So far, it seems as though peo­ple have been treat­ing them like non-in­ten­sive posts (e.g. friendly com­men­tary/​push­back, no point-by-point crit­i­cism). That seems good to me, but my vi­sion of how Short­form should work may change as the fea­ture gets used more of­ten.

“Triv­ial, but in a Derek Parfit way” is hon­estly the high­est com­pli­ment I could ever re­ceive.

Thanks for post­ing this ques­tion! It seems pretty tricky to figure out the con­nec­tions be­tween short-term an­i­mal welfare and long-term value, and I’m glad you sparked more dis­cus­sion on the sub­ject.

(I think “if the con­tribu­tive ax­iolog­i­cal value of peo­ple is nega­tive, then prefer­ring smaller pop­u­la­tions is con­sis­tent with—in­deed, im­plied by—to­tal­ism in pop­u­la­tion ethics” is a valid point, and ob­vi­ously so. It is also men­tioned by Spears in the pa­per I cite above. I there­fore find it quite ir­ri­tat­ing that the par­ent com­ment was ap­par­ently strongly down­voted. Cu­ri­ous if I’m miss­ing a rea­son for this?

NB I also think the point is triv­ial and has an im­plau­si­ble premise, but IMO it is the hal­l­mark of good philos­o­phy that each in­di­vi­d­ual state­ment seems triv­ial—e.g., Rea­sons and Per­sons fea­tures an am­ple amount of such claims that might strike some read­ers as triv­ial or pedan­tic.)

Ques­tion 13 seems un­der-speci­fied to me, speci­fi­cally this part: “Their mem­bers are equally happy.” Does this mean their level of welfare is the same, but it could be at any level for the pur­poses of this ques­tion? Does the use of “happy” in par­tic­u­lar mean the ques­tion as­sumes this con­stant level of welfare is net pos­i­tive? Could the mag­ni­tudes of hap­piness and suffer­ing differ be­tween peo­ple as long as the “net welfare” is pos­i­tive, as­sum­ing it’s pos­si­ble to make that ag­gre­ga­tion?

I think these ques­tions mat­ter be­cause they in­fluence your in­ter­pre­ta­tion of the an­swers as ei­ther a re­sult of pop­u­la­tion eth­i­cal fac­tors, or other things like the re­spon­dents’ be­liefs about the moral weight of hap­piness vs suffer­ing. Some­one could co­her­ently ac­cept to­tal­ism yet con­sider the smaller world bet­ter if, for in­stance, they think the higher num­ber of cases of the ex­treme tails of suffer­ing in the larger pop­u­la­tion (just be­cause there are more peo­ple that things could go very wrong for) makes it worse.

A pri­ori I ex­pect suffer­ing fo­cused in­tu­itions to be in the minor­ity, but in any case it’s not ob­vi­ous that the an­swers to #13 re­veal non-to­tal­ist or ir­ra­tional pop­u­la­tion ethics among the re­spon­dents.

I think you ac­ci­den­tally posted this twice.

I ap­pre­ci­ate the other com­ments about how the model does not take into ac­count the base im­pact of the char­i­ties. Also:

1) Do trustees take on some form of le­gal re­spon­si­bil­ity for char­i­ties? If so, it could be risky to get in­volved with a char­ity you know lit­tle about.

2) Do you know how many other trustees are typ­i­cally already in­volved? If you are one of three, I could see you hav­ing an eas­ier time in­fluenc­ing a char­ity than if you are, say, one of seven.

I think the biggest im­prove­ment would be cor­rect­ing the fact that this model (ac­ci­den­tally, I think) as­sumes that im­prov­ing any ar­bi­trary high bud­get char­ity by 5% is equally as im­pact­ful as im­prov­ing a Givewell equiv­a­lent char­ity by 5%. Most char­ity’s im­pact is an or­der of mag­ni­tude smaller.

You could solve this with a mul­ti­plier for the char­ity’s im­pact at baseline.

If I un­der­stand cor­rectly, you figure that if you be­come a trustee of a £419668/​year bud­get char­ity, if only you can im­prove the cost effec­tive­ness by 5%, then you can di­vide that by 42 hours a year, to get £419668*5%/​42 hours=£500/​hour in the value of your donated time. (A style tip—it would be helpful to put the key equa­tion de­scribing roughly what you’ve done in the de­scrip­tion, to make it all leg­ible with­out hav­ing to go into the spread­sheet.)

I think it is fair to say that, were you to suc­cess­fully perform this feat, you have in­deed done some­thing roughly as im­pact­ful as pro­vid­ing a £500/​hour value to the char­ity you are trustee-ing for. So, if you im­proved a Givewell-top-char­ity-equiv­a­lent’s cost effec­tive­ness by 5% for a year, then maybe you could fairly take 5% of that char­ity’s yearly bud­get and di­vide it by your hours for that year, as you’ve done, to calcu­late your Givewell-top-char­ity-equiv­a­lent im­pact in terms of how it would com­pare to donated dol­lars.

But if you im­prove a £419668/​yr bud­get char­ity which is only 1% as cost-effec­tive as a Givewell-top-char­ity-equiv­a­lent by 5%, then that makes your hourly im­pact 1%*£419668*5%/​42 hours = £5/​hour of Givewell-top-char­ity-equiv­a­lent im­pact—you’d be bet­ter served work­ing a bit ex­tra and donat­ing 5 dol­lars to Givewell.

Yeah, I don’t think that peo­ple who worked at Lev­er­age are pretty likely to be racist.

A rea­son I con­sider what I de­scribed likely is not least that I find it more likely that fu­ture soft­ware sys­tems will con­sist in a mul­ti­tude of spe­cial­ized sys­tems with quite differ­ent de­signs, even in the pres­ence of AGI, as op­posed to most ev­ery­thing be­ing done by copies of some sin­gu­lar AGI sys­tem.

Can you ex­plain why this is rele­vant to how much effort we should put into AI al­ign­ment re­search to­day?

Thanks, these look like some in­ter­est­ing fea­tures.

Are there /​ should there be any so­cial norms re: re­ply­ing to some­one else’s short­form? They seem in­tu­itively sort of ‘pri­vate prop­erty’ to me.

I agree. If none of the char­i­ties rep­re­sented are ones which do things you think are valuable or can con­ceive of be­com­ing valuable I sug­gest peo­ple turn down the offer.

How­ever, per­haps you could get some bud­get moved from dog and cat shelters into fac­tory farm­ing or change the types of protests that your mem­bers at­tend. Some types of change would be effi­ciency though oth­ers could be new av­enues or pub­lish­ing im­pact. I think you are right to say you could get some sense of likely­hood of pos­i­tive im­pact on be­ing offered the trustee­ship.

Can you think of a way to in­clude this kind of vari­abil­ity sim­ply? Other­wise I guess there is no way to know whether this is a good idea or not and I’m not sure what to do with that.

Sorry for the lack of de­tail. First post and try­ing to figure out fo­rum.

How big a prob­lem is mi­suse of pes­ti­cides by la­bor­ers and farm­ers?

I am cur­rently work­ing on a pro­ject in In­dia try­ing to get la­bor­ers and farm­ers to use pes­ti­cides cor­rectly by us­ing be­hav­ioral sci­ence. Here are some of the con­se­quences that we are look­ing at from mi­suse.

Health

- Users do not wear gear, or even smoke and eat while spray­ing, and reg­u­larly in­hale pes­ti­cide and get it all over their skin. This can cause “headaches, nau­sea, eye ir­ri­ta­tion, skin rashes, and flu-like symp­toms. Long term chronic health prob­lems in­clude neu­rolog­i­cal and re­pro­duc­tive di­s­or­ders and var­i­ous forms of can­cer”

- Users of­ten do not shower, and so come home to their fam­ily cov­ered in pes­ti­cide, af­fect­ing the health of their fam­ily as well (es­pe­cially bad for preg­nant women)

- Users will some­times use pes­ti­cide con­tain­ers to store food and wa­ter. This can be dis­as­trous.

- Users will spray near rivers, or dis­card of pes­ti­cide con­tain­ers in fields or near rivers which can cause health is­sues for the sur­round­ing communities

- Agri­cul­ture is the in­dus­try with the most child la­bor, and chil­dren are more af­fected than adults. Thus the effects are longer last­ing and more se­vere than an ini­tial anal­y­sis might indicate

- Pests de­velop re­sis­tance due to mi­suse, mean­ing stronger chem­i­cals must be used, thus ex­as­per­at­ing the is­sues above.

En­vi­ron­ment and wildlife

- Pes­ti­cides in the ecosys­tem may be hurt­ing fish and other wildlife

- Mi­suse of pes­ti­cides leads to bugs get­ting sprayed and dy­ing slowly, or get­ting close to death and then re­cov­er­ing, de­vel­op­ing re­sis­tance. (Rele­vant if you care about the welfare of such pests)

- In­creased pro­duc­tivity of fields would mean less land would be needed to pro­duce the same amount of food.

- Pests de­velop re­sis­tance due to mi­suse, mean­ing stronger chem­i­cals must be used, thus ex­as­per­at­ing the is­sues above.

Money

- Mi­suse of pes­ti­cides leads to lower yields of crops which means lower rev­enue for the farmers

- Lower rev­enue de­presses the econ­omy, puts la­bor­ers out of work, and means the farmer is less likely to buy pro­tec­tive gear in the fu­ture or take other pre­cau­tions.

- And not all money is equal “growth of 1% in GDP from the agri­cul­tural sec­tor, re­sults in an in­crease of ex­pen­di­tures of the three poor­est deciles, by at least 2.5 times that of the rest of the econ­omy.” (http://​​www.expo2015.org/​​mag­a­z­ine/​​en/​​econ­omy/​​agri­cul­ture-re­mains-cen­tral-to-the-world-econ­omy.html)

- Pests de­velop re­sis­tance due to mi­suse, mean­ing stronger chem­i­cals must be used, thus ex­as­per­at­ing the is­sues above.

Food

- Mi­suse of pes­ti­cides leads to lower yields of crops mean­ing that food prices go up and there is less food for ev­ery­one.

- Pests de­velop re­sis­tance due to mi­suse, mean­ing stronger chem­i­cals must be used, thus ex­as­per­at­ing the is­sue above.

The pa­per ‘Pes­ti­cide Risk Per­cep­tions, Knowl­edge, and At­ti­tudes of Oper­a­tors, Work­ers, and Res­i­dents: A Re­view of the Liter­a­ture’ by Kyr­i­aki Re­moundou, Mary Bren­nan, Andy Hart & Lynn J. Frewer is a good re­source for get­ting an idea of the types of prob­lems and mi­suse that are oc­cur­ring, and some of the com­pli­ca­tions sur­round­ing it. But I don’t know any­one who has done work at try­ing to quan­tify how big a prob­lem it ac­tu­ally is. I sus­pect that the num­ber of peo­ple who have suffered ill health effects from mi­suse of pes­ti­cides is ex­tremely high, but it is not my do­main of ex­per­tise and so I am un­fa­mil­iar with the re­search.

Hey JP, don’t worry I won’t hold you to any­thing :-) I know you guys have a lot on. I think the fea­ture of set­ting your own mod­er­a­tion guidelines at a cer­tain karma level is a good one. It en­courages top posters and also en­courages the whole com­mu­nity to take more re­spon­si­bil­ity for good qual­ity con­ver­sa­tions. If you did get the chance to figure out how this was cur­rently work­ing on the EA Fo­rum (e.g. at what karma level to in­di­vi­d­u­als get what mod­er­a­tion fea­tures) and per­haps en­act some­thing similar to the LessWrong ver­sion that would be cool. Let me know what you end up de­cid­ing to do.

Thanks Saulius for point­ing this out. We spent a con­sid­er­able amount of time try­ing to find a more rele­vant num­ber, but that was the closes proxy we’ve found. If you have any other source we could use, that would in­crease the ac­cu­racy of the es­ti­mate.

It prob­a­bly de­pends on what you value. Based on your guess, for 42 hours of my time I could po­ten­tially cause:

-5% more cats adopted from a par­tic­u­lar shelter

-5% in­crease in aware­ness of autism at part­ner workplaces

-5% more opera performances

-5% more women at­tend­ing a par­tic­u­lar ‘women’s em­pow­er­ment’ conference

-5% more re­li­gious conversions

-5% in­crease in cita­tions from re­search com­mis­sioned by a par­tic­u­lar charity

-5% in­crease in protest attendance

Most of these things I imag­ine mov­ing from ‘good but not cost effec­tive’ to ‘good and still not all that cost effec­tive’ or ‘ba­si­cally use­less’ to ‘still use­less’. I don’t re­ally imag­ine many of these char­i­ties chang­ing cat­e­gories from ‘use­less’ to ‘good’ be­cause of a 5% in­crease in pro­duc­tivity.

When an­a­lyz­ing 35 In­de­pen­dent Pie­ces of Ev­i­dence for Why New Cor­po­rate Cam­paigns Might (or Might Not) Work, I’ve found some com­mon­al­ities:

1. Fol­low-through rate is lower in the case of com­pa­nies that af­fect the high­est num­ber of hens.

2. Fi­nan­cial situ­a­tion: When com­pa­nies broke or de­layed the com­mit­ment, they blamed the re­ces­sion or claimed that there were in­suffi­cient con­sumer de­mand and lack of funds for mak­ing the nec­es­sary changes on farms.

3. ACE claimed that the suc­cess of the cam­paigns de­pends on the pub­lic per­cep­tion of tar­geted is­sues. I have not fack-check that though.

Char­ac­ter­is­tics that don’t cor­re­late with fol­low-through:

1. No cor­re­la­tion be­tween the num­ber of pledges and the % of cage-free eggs in a given coun­try be­fore the cam­paign. Mean­ing that cam­paigns do not seem to be sim­ply rid­ing already ex­ist­ing trends.

I wouldn’t be sur­prised if those fac­tors were some­what pre­dic­tive:

1. If the pledge was made vol­un­tary form the com­pany or was forced upon by strong nega­tive cam­paign.

2. If a given com­pany gave it­self some wig­gle room in the phras­ing of their com­mit­ments, which they could later be used as a jus­tifi­ca­tion for break­ing their com­mit­ments.

3. What brand and pub­lic image they try to cre­ate

The model does noth­ing to con­sider the vari­abil­ity in im­pact­ful­ness of char­i­ties. Many peo­ple think that the ma­jor­ity of char­i­ta­ble in­ter­ven­tions achieve noth­ing, and it seems odd to as­cribe value to a char­ity whose work isn’t achiev­ing any­thing.

Think­ing again, I’m not sure I agree. The model as­sumes a 5% in­crease in effect. For a char­ity do­ing no im­pact­ful work that might mean mak­ing it slightly im­pact­ful, which doesn’t seem un­rea­son­able—get­ting them to make im­pact as­sess­ments could do far more good than this. What do you think?

1. Most of my time will be spent do­ing similar things to last year, al­though I’ve only been work­ing full time for 9 months. I think in­no­va­tion, at least for me, hap­pens over time rather than as a one off pro­cess, with ideas be­ing formed and then be­ing tested to see what peo­ple like. One re­cent ex­am­ple is the com­mu­nity spotlight, which was an idea a few weeks ago and now has be­come a reg­u­lar post.

In terms of tak­ing EA to the next level, there isn’t a plan to grow or hire more peo­ple but rather co­or­di­nate and sup­port nascent or­gani­sa­tions in Lon­don that can start do­ing other work and grow in­de­pen­dently.

2. I will have a look

There is a Face­book group and a small page on the web­site.