EA Hotel Fundraiser 6: Concrete outputs after 17 months

We have com­piled a list of con­crete out­puts that EA Ho­tel res­i­dents have pro­duced dur­ing their stays.

Like last time, we note that in­ter­pret­ing this list comes with some caveats:

  • While the out­put per dol­lar is high, this might not be the best met­ric. A bet­ter met­ric might be marginal out­put di­vided by marginal out­put for a coun­ter­fac­tual dona­tion, but this is hard to es­ti­mate. As a proxy, we sug­gest look­ing at out­put per per­son-day. Does this seem high com­pared to an av­er­age EA?

  • This list doesn’t cover ev­ery­thing of value that has been sup­ported by the Ho­tel. Some res­i­dents have spent months work­ing on things they haven’t pub­lished yet. Some have spent most of their time do­ing self-ther­apy, read­ing books or hav­ing a lot of re­search con­ver­sa­tions. Some have been de­vel­op­ing pre­sen­ta­tions and work­shops. The Ho­tel is in a unique po­si­tion to sup­port this kind of hard-to-ver­ify work. To in­crease trans­parency, guests are en­couraged to share in­for­mal progress up­dates (pre­sented in em­bed­ded Google Docs on our web­site), to give out­siders an idea of what they are do­ing with their time.

  • Out­puts are pre­sented along with their coun­ter­fac­tual like­li­hood of hap­pen­ing with­out the Ho­tel. This gives an idea of the value the Ho­tel is adding, es­pe­cially con­sid­er­ing most res­i­dents would have been do­ing less in the way of EA-fo­cused work with­out the Ho­tel. How­ever, it’s worth con­sid­er­ing the pos­si­bil­ity that res­i­dents may have done things of equal or higher value oth­er­wise. In the fu­ture, we will go fur­ther and ask res­i­dents to es­ti­mate the coun­ter­fac­tual im­pact of the work they have done at the Ho­tel rel­a­tive to the work they ex­pect they would have done oth­er­wise. We will touch on this in the next post in this se­ries, where we take a look at res­i­dent case stud­ies.

An up-to-date, live ver­sion of the list of out­puts can be found at ea­ho­tel.org/​​out­puts.

To­tal ex­penses as of Oc­to­ber 2019

Money: So far ~£110,400* has been spent on host­ing our res­i­dents, of which ~£17,900 was con­tributed by res­i­dents.

Time: ~7,600 per­son-days spent at the Ho­tel.

Sum­mary of con­crete out­puts, since the Ho­tel’s in­cep­tion in May 2018

  • The in­cu­ba­tion of 3 EA pro­jects with po­ten­tial for scal­ing.

  • 18 on­line course mod­ules followed

  • 2.5 on­line course mod­ules produced

  • 46 posts on LessWrong and the EA Fo­rum (with a to­tal of ~1500 karma)

  • 2 pa­pers published

  • 3 AI Safety events, 1 ra­tio­nal­ity work­shop, 2 EA re­treats hosted; and 2 EA re­treats organised

  • 2 in­tern­ships and 2 jobs earned at EA organisations

For the full list see the ded­i­cated Ho­tel page here.

New out­puts as of Oc­to­ber 2019

Here we pre­sent new out­puts since our last sum­mary; those from April-Oc­to­ber 2019.

Ex­penses for this pe­riod:

Money: ~£43,900* has been spent on host­ing our res­i­dents, of which ~£10,300 was con­tributed by res­i­dents.

Time: ~3,600 per­son-days spent at the Ho­tel.

Out­puts Key

Ti­tle with link [C] (K)
C = coun­ter­fac­tual like­li­hood of hap­pen­ing with­out the EA Ho­tel.
K = Karma on EA Fo­rum, Less Wrong, (Less Wrong; Align­ment Fo­rum) [cor­rect as of 8th Nov 2019].

Davide Zagami:

Coau­thored the pa­per Cat­e­go­riz­ing Wire­head­ing in Par­tially Embed­ded Agents, and pre­sented a poster at the AI Safety Work­shop in IJCAI 2019 [15%]

Linda Linse­fors:

Or­ga­nized the AI Safety Learn­ing By Do­ing Work­shop (Au­gust and Oc­to­ber 2019)

Or­ga­nized the AI Safety Tech­ni­cal Un­con­fer­ence (Au­gust 2019)


Dis­tance Func­tions are Hard [50%] (40; 14)

What are con­crete ex­am­ples of po­ten­tial “lock-in” in AI re­search? [1%] (17; 9)

Non-an­throp­i­cally, what makes us think hu­man-level in­tel­li­gence is pos­si­ble? (10)

The Mo­ral Cir­cle is not a Cir­cle [1%] (25)

Cog­ni­tive Dis­so­nance and Veg*nism (7)

What are we as­sum­ing about util­ity func­tions? [1%] (17;8)

8 AIS ideas [1%] (39)

Lu­minita Bo­gatean:

Courses: Python Pro­gram­ming: A Con­cise In­tro­duc­tion [20%]

Sa­muel Knoche:

Code for Style Trans­fer, Deep Dream and Pix2Pix im­ple­men­ta­tion [5%]

Code for lightweight Python deep learn­ing library [5%]

Markus Salmela:

Joined the de­sign team for the up­com­ing AI Strat­egy role-play­ing game In­tel­li­gence Ris­ing and or­ganised a se­ries of events for test­ing the game [15%]

David Kristoffers­son:

Defined a re­cruit­ment plan for a re­searcher-writer role and pub­li­cized a job ad [90%]

Or­ga­niz­ing AI Strat­egy and X-Risk Un­con­fer­ence (AIXSU) [1%]

Denisa Pop:

Re­search­ing and de­vel­op­ing pre­sen­ta­tions and work­shops in Ra­tional Com­pas­sion: see How we might save the world by be­com­ing su­per-dogs [0%]

Be­com­ing In­terim Com­mu­nity Man­ager at the Ho­tel and offer­ing res­i­dents coun­sel­ing/​coach­ing ses­sions (pro­duc­tivity & well-be­ing) [0%]

Matt Gold­en­berg:

Or­ga­nizer and in­struc­tor for the Athena Ra­tion­al­ity Work­shop (June 2019)

An­ders Huit­feldt:

Scien­tific Ar­ti­cle: Huit­feldt, A., Swan­son, S. A., Sten­srud, M. J., & Suzuki, E. (2019). Effect het­ero­gene­ity and vari­able se­lec­tion for stan­dard­iz­ing causal effects to a tar­get pop­u­la­tion. Euro­pean Jour­nal of Epi­demiol­ogy. https://​​doi.org/​​10.1007/​​s10654-019-00571-w

Post on EA Fo­rum: Effect het­ero­gene­ity and ex­ter­nal val­idity (6)

Post on LessWrong: Effect het­ero­gene­ity and ex­ter­nal val­idity in medicine (43)

Kris Gu­lati:

Distinc­tion in MU123 and MST124 (Math­e­mat­ics Mo­d­ules) via the Open Univer­sity.

Com­pleted ‘Jus­tice’ (Har­vard MOOC; Ver­ified Cer­tifi­cate).

Com­pleted Units 1 (In­tro­duc­tion) 2 (Math­e­mat­i­cal Type­set­ting), MST125 (Pure Maths mod­ule), The Open Univer­sity.

Com­pleted Unit 1, M140 (Statis­tics), The Open University

Com­pleted Week 1, GV100 (In­tro to Poli­ti­cal The­ory), Lon­don School of Eco­nomics [Au­dit­ing mod­ule].

Max Carpen­dale:

Posts on the EA Fo­rum:

In­ter­view with Jon Mal­latt about in­ver­te­brate con­scious­ness [50%] (81; win­ner of 1st place EA Fo­rum Prize for Apr 2019)

My recom­men­da­tions for RSI treat­ment [25%] (60)

Thoughts on the welfare of farmed in­sects [50%] (31)

In­ter­view with Shel­ley Adamo about in­ver­te­brate con­scious­ness [50%] (37)

My recom­men­da­tions for grat­i­tude ex­er­cises [50%] (39)

In­ter­view with Michael Tye about in­ver­te­brate con­scious­ness [50%] (32)

Got a re­search po­si­tion (part-time) at An­i­mal Ethics [25%]

Nix Goldowsky-Dill:

EA Fo­rum Com­ment Prize ($50), July 2019, for “com­ments on the im­pact of cor­po­rate cage-free cam­paigns” (11)

Rhys Southan:

Edited and par­tially re-wrote a book on meat, treat­ment of farmed an­i­mals, and al­ter­na­tives to fac­tory farm­ing—earn­ing enough money to start pay­ing rent at the Ho­tel. The name of the book or its au­thor can’t be named yet, as a non-dis­clo­sure agree­ment was signed, but it will be ver­ifi­able. [70%]

Wrote an aca­demic philos­o­phy es­say about a prob­lem for David Be­natar’s pes­simism about life and death, and sub­mit­ted it to an aca­demic jour­nal. It is cur­rently await­ing scores from re­view­ers. [10%]

Re­cently se­cured a paid job writ­ing an in­dex for a book about death and dy­ing by moral philoso­pher Frances Kamm, and will use the money to con­tinue pay­ing rent to the ho­tel. [20%]

Events made pos­si­ble by the EA Hotel

(Note that these already ap­pear in the above list un­der the main or­ga­nizer/​lec­turer’s name)

Athena Ra­tion­al­ity Work­shop (June 2019) (ret­ro­spec­tive)

AI Safety Learn­ing By Do­ing Work­shop (Au­gust and Oc­to­ber 2019)

AI Safety Tech­ni­cal Un­con­fer­ence (Au­gust 2019) (ret­ro­spec­tive writ­ten by a par­ti­ci­pant)

Our Ask

Do you like this ini­ti­a­tive, and want to see it con­tinue for longer, on a more sta­ble foot­ing? Do you want to cheaply buy time spent work­ing full-time on work re­lat­ing to EA, whilst si­mul­ta­neously fa­cil­i­tat­ing a thriv­ing EA com­mu­nity hub? Do you want to see more work in the same vein as the above? Then we would like to ask for your sup­port.

We are very low on run­way. Our cur­rent short­fall is ~£5k/​month from Jan­uary on­ward (thanks to the gen­er­ous donors who donated in the last week!). See the Ho­tel’s fundraiser page for more de­tails.

To donate, please or use our PayPal MoneyPool (0% fees, ~2% cur­rency con­ver­sion losses) or GoFundMe (2.9% fees). If you’d like to give reg­u­lar sup­port, we also have a Pa­treon (10%+ fees/​losses for non-USD donors!). Con­tact us to donate di­rectly via bank trans­fer and save in fees/​cur­rency con­ver­sion (Revolut and Trans­fer­wise are good op­tions with 0-1% losses in fees and cur­rency ex­change).

For an added bonus equiv­a­lent to tax-de­ductibil­ity, we are on Effec­tive Altru­ist Dona­tion Swap (in the Meta sec­tion). It should be pos­si­ble to get your dona­tion swapped with some­one from a coun­try that doesn’t offer tax de­ductibil­ity to any char­ity.

Pre­vi­ous posts in this se­ries:

See also the pitch doc­u­ment we’ve sent to po­ten­tial fun­ders.

Writ­ten by oth­ers:

*this is the to­tal cost of the pro­ject to date (31st Oc­to­ber 2019), not in­clud­ing the pur­chase of the build­ing (£132,276.95 in­clud­ing build­ing sur­vey and con­veyanc­ing)