Halstead

• Halstead 17 May 2019 13:31 UTC

  2 points

  in reply to: Larks's comment on: The case for de­lay­ing so­lar geo­eng­ineer­ing research

  I agree it’s not tech­ni­cally the right name, but peo­ple gen­er­ally know what it means which was im­por­tant for a blog­post. In the pa­per I ac­tu­ally call it the miti­ga­tion ob­struc­tion ar­gu­ment. I ex­plic­itly dis­cuss the ir­ra­tional­ity as­sump­tion re­quired for the miti­ga­tion ob­struc­tion ar­gu­ment in my pa­per. I think the ques­tion of how ir­ra­tionally peo­ple/​gov­ern­ments will re­spond to re­search is an open one.

• Halstead 3 May 2019 8:26 UTC

  11 points

  in reply to: Sean_o_h's comment on: Cen­tre for the Study of Ex­is­ten­tial Risk Six Month Re­port: Novem­ber 2018 - April 2019

  3. I have a scep­ti­cal prior against EU stud­ies of sci­en­tific is­sues be­cause the EU has taken an anti-sci­ence stance on many is­sues un­der pres­sure from the en­vi­ron­men­tal move­ment—see e.g. the effec­tive pro­hi­bi­tion of GMOs. The fact that the re­port you cite ad­vo­cates for in­creased or­ganic farm­ing adds weight to my scep­ti­cism. The re­port also says that the es­ti­mate of the eco­nomic costs is ex­tremely un­cer­tain and po­ten­tially a mas­sive over­es­ti­mate.

  4. There are many things in the world that im­pose sub­stan­tial eco­nomic costs, in­clud­ing in­effi­cienct tax­a­tion, labour mar­ket reg­u­la­tion, failure to in­vest in R&D, etc. While they may in­deed cre­ate eco­nomic costs, I fail to see the con­nec­tion to ex­is­ten­tial risk.

  5. While it is a small part of your port­fo­lio, there is limited poli­ti­cal at­ten­tion for ex­is­ten­tial risk, and if CSER does start ad­vo­cat­ing for the view that bio­di­ver­sity loss de­serves se­ri­ous con­sid­er­a­tion as a fac­tor rele­vant to ex­is­ten­tial risk, that comes at a cost. In this case, the fact that Partha Das­gupta is an in­fluen­tial per­son is a nega­tive be­cause he risks dis­tract­ing poli­cy­mak­ers from the gen­uine risks

• Halstead 3 May 2019 8:05 UTC

  10 points

  in reply to: HaydnBelfield's comment on: Cen­tre for the Study of Ex­is­ten­tial Risk Six Month Re­port: Novem­ber 2018 - April 2019

  There are lots of risk fac­tors for so­cietal re­silience to catas­tro­phes, in­clud­ing all con­tem­po­rary poli­ti­cal and eco­nomic prob­lems. The key ques­tion is how much of a risk they are and I have yet to see any ev­i­dence that bio­di­ver­sity loss is among the top ones.

• Halstead 2 May 2019 16:28 UTC

  47 points

  in reply to: HaydnBelfield's comment on: Cen­tre for the Study of Ex­is­ten­tial Risk Six Month Re­port: Novem­ber 2018 - April 2019

  Can you ex­plain what the mechanism is whereby bio­di­ver­sity loss cre­ates ex­is­ten­tial risk? And if bio­di­ver­sity loss is an ex­is­ten­tial risk, how big a risk is it? Should 80k be get­ting peo­ple to go into con­ser­va­tion sci­ence or not?

  There are in­de­pen­dent rea­sons to think that the risk is neg­ligible. Firstly, ac­cord­ing to wikipe­dia, dur­ing the Eocene pe­riod ~65m years ago, there were thou­sands fewer gen­era than to­day. We have made ~1% of species ex­tinct, and we would have to con­tinue at cur­rent rates of species ex­tinc­tions for at least 200 years to re­turn to Eocene lev­els of bio­di­ver­sity. And yet, even though sig­nifi­cantly warmer than to­day, the Eocene marked the dawn of thou­sands of new species. So, why would we ex­pect the world 200 years hence to be in­hos­pitable to hu­mans if it wasn’t in­hos­pitable for all of the species emerg­ing in the Eocene, who are/​were sig­nifi­cantly less nu­mer­ous than hu­mans and sig­nifi­cantly less ca­pa­ble of a ra­tio­nal re­sponse to prob­lems?

  Se­condly, as far as I am aware, ev­i­dence for pres­sure-in­duced non-lin­ear ecosys­tem shifts is very limited. This is true for a range of ecosys­tems. Lin­ear ecosys­tem dam­age seems to be the norm. If so, this leaves more scope for learn­ing about the costs of our dam­age to ecosys­tems and cor­rect­ing any dam­age we have done.

  Thirdly, ecosys­tem ser­vices are over­whelm­ingly a func­tion of the re­la­tions within lo­cal ecosys­tems, rather than of global trends in bio­di­ver­sity. Upon dis­cov­er­ing Hawaii, the Poly­ne­si­ans elimi­nated so many species that global decadal ex­tinc­tion rates would have been ex­cep­tional. This has next to no bear­ing on ecosys­tem ser­vices out­side Hawaii. Hu­man­ity is an in­tel­li­gent species and will be able to see if other re­gions are suffer­ing from bio­di­ver­sity loss and make ad­just­ments ac­cord­ingly. Why would all re­gions be so stupid as to ig­nore les­sons from el­se­where? Also, is bio­di­ver­sity ac­tu­ally de­creas­ing in the rich world? I know for­est cover is in­creas­ing in many places. Pop­u­la­tion is set to de­cline in many rich coun­tries in the near fu­ture, and en­vi­ron­men­tal im­pact per per­son is de­clin­ing on many met­rics.

  I also find it sur­pris­ing that you cite the Kareiva and Car­ranza pa­per in sup­port of your claims, for this pa­per in fact di­rectly con­tra­dicts them:

  “The in­ter­est­ing ques­tion is whether any of the plane­tary thresh­olds other than CO2 could also por­tend ex­is­ten­tial risks. Here the an­swer is not clear. One bound­ary of­ten men­tioned as a con­cern for the fate of global civ­i­liza­tion is bio­di­ver­sity (Ehrlich & Ehrlich, 2012), with the pro­posed safety thresh­old be­ing a loss of greater than 0.001% per year (Rock­ström et al., 2009). There is lit­tle ev­i­dence that this par­tic­u­lar 0.001% an­nual loss is a thresh­old—and it is hard to imag­ine any data that would al­low one to iden­tify where the thresh­old was (Brook, Ellis, Per­ring, Mackay, & Blomqvist, 2013; Len­ton & Willi­ams, 2013). A bet­ter ques­tion is whether one can imag­ine any sce­nario by which the loss of too many species leads to the col­lapse of so­cieties and en­vi­ron­men­tal dis­asters, even though one can­not know the ab­solute num­ber of ex­tinc­tions that would be re­quired to cre­ate this dystopia.

  While there are data that re­late lo­cal re­duc­tions in species rich­ness to al­tered ecosys­tem func­tion, these re­sults do not point to sub­stan­tial ex­is­ten­tial risks. The data are small-scale ex­per­i­ments in which plant pro­duc­tivity, or nu­tri­ent re­ten­tion is re­duced as species num­bers de­cline lo­cally (Vel­lend, 2017), or are lo­cal ob­ser­va­tions of in­creased vari­abil­ity in fish­eries yield when stock di­ver­sity is lost (Sch­indler et al., 2010). Those are not ex­is­ten­tial risks. To make the link even more ten­u­ous, there is lit­tle ev­i­dence that bio­di­ver­sity is even de­clin­ing at lo­cal scales (Vel­lend et al., 2013, Vel­lend et al., 2017). To­tal plane­tary bio­di­ver­sity may be in de­cline, but lo­cal and re­gional bio­di­ver­sity is of­ten stay­ing the same be­cause species from el­se­where re­place lo­cal losses, albeit ho­mog­e­niz­ing the world in the pro­cess. Although the ma­jor­ity of con­ser­va­tion sci­en­tists are likely to flinch at this con­clu­sion, there is grow­ing skep­ti­cism re­gard­ing the strength of ev­i­dence link­ing trends in bio­di­ver­sity loss to an ex­is­ten­tial risk for hu­mans (Maier, 2012; Vel­lend, 2014). Ob­vi­ously if all bio­di­ver­sity dis­ap­peared civ­i­liza­tion would end—but no one is fore­cast­ing the loss of all species. It seems plau­si­ble that the loss of 90% of the world’s species could also be apoc­a­lyp­tic, but not one is pre­dict­ing that de­gree of bio­di­ver­sity loss ei­ther. Tragic, but plau­si­ble is the pos­si­bil­ity of our planet suffer­ing a loss of as many as half of its species. If global bio­di­ver­sity were halved, but at the same time lo­cally the num­ber of species stayed rel­a­tively sta­ble, what would be the mechanism for an end-of-civ­i­liza­tion or even end of hu­man pros­per­i­tysce­nario? Ex­tinc­tions and bio­di­ver­sity loss are eth­i­cal and spiritual losses, but per­haps not an ex­is­ten­tial risk.”

• Halstead 18 Apr 2019 20:06 UTC

  10 points

  in reply to: dpiepgrass's comment on: Does cli­mate change de­serve more at­ten­tion within EA?

  En­ergy for Hu­man­ity is a great un­der­funded pro-nu­clear NGO work­ing in the EU. Clean Air Task Force and Third Way are also great.

  I also think the cur­rent em­pha­sis on so­lar and wind in some places could be a bar­rier to sen­si­ble low car­bon poli­cies in the long-term, es­pe­cially as they don’t go very well with nu­clear. It also doesn’t make a great deal of sense to com­bine in­ter­mit­tent re­new­ables with nu­clear, as France bizarrely re­cently con­sid­ered do­ing, since it just makes nu­clear run be­low ca­pac­ity when the sun is shin­ing, which doesn’t make eco­nomic sense.

• Halstead 18 Apr 2019 20:02 UTC

  2 points

  in reply to: bdixon's comment on: Does cli­mate change de­serve more at­ten­tion within EA?

  I’ll fo­cus on point 2 be­cause I think it is the most im­por­tant. I don’t see the ar­gu­ment for it be­ing true that for the vast ma­jor­ity of peo­ple, work­ing on cli­mate change promises more lev­er­age on the prob­lem of nu­clear war, than does work­ing di­rectly on nu­clear war. Nu­clear war is eas­ier to make progress on, more ne­glected and more im­por­tant than cli­mate change.

• Halstead 18 Apr 2019 19:56 UTC

  4 points

  in reply to: StevenKaas's comment on: Does cli­mate change de­serve more at­ten­tion within EA?

  Yes I think you are in fact right that plau­si­ble pri­ors do seem to ex­clude ECS above 5 de­grees.

  You pick out a ma­jor prob­lem in draw­ing con­clu­sions about ECS—the IPCC does not ex­plain how they ar­rive at their pdf of ECS and the es­ti­mate seems to be pro­duced some­what sub­jec­tively from var­i­ous cur­rent es­ti­mates from in­stru­men­tal and pa­le­o­cli­matic data and from their own ex­pert judge­ment as to what weight to give to differ­ent stud­ies. I think this means that they give some weight to pdfs with a very fat tail, which seems to be wrong, given their use of uniform pri­ors. This might mean that their tail es­ti­mate is too high

• Halstead 18 Apr 2019 19:34 UTC

  44 points

  in reply to: Ben_Kuhn's comment on: Does cli­mate change de­serve more at­ten­tion within EA?

  I agree that the en­vi­ron­men­tal move­ment is ex­tremely poor at op­ti­mi­sa­tion. This be­ing said, there are a num­ber of very large philan­thropists and char­i­ties who do take a sen­si­ble ap­proach to cli­mate change, so I don’t think this is a case in which EAs could march in and to­tally change ev­ery­thing. Much of Cli­mate­works’ giv­ing takes a broadly EA ap­proach, and they over­see the giv­ing of nu­mer­ous multi-billion dol­lar foun­da­tions. Gates also does some sen­si­ble work on the en­ergy in­no­va­tion side. Nev­er­the­less, most money in the space does seem to be spent very badly, e.g. on op­pos­ing nu­clear power. This con­sid­er­a­tion might even make the en­vi­ron­men­tal move­ment net nega­tive wrt cli­mate, though I haven’t crunched any num­bers on that.

  I would also add that sen­si­ble EA an­swers in this space face sub­stan­tial op­po­si­tion from the en­vion­men­tal move­ment. I think a ra­tio­nal anal­y­sis ar­gues in favour of ad­vo­cat­ing for nu­clear and car­bon cap­ture, for en­ergy in­no­va­tion in gen­eral, and for fi­nan­cial in­cen­tives for pre­vent­ing de­foresta­tion. All of these things are op­posed quite strongly by differ­ent con­stituen­cies in the en­vi­ron­men­tal move­ment. Maybe the one thing most peo­ple can agree on is car­bon pric­ing, but that is hard to get through for other reasons

• Halstead 17 Apr 2019 21:12 UTC

  21 points

  in reply to: StevenKaas's comment on: Does cli­mate change de­serve more at­ten­tion within EA?

  On Bayesi­anism—this is an im­por­tant point. The very heavy tailed es­ti­mates all use a “zero in­for­ma­tion” prior with an ar­bi­trary cut-off at eg 10 de­grees or 20 de­grees. (I dis­cuss this in my write-up). This is flawed and more plau­si­ble pri­ors are available which thin out the tails a lot.

  How­ever, I don’t think you need this to get to there be­ing sub­stan­tial tail risk. Eye­bal­ling the ECS es­ti­mates that use plau­si­ble pri­ors, there’s still some­thing like a 1-5% chance of ECS be­ing >5 de­grees, which means that from 1.5 dou­blings of GHG con­cen­tra­tions, which seems plau­si­ble, there’s a 1-5% of ~7 degrees

• Halstead 17 Apr 2019 21:04 UTC

  40 points

  on: Does cli­mate change de­serve more at­ten­tion within EA?

  Thanks for this. It’s use­ful for the com­mu­nity to think about this kind of thing and this is well-ar­gued.

  Overall

  1. It’s a good point that since the top AI fields seem over­sub­scribed, it might be worth some peo­ple mov­ing into the next best causes. Another pos­si­bil­ity is that they should wait un­til the num­ber of or­gani­sa­tions catches up with the num­ber of peo­ple. It might even be that the most valuable op­tions is hav­ing a re­serve of a large num­ber of peo­ple who could, with some prob­a­bil­ity, be a good fit for the high­est-im­pact orgs, even though most of these peo­ple never end up work­ing for high-im­pact orgs. This puts a new slant on the de­mand­ing­ness of EA: rather than mak­ing sac­ri­fices by donat­ing, EAs make sac­ri­fices by be­ing pre­pared to ac­cept the sub­stan­tial prob­a­bil­ity of them­selves never hav­ing im­pact. This would be hard to take psy­cholog­i­cally, but might be the right thing to do in a crowded tal­ent space.

  2. On in­di­rect risks, an­other point I make in the FP re­port is that while cli­mate change is an in­di­rect stres­sor of other risks, this sug­gests to me that work­ing on those ter­mi­nal risks di­rectly would be a bet­ter bet than work­ing on cli­mate change since cli­mate change is such an in­di­rect stres­sor, is very crowded and seems difficult to make progress on. What do you think of that ar­gu­ment?

  ITN

  3. I don’t think it is right that prob­lems with high tractabil­ity should be de-pri­ori­tised. I think what you mean is that we should fo­cus on things that shift the long-term tra­jec­tory of hu­man­ity. But these could be highly tractable. e.g. the prob­lem of not start­ing nu­clear war was tractable for Vasili Arkhipov, but plau­si­bly had large long-term effects. Hav­ing looked at it in some depth, cli­mate change does look an in­tractable prob­lem over­all and this is in­deed a rea­son not to work on it.

  4. Another good point on how there could be in­creas­ing re­turns to scale in cli­mate change, as we could af­fect the huge pool of funds go­ing to the space through en­gage­ment.

  5. Really, the ITN per­haps shouldn’t be used when we have cost-effec­tive­ness es­ti­mates. On the 80k ren­der­ing, ITN is liter­ally a cost-effec­tive­ness es­ti­mate. But we now have cost-effec­tive­ness es­ti­mates of cli­mate char­i­ties. If we can make plau­si­ble es­ti­mates of the im­pact of bio, AI and nu­clear, then we should use those, rather than ap­peal­ing to the ITN. similarly, for use of time as well as money.

  5. It is pre­ma­ture to say that work on cli­mate change could be tractable. I think care­ful anal­y­sis is needed to figure out whether the things you list are in­deed a good bet com­pared to other things that EAs could do.

  Details

  6. Cli­mate Ac­tion Tracker sug­gests that on cur­rent policy, we are in for 3.1 to 3.5C, which is differ­ent to the ‘baseline’ tra­jec­tory es­ti­mate that you give. I think the cur­rent policy tra­jec­tory is most rele­vant for that part of your ar­gu­ment. (But note that this is only by 2100)

  7. The im­pact of cli­mate change on food pro­duc­tion is in fact pre­dicted to be fairly mod­est, as I dis­cuss here. Yields might fall by 10-20% but this will be in the con­text of ris­ing pro­duc­tivity and im­prove­ment in the other fac­tors that de­ter­mine the sup­ply of food.

  8. The em­pha­sis on wa­ter short­age through­out is a bit overblown. We don’t need to ra­tion wa­ter, we just need to price it prop­erly (which is effi­cient ra­tioning). If we did that, there would be no wa­ter prob­lems to­day or in the fu­ture, any­where (pro­vided peo­ple had enough money).

• Halstead 31 Mar 2019 19:52 UTC

  2 points

  in reply to: dpiepgrass's comment on: The case for de­lay­ing so­lar geo­eng­ineer­ing research

  2. I don’t think this is right, for rea­sons dis­cussed in this Na­ture pa­per. Firstly, so­lar geo­eng­ineer­ing could be used to slow the rate of warm­ing even if it is de­ployed tem­porar­ily. You could de­ploy it over e.g. a fifty year pe­riod and thereby de­lay the point at which we reach peak warm­ing, and then ta­per it out grad­u­ally. Se­condly, as you say, an ex­cep­tion is if CO2 emis­sions stay above zero. So­lar geo­eng­ineer­ing could in prin­ci­ple buy us time to abate emis­sions and to take CO2 out of the at­mo­sphere in which case it would not have to be de­ployed for the full life­time of CO2 in the at­mo­sphere. In this case, so­lar geo would slow the rate of warm­ing and re­duce peak warm­ing.

  Thirdly, I don’t see why so­lar geo­eng­ineer­ing would ever be stopped sud­denly once we started. The rea­sons for this are dis­cussed in the Parker and Irv­ine piece on so­lar geo­eng­ineer­ing. All coun­tries would have a rea­son to pre­vent it from stop­ping sud­denly and would have the means to do so given how cheap it is. A catas­tro­phe caus­ing ter­mi­na­tion would have to be ex­traor­di­nar­ily spe­cific.

  3. To clar­ify, is your point here that we should fo­cus on miti­ga­tion be­cause then we’ll be left with some spare oil come a later catas­tro­phe?

[Link] New Founders Pledge re­port on ex­is­ten­tial risk

• Halstead 27 Mar 2019 18:54 UTC

  2 points

  in reply to: andrewpei's comment on: The case for de­lay­ing so­lar geo­eng­ineer­ing research

  Hello,

  I’m not com­pletely sure I fol­low why your first para­graph is a cri­tique. I don’t ex­pect gov­er­nance to im­prove on its own. My claim is that we do not need 50 years of gov­er­nance re­search to get gov­er­nance to a suffi­ciently good level should we need to de­ploy so­lar geo­eng­ineer­ing in the fu­ture. The hope is that we will be wise enough not to have to use it be­cause we will start se­ri­ous miti­ga­tion, and I’m wor­ried that geo­eng­ineer­ing re­search could be one of many fac­tors that could de­rail those efforts.

  It is true that de­vel­op­ing geo­eng­ineer­ing tech­nol­ogy would cre­ate in­cen­tives to im­prove gov­er­nance mechanisms for geo­eng­ineer­ing. I’m not sure why that is a cri­tique of my ar­gu­ment.

  I agree that war is un­likely for the rea­sons you out­line.

• Halstead 26 Mar 2019 16:47 UTC

  10 points

  in reply to: Milan_Griffes's comment on: Apology

  Was deleted for tone, no in­ter­est­ing content

• Halstead 25 Mar 2019 22:42 UTC

  9 points

  in reply to: Holly_Elmore's comment on: Apology

  ok thanks, un­der­stood. i hope it wasn’t grasp­ing at straws, but maybe this de­bate has got too side­tracked and should draw to a close.

• Halstead 25 Mar 2019 19:23 UTC

  11 points

  in reply to: Holly_Elmore's comment on: Apology

  We were de­bat­ing the claim “Hmm, it is not at all clear to me that the ac­cu­sa­tions that are be­ing dis­cussed here [the Brown ac­cu­sa­tions] are sep­a­rate from the ac­cu­sa­tions that ap­pear to have caused his apol­ogy.” Ju­lia Wise’s com­ments has con­firmed that the claims were sep­a­rate. The term ‘sep­a­rate’ here means ‘differ­ent in­stance of sex­ual ha­rass­ment’.

• Halstead 25 Mar 2019 19:11 UTC

  10 points

  in reply to: Holly_Elmore's comment on: Apology

  The ques­tion is about prob­a­bil­ities of guilt/​in­no­cence. If you have mul­ti­ple peo­ple ac­cuse you of sex­ual or non-sex­ual ha­rass­ment over the course of at least 7 years in differ­ent com­mu­ni­ties, then you are ei­ther ex­tremely un­lucky or you have ac­tu­ally ha­rassed peo­ple. He also ad­mits guilt

• Halstead 25 Mar 2019 19:03 UTC

  3 points

  in reply to: Habryka's comment on: Apology

  I know the two are not the same—this ar­gu­ment was about your claim: “Hmm, it is not at all clear to me that the ac­cu­sa­tions that are be­ing dis­cussed here [the Brown ac­cu­sa­tions] are sep­a­rate from the ac­cu­sa­tions that ap­pear to have caused his apol­ogy.”

• Halstead 25 Mar 2019 18:58 UTC

  −10 points

  in reply to: Holly_Elmore's comment on: Apology

  .

• Halstead 25 Mar 2019 18:53 UTC

  1 point

  in reply to: Holly_Elmore's comment on: Apology

  .