feel like it’d be eas­ier for me to rate things on a three- or five-point scale or even with a slider. As it is, I’m tempted to tick al­most all the boxes on all prob­lems.

  3. I re­ally like your ini­tial se­lec­tion. I had a hard time vot­ing on an­i­mal welfare be­cause my as­sess­ments of farmed and wild an­i­mal welfare are so differ­ent. Then I no­ticed that you had an­ti­ci­pated my prob­lem and added the sub­sec­tions!

  4. The term tractable seems more in­tu­itive to me than solv­able, since the sec­ond can be mi­s­un­der­stood as whether the prob­lem can be fully solved as op­posed to the defi­ni­tion that Saulius cites.

  5. I’d like to make a dis­tinc­tion (and maybe a gran­u­lar one) be­tween whether I think a prob­lem is not im­por­tant/​ne­glected/​tractable or whether I have no opinion on it. E.g., I cur­rently don’t know whether the Pain Gap is tractably ad­dress­able, but not tick­ing the box feels as if I were say­ing that it’s not tractable.

  6. Not show­ing other peo­ple’s rat­ings could help avoid an­chor­ing peo­ple.

• Denis Drescher 5 Aug 2019 10:12 UTC

  5 points

  on: What types of or­ga­ni­za­tions would be ideal to be dis­tribut­ing fund­ing for EA? (Fel­low­ships, Or­ga­ni­za­tions, etc)

  A re­mote re­search or­ga­ni­za­tion for fund­ing and co­or­di­nat­ing on­go­ing in­di­vi­d­ual re­mote re­search efforts.

  I’d be in­ter­ested in this! In terms of co­or­di­na­tion, it could (1) pool in­for­ma­tion on what or­ga­ni­za­tions and in­di­vi­d­u­als are already work­ing on to al­low co­op­er­a­tion and men­tor­ship, and avoid du­pli­ca­tion un­less du­pli­ca­tion is use­ful; (2) give guidance on pri­ori­ti­za­tion of re­search ideas; and (3) provide a cu­rated, ac­cessible, and well cat­e­go­rized out­let (e.g., blog) for re­searchers that is suffi­ciently high-qual­ity that a lot of peo­ple read it.