Comments
We want to help promising EAs maximise their value.
There are many smart and highly motivated people in the EA movement who lack visible achievements. As a movement, we are underusing their talents. By operating at minimal costs, the EA Hotel is able to make high EV/$ investments in large numbers of them as a form of hits-based giving, allowing them to move from software to AI Safety, move from part time to full time study, write original research, and many more possibilities.
Whilst the EA Hotel is open to people pursuing work across all EA cause areas, we note that approximately half of our residents have been focused on AI Safety/x-risk (~45% of person-days spent at the EA Hotel to date have come from those working in the AI Safety/x-risk cause area).
At the moment, it’s hard for talented and motivated EAs to leave stable employment and work on valuable independent projects. Three factors typically make this difficult:
In short, we are giving many EAs an opportunity to do important work, in an environment that catalyses it.
We’ve been able to attract an impressive set of people, from AI Safety researchers, to PhDs, to highly productive entrepreneurs, with an average of 4.8 years work experience (based on a survey earlier this year).
We expect most of the value of the EA Hotel will come from a handful of individuals whose career trajectory we can help change. That said, it looks like the counterfactual impact for the median resident is already high, with an average of 2.2x productivity increase reported in a survey earlier this year. A summary of outputs so far (see here for expanded list):
(Inputs: Money: ~£110.4k, of which ~£17.9k was contributed by residents
Time: ~7.6k person-days spent at the hotel.)
Based on self-reporting (see square brackets on outputs page), it’s estimated that over half of the value of these outputs can be counterfactually attributed to the EA Hotel. Other, less tangible value has also been reported by our residents. We follow with some case studies.
Before moving to Blackpool, Derek was studying for a Masters in Health Economics and Decision Modelling. Out of runway, he initially came to the EA Hotel to work on the 2019 Global Happiness and Well-Being Policy Report, which was published in January. He subsequently got a job at Rethink Priorities.
After some time away, he has returned to the Hotel. Still with Rethink, he is currently working on a research agenda for the creation of a wellbeing-adjusted life-year metric.
How the Hotel has helped
It gave him somewhere to stay when he had run out of money, increased his productivity and runway, and expanded his network of potential collaborators.
The counterfactual
Derek would have done similar work, but perhaps to a lower standard. He’s not sure how he would have paid his rent for the first couple months before getting his first paycheque, and would not have been able to save as much. He is also confident that he would not have attended EAG London 2018, where he gained a number of valuable contacts in addition to those he met at the Hotel.
Linda is doing AI Safety research. After interning at MIRI, she went back to her home town in Sweden and found herself largely unmotivated to continue her research alone. She went to the EA Hotel to continue her research and studies among like minded people.
During her stay here Linda also explored other career paths. She did some web programming for Matt Goldenberg and co-organised Athena Rationality Workshop. After rediscovering the joy of organising, Linda has started to organize AI Safety events. So far she has done two workshops and one unconference, which have all been appreciated, and she will most likely do more such events in the future.
How the Hotel has helped
Life at the Hotel has made Linda much more motivated and allowed her to put considerably more focused hours into studying. She also managed to find some collaboration, which gave birth to two posts on the AI Alignment Forum.
Living in the same house as most of the RAISE team allowed Linda to give high quality, high bandwidth feedback on the AI Safety course they were developing.
Linda has received a great amount of mental debugging help and emotional support from other Hotel residents. Because of this, her mental health has improved a lot which has increased her productivity and reliability. Without this upgrade she would not have had the confidence that she would be reliable enough to organise events.
Being able to use the Hotel as a venue for events has made Linda able to do more events for less effort. The fact that EA Hotel already have regular catering service (Jacob cooks for us) saves lots of work.
The counterfactual
She would have continued her life in Sweden, doing research and studying at a rate of approx 30% of the current.
None of the events would have happened.
Max is a researcher working on invertebrate sentience and suffering who was living in Vancouver. He has recently worked with Rethink Priorities on this subject and has been successful in getting funding from the EA community for his research. Max also started working part-time for Animal Ethics as of September 2019. Max does research on wild animal suffering for Animal Ethics and he is currently working on creating an online course on wild animal suffering.
How the Hotel has helped
It extended his runway considerably. Max suspects that he would have ran out of money to pay rent if not for the Hotel, and really isn't sure what he would've done in that case. Max thinks that there is a 75% chance he wouldn't have got his current job at Animal Ethics if it wasn't for the Hotel. Max does not think that he has been more productive at work while at the Hotel, but he thinks that the Hotel has been very helpful for personal growth and his lifetime impact as an EA.
The counterfactual
Being at the Hotel has given him the opportunity to travel much more easily than he would be able to in Canada. He has gone to several EA conferences and retreats that he would not otherwise have been able to go to. It gave him the opportunity to live abroad in Europe, which he thinks has benefited him greatly. He expects that his network has expanded and will continue to expand due to the Hotel.
More case studies can be found in Appendix A.
Quitting your job to work on a new project or study a new field presents a financial risk - or for some a financial impossibility. We enable those with the potential to do great EA work to do it regardless of circumstance.
Staying at the Hotel is incredibly cheap as we have bought the building outright. Hospitality costs for a resident are ~£80/week (including mentoring). As the cost of living in Blackpool is low, we can cover the remaining costs with a stipend for grantees of £30/week.
We offer grants of free or subsidised accommodation and board, and a stipend, to those applying to do EA work, subject to passing through our vetting procedure (see Grant Making Policy).
Up until now, our default has been providing free stays to anyone doing EA work that we offer a place to, and asking those with salaries or significant savings (>24 months runway) to volunteer payment of cost price for their stay. Given our application for charitable status, we have revised our model: applicants request the level of subsidy based on their needs.
Ultimately, we’d like to sustain the Hotel primarily via pledges from successful alumni - it’s cheap enough that it could plausibly be funded by just a handful of successful benefactors (so far we have already had one such alumnus who has made significantly more in donations to the Hotel than his stay cost). Until we reach that point, we’ll experiment with various (honesty-based) means-tested approaches, balancing the need for at least partial sustainability with administrative efficiency and more importantly a desire not to deter great candidates. For example, we’re currently taking the following approach:
We may further update this model as we learn more about how it plays out in practice.
This approach will help cover some of our costs, but will necessarily still leave a shortfall as we are especially interested in funding work that wouldn’t happen otherwise (i.e. work by those not in a position to pay for their stay).
In our current situation, factoring in the above, we need ~£5,000/month in additional funding to sustain the Hotel. Note that this is an approximate figure given our costs are variable dependent on different levels of guest contribution, occupancy, staffing and maintenance costs. Our current runway extends until the beginning of March 2020 (see our Fundraiser page for up to date figures).
What various levels of funding would provide
£15,000 (3 months funding) would allow us to run a hiring round for a Community & Projects Manager, which could in turn bring in money from EA Grants/CEA, who are interested in providing funding for this role conditional on us finding an excellent candidate for it.
£60,000 (12 months funding) would allow us to provide 8 slots for free stays filled with great people for the next 12 months, and 18 6-weeks’ free stay grace periods for new applicants.
£100,000 would give us an industry-standard runway, and flexibility to provide more free stays to great people over the next 18 months.
Low cost EA hubs have been talked about a lot but have rarely materialised, due to high activation costs and coordination difficulties. We have built the community, shown preliminary results in our first 18 months and now need outside funds to continue providing cheap/free stays to great people, and bridge the gap to long-term sustainability.
Our current funding situation is precarious, without further funds we’ll need to start winding down and giving residents notice in under 3 months. Under reasonable assumptions, we could reach sustainability within 3-5 years, once we have sufficient alumni pledging support. But we need your help to bridge the gap and make this happen. If you like this experiment and want it to continue, please donate here.
Previous posts in this series:
Written by others:
We discuss additional case studies, assessing the impact of the Hotel on their work from their own perspective, and then link to additional bios for past guests.
Fredi is currently writing his master’s thesis on the Demandingness Objection to ethical theories and he did an internship at Animal Ethics, working on wild animal welfare assessment methods.
How the Hotel has helped
He heard about the internship position through one of his many new acquaintances he made at the Hotel. The Hotel enabled him financially to do the unpaid internship.
The atmosphere of the Hotel keeps him motivated and increases his productivity.
The counterfactual
Fredi would still write his master’s thesis, but would’ve not been able to start the internship. He would work part time as a coach to support himself, like he did for the last ten years. Maybe, he would have drifted out of the EA community.
Kris has an undergraduate degree in Politics and a two-year MSc in Econometrics and Economics from The University of Nottingham. He is currently applying to PhD programmes in Economics, whilst learning the Mathematics needed for graduate school. His primary research interests are in Growth Theory and Empirics with a wide range of applications including: developing countries, political economy, and the repercussions of technological innovation.
How the Hotel has helped
The Hotel has helped substantially. Without the Hotel he would have been forced to join a PhD programme at Nottingham or Manchester (where he had applied to and been accepted). The Hotel has given him the opportunity to apply to the top PhD programmes in the UK (Oxford, Cambridge, LSE etc.). The Hotel has also allowed him to continue developing the mathematical/statistical knowledge needed for a top tier PhD programme. It has also refined his research interests, with a greater focus on EA-related causes.
The counterfactual
Kris says: “Although the counterfactual will not be completely known until I hear back from the PhD programmes, I still believe the counterfactual impact of the Hotel on me is very large. I would have not been able to take a year out (for financial reasons) to apply to PhD programmes without the Hotel. Firstly, academia is quite elitist and so if I get into a better school, the potential rewards (both in terms of research quality and salary which would be donated) are very large. Secondly, the knowledge I’ve gained will translate to a better research output, irrespective of where I ultimately end up as a researcher. I now aim to focus 100% of my research on EA-related causes. If I had not stayed at the Hotel, I may have dedicated some of my research capacity to less important causes. Thirdly, I strongly feel that my commitment to EA has dramatically improved. Value drift seems to be a concern amongst EAs. However, I truly believe interacting with the other EAs in the Hotel and the network I’ve gained has made me into a lifelong EA. Prior to coming to the Hotel, I felt slightly alienated with the views I had held. However, this has changed dramatically. If these effects were to be quantified, I believe they are by far in excess of the costs.”
Samuel is studying Machine Learning. Since he arrived at the Hotel, he has worked through the Introduction to Statistical Learning textbook, and made great strides in his study of deep learning. He has replicated many papers (DQN, World Models, Deep Dream, Style Transfer, Pix2Pix…) and hopes to get an earning to give position in ML research, and ultimately contribute to AI safety.
How the Hotel has helped
For an autodidact, the Hotel is ideal. It allows one to have full control over one’s learning progress, while also providing the social support to stave off depression, and stay productive. Samuel therefore also thinks that his experience successfully self-teaching himself deep learning will give him more credibility in calling for education reforms. He thinks that universities and startups in the education space have a lot to learn from the EA Hotel model.
The counterfactual
Samuel would have either taken a junior software engineering position, started some boot-camp/coding school or tried self-teaching at a significantly slower pace while living with his parents. In any way, his long term career plans would have probably taken at least 2 years longer, and his commitment to effective altruism would have been significantly lower than it is now.
Chris originally attended the EA Hotel in order to work on the problem of Infinite Ethics. After finding the problem significantly harder than expected, he decided to split his time between trying to make headway on this problem and learning more about AI Safety. Chris used his time at the Hotel to read Superintelligence, the RAISE coursework material and the Alignment Forum Sequences. He also developed the forgetting approach to Logical Counterfactuals at the Hotel. He believes that this is the most promising area of AI safety for him to work on and that he has already made significant progress, most notably in his Alignment Forum post Deconfusing Logical Counterfactuals. He also found out about his current position at Triplebyte from another person at the EA Hotel. This job is allowing him to support himself while working relatively few hours so that he can pursue research on the side and to work remotely so that he can attend programs like AI Safety Camp.
How the Hotel has helped
The Hotel allowed him to find out about the opportunity to work at Triplebyte. It freed up his time without which his investigations into Logical Counterfactuals might not have proceeded very far, instead ending up on a long list of things to pursue some day and perhaps even forgotten.
The counterfactual
Chris probably would have taken a normal software developer position and not have anywhere near as much time to pursue research on the side or been able to join AI Safety Camp. He would likely still be in Australia and have much less opportunity to learn from others who are interested in AI Safety.
Denisa has a PhD in Human-Animal Interaction and was planning to pursue a career in academia. The Hotel allowed her to focus her research entirely on high impact topics. She has since developed and delivered workshops on her psychological framework, “Rational Compassion”, as a tool for aspiring EAs. Separately, she is also able to bring her 5 years' experience as a counseling psychologist to the Hotel to offer mental health sessions. She has also helped EA Netherlands with organising conferences and retreats and currently works as an Interim Community Manager at the Hotel.
How the Hotel has helped
She feels the Hotel provided her a supportive, encouraging and motivating environment, with people who share similar values. She thinks that living, working and doing activities together with this type of people amplifies the benefits of being part of an EA community.
The counterfactual
Denisa would have probably been enrolled in a post-doctorate programme (most likely on a non-EA topic) or doing mental health counseling for non-EAs (she is also a licensed psychologist). By not being able to completely put her values into action and being surrounded by people who don't share her life views, she believes that her productivity and life satisfaction would have been decreased.
Markus is a developer for the upcoming AI Strategy role-playing game Intelligence Rising. He also works on understanding the effects of technologically extended human lifespans from an existential risk perspective. He previously finished a research collaboration on x-risk forecasting while at the Hotel. His background is mostly in Philosophy and Social Sciences.
How the Hotel has helped
The Hotel provided essential financial support, and the relatively easy access to London made it much more likely that Markus could use his contacts to join the development of Intelligence Rising. Other Hotel residents were often helpful for his projects, giving good suggestions and being generally supportive, which led to noticeably greater motivation and productivity.
The counterfactual
In the 3-4 years before coming to the Hotel, Markus did occasional minor contributions to EA projects, equivalent to about two weeks full time work annually, while not making any donations. Without the Hotel he would likely have continued at this rate, unless he received about the same amount of funding from elsewhere to pay for living expenses.
Davide initially joined the RAISE team and developed lessons on Inverse Reinforcement Learning, Computability, Logic and Set Theory. After the project got discontinued, he attended AISC for ten days and then co-authored a workshop paper about wireheading while at the Hotel. Now he is allocating part of his time to thinking and reading more about technical AI Alignment, and part of his time to self-studying Data Science and Machine Learning.
How the Hotel has helped
Davide thinks that the EA Hotel provides him time and space to think about AI Alignment and to educate himself about Machine Learning, as well as other topics. His general productivity improved considerably, as well as his ability to bounce off ideas with other people in real time.
The counterfactual
Originally, Davide would have accepted a job offer in Italy as a software engineer for a non EA cause (ML is also not very popular in Italy right now). He thinks this may have caused him to gradually fall out of touch with EA ideas, or at least to delay his EA work by about one year. He may have studied AI Safety as an autodidact from time to time, and at a much reduced pace, as he did for about six months before joining the Hotel (there is no EA/AI community in Italy, whereas the Hotel enabled him to attend events such as AISC, HAAISC, IJCAI, and TAISU). If he had to leave now, he would spend much less time thinking about AI. He may be able to find a job as a Data Scientist in London, if he is lucky, or he could go back to something similar to that job opportunity in Italy, more likely.
Anon used to be a musician, encouraged to pursue a career in the arts, even though they privately thought this was a waste of time and were more interested in EA. They came to the Hotel to retrain. They are now self-studying full time, aiming to become a researcher or operations specialist.
How the Hotel has helped
The Hotel has given them the freedom to leave home and begin to put their values into action. They can study what they want, without having to observe a curriculum inflated with useless trivia. They can talk to and learn from plenty of intelligent people who they would never have approached had they not been in their immediate environment. They are a lot happier at the Hotel than they would’ve been otherwise.
The counterfactual
They would still be self-studying, but at a much slower rate, whilst working full-time outside of the EA sphere. It would take considerably longer for them to make progress under these alternate circumstances, and they doubt they would have been able to sustain the motivation to persevere whilst staving off depression.
Anon recently completed an MS in mathematics, with their thesis research focusing on mathematical logic. Having finished, they moved to the Hotel 3 months ago to focus on AI safety research. They've been trying to find new angles of attack on the problem(s), and drafted up a few ideas along the way. Their posts on the Alignment Forum have been relatively well-received, not to mention the 8 other ideas they dropped on the Open Thread, a few of which they are currently drafting up.
How the Hotel has helped
The Hotel provides an environment where Anon can do research without having to worry about finances or even preparing food. They can also be more clued-in to EA and bounce ideas off people. In addition, there have already been 3 AI safety-related workshops in Anon's time here (organized by Linda) which have been helpful for Anon in various ways: giving talks, airing ideas, having conversations with people, hearing other ideas, and, in the case of the Learn-By-Doing workshops, getting an excuse to spend an afternoon rethinking the Alignment Problem from the ground up. Also, Anon could attend EAG London thanks to being at the Hotel, which they found valuable.
The counterfactual
Anon isn’t sure but thinks they would have continued on in grad school, either in math or theoretical CS, and continued doing research they found interesting but ultimately irrelevant to EA.
*these residents preferred not to be identified by name
RAISE (Road to AI Safety Excellence) was a charity that aimed to lower the cost of studying and teaching AI Safety. They were a team of 1.5 FTEs, in different countries, until they moved to the EA Hotel together and became a team of 4.3 FTEs, greatly increasing their production rate.
How the Hotel helped
They had some funding, but only enough to support one of the two people working on the project. The founder of the project was running out of runway. Part of the reason was that, despite their best efforts, production speed was not on a satisfactory level.
The Hotel allowed RAISE to add 3 more people to their production team, increasing their production speed fourfold. It helped the project survive - when funding dried up entirely in November 2018, the project was able to keep afloat at nearly full capacity thanks to the Hotel. 3 weeks later, they were given an experimental grant, covering all of their expenses.
The project is now defunct, after consulting with the EA Long Term Future Fund.
The counterfactual
A startup is an experiment. The goal is not to succeed, but to test a business model in combination with a few initial conditions (founders, their capital, etc). In this particular case, the lesson was quite valuable for the people involved. Without the extra 8 months that the Hotel appended to the existence of RAISE, the experiment would not have been carried out as thoroughly, and the takeaways would have been as meager as “they didn’t want to fund it but we don’t know why”.
Hoagy is currently working on an international project trying to advance interpretability in reinforcement learners in his spare time while working for RAISE during the day. Hoagy has a degree in Economics and Philosophy from Oxford University.
How the Hotel has helped
Before coming to the Hotel he had a growing interest in AI safety but had no friends in the area and lacked some skills to contribute, especially in programming. He now has a fast-growing network in the field and the ability to contribute to projects in a variety of ways. He is now trying to work on AI safety full time. He currently has an offer from the Civil Service Fast Stream as an economist and plans to use this as a jumping point to enter an AI policy role in the UK government if he cannot find a role as a researcher.
[Update (October 2019): Was a part of MIRI summer fellows which he would have (90% confidence) been able to attend without having gone to the Hotel. He is now working in the Home Office and actively pursuing opportunities to move into AI policy in government.]
The counterfactual
Hoagy would have continued self-study in mathematics and searching for opportunities both in and out of AI safety. He would most likely have begun working at a non-EA corporate role in London (60%) or found something more exciting (40%).
Edward is a researcher and aspiring effective altruist currently working on researching objections to effective altruism, particularly from the left-wing political sphere. Parallel to this, Edward has, on the advice of others at the Hotel, applied for further education in International Relations, and is now beginning to apply for jobs that would increase his career capital.
How the Hotel has helped
The Hotel has helped in two primary ways. The first was to give the funding necessary for Edward to research and begin writing his article(s) on the topic of left-wing criticisms of effective altruism. The second, and more significant one, is the way in which the community of the Hotel has encouraged Edward to shift his focus toward the future. In particular, he has been inspired to pursue a career mitigating existential risk, and taken steps toward achieving this.
The counterfactual
Primarily, Edward would likely not have identified as part of the effective altruist community, and would have drifted away from the movement completely. More concretely, he would not have been inspired to pursue an EA-aligned career, or taken steps to achieve this by applying for relevant further education and jobs, instead likely relegating this to a vague aspiration. Finally, he would not have had the time to pursue research which he had long been interested in, and sincerely hopes will be useful to the EA community for both outreach and intra-EA conversations when published.
Guests have fortnightly check-ins to discuss their work with the Community & Projects Manager, and fill out a monthly survey as part of a regular group meeting. We think it’s tricky to gauge the outcome of projects over the relatively short time period as has so far elapsed with the EA Hotel, especially considering that most of our guests are at an early stage in their career, but are tracking stories (ongoing counterfactual value self-reports are a requirement of longer stays at the Hotel) and collecting outputs. In the future we intend to look at metrics such as impact-adjusted significant plan changes (following 80,000 Hours) and counterfactual additional money donated to EA charities.
We are making an in-depth attempt to estimate the (relative) EV of the Hotel.
Grants operate at a different level of granularity - it takes a certain amount of time to apply for and process them, and this will inevitably be longer for higher value grants. Since grants range in value from the equivalent of several months to several years of a one-person stay at the Hotel, they’ll inevitably be sought by people who need more resources and are typically further down the path. In other words, far fewer people will plausibly receive them than will stay at the Hotel.
One might consider EA Grants/EA Funds as EA equivalents to venture capital funds, whereas the Hotel is more like an incubator. Just as in the for-profit world, we think both roles are distinct but complementary in getting the most from the EA ecology.
That said, some residents probably will end up paying with an EA grant, since grant and Hotel aren’t exclusive - and living in a low-cost hub would act as a multiplier on their grant’s value.
Brand association: other EA organisations have found that the ideas of effective altruism are hard to explain in low-fidelity media. As long as we don’t have considerable PR expertise on hand, we will reject all media requests.
Interpersonal conflict: given many people in close proximity, incidents of interpersonal conflict will sometimes arise. We take the risk posed by such conflicts seriously, and have systems in place to both prevent conflicts emerging in the first place, and to mediate and prevent escalation if they do.
Supporting projects with negative value: This is mitigated by having a community on hand that is proactive in giving correcting feedback. We are also have some respected community members acting as advisors particularly on this issue.
EA Grants, EA Long-Term Fund and EA Meta Fund
Many of the pros and cons here are essentially two sides of a coin:
Comparative advantages of EA Grants/Funds:
Comparative advantages of the Hotel:
In addition, we see three non-analogous benefits to the Hotel:
Charity Entrepreneurship - projects run by Charity Entrepreneurship are well run and have been successful. However, Charity Entrepreneurship seem to have plenty of runway at the moment and (as far as we are aware) comparatively little room for more funding.
Initially, it will stay open and charge people cost price to stay, probably removing management and some amenities. This would remove many of the benefits of the Hotel, and it would be going against the main initial idea (of providing a kind of “UBI” for living costs for those doing useful EA work). The size and value of the community at the Hotel would very likely be substantially reduced. It would in effect just become a large EA group house.
If this proves financially unviable, the Hotel would be rented to a third party, or sold. A knock-on effect of the project ceasing due to lack of funding would be a diminished likelihood of similar projects being attempted, despite their grass-roots popularity among the EA movement.
"Max is a researcher working on invertebrate sentience and suffering who was living in Vancouver. He has recently worked with Rethink Priorities on this subject and has been successful in getting funding from the EA community for his research. Max also started working part-time for Animal Ethics as of September 2019. ... Max suspects that he would have ran out of money to pay rent if not for the Hotel"
Was Max working for these organisations the whole time and they just don't pay people enough to make rent? Or were these seperate gigs with gaps in between, and the gaps are what threatened his financial security?
I'd prefer charities to pay their staff a living wage rather than the Hotel subsidise living costs.
To be clear, Rethink Priorities pays good salaries that are more than enough to make rent. Those of our employees who live at the EA Hotel pay cost price and are not subsidized by the Hotel.
I came to the hotel as I was finishing a contract for Rethink Prioritites, worked for them there for one month, then did indepenent research. Now I am employed at an EA org again, and I am paying cost price.
Thanks Max for being willing to talk about your finances online. I know that's a sensitive topic.