Love to see this! I read Marina’s book senior year of high school and it was hauntingly beautiful. (Her essays were collected posthumous by family). Worth checking out!
barkbellowroar
Karma: 305
Hi David,
I’m intrigued by what you are trying to do with this LLM project. Prompt-engineering is very much a trial and error process—here’s what I got from walking through it a few times.
To clarify my understanding of your objective here is:
Given the 80,000 hours podcast episodes
Identify the episodes discussing quantitative social sciences, economics and policy
Within those episodes identify questions brought up in conversation
Filter those questions by if they are specific and measurable (operationalizable)
From there, identify the questions that would impact a donor’s decision-making in deciding whether or not to fund further research
(gleaning from your examples) add a direct quote and link to relevant literature
Does that sound right? If so here, here’s what I got (a few different threads) (top to bottom)
1st attempt:
Make a research agenda for one specific episode
identify donor relevant funding questions
2nd attempt:
first identify all the relevant podcasts (but 10 seems to be the limit for any kind of listing)
I narrowed it to just one (quant social).
I then tried using 80k podcast filter to help narrow it but perplexity says it can’t read the filterable listings of the transcripts.
3rd attempt:
I asked it to pull all episodes related to economics (caps it at 10)
Pull questions from each episode (only does it for 3 episodes)
Identify questions that are measurable and influence donor decision-making
Asked it to add quotes but it lost the thread I guess and didn’t connect to the above list of questions.
So I tried narrowing it down to one episode and filter through again adding a quote and a link to a piece of related research.
Curious to know if these results are along the lines of what you are trying to achieve?
My guess for systematizing this would be defining a prompt flow that gets the results you want and then replicating it across each individual podcast episode. You could use a separate thread to get it to create a list of episodes relevant to topics to narrow the amount you have to search through.
Long-term it would be cool to see 80k have something like tim ferriss’ blogbot given how much content they have (and the information that is buried in each podcast episode).
Two general notes on UXD:
1) It could be worth conducting user tests of whether people find the site’s landing page being the forum feed overwhelming. It’s hard to get your bearings on that page versus say the “best of the forum” page. People typically like to be guided initially in an experience and get a feel for what’s available, then explore. Or even just a pop-up with “learn about the forum” (it takes a minute to find the link for that page on the sidebar and these days people bounce within seconds).
2) In the spirit of the intranet comment above, I’d love to see the CRM hidden behind a logged-in view; this could just be my “safety” lens but having a giant list of confirmed EAs in public view seems problematic (sadly, something to consider these days). Maybe ask community health team for their view on it? but when it rolled out I was a little irked to see it created without first asking users if they want to be placed on such a definitive, public listing.
Thanks for writing this Sarah; nice to have leaders sharing thinking/requesting feedback!
My quick ‘off-the-cuff’ thoughts in response:
I definitely agree the Forum is critical as a “space” in EA, particularly for those of us who aren’t living in a hub, have access to conferences, nor work directly in an organization. It can often feel at times like being a buoy at sea without that direct connection—the handful of times I’ve drifted from EA over the years it has been the Forum pulling me back in (by helping me get up to speed on what’s going on especially in cases like the FTX stuff … and contests are often one of the few ways I feel I can participate—i.e. open phil’s worldview prize, red teaming, future fund RFP).
I have at times seen established EAs “punch-down” on the “extremely online” EAs and I’ve felt that to be a shallow judgement of those of us not privy to direct engagement opportunities. It definitely gave me the impression the movement and community are two separate entities not always aligned. I’d like to think the Forum could better blur that distinction to avoid hubs becoming silos with strong views (which I think contributes to the confusion around the public perception of EA).
That aside, the Forum right now is confusing because it’s providing multiple services in one; a newspaper, an opportunities/classifieds board, a library (the wiki) and a discussion space though not as free flowing as slack/discord. Now with groups and CRM and event tracking it’s becoming a catch-all for EA information. Given this…
Has the forum team considered reframing the Forum as an intranet? (Note, the intranet model is often associated with corporations but it’s used in a lot of contexts for brand community sites, colleges and social groups).
I’m thinking one of the limitations of the Forum is that the name implies one feature, but now the Forum is so much more than that—if we were to shift away from centralizing “the Forum” aspect and make it the “Hub” or “EA Online”… whatever it’s called, the idea is to step back and reframe the online space as more than just a “forum” because that’s what it’s becoming, as evidenced by CEA taking on the EA Hub and the Opps Board recently, it’s clear the goal is for CEA to manage the movement’s digital infrastructure.
I see the intranet design helping resolve issues like low engagement. If people can accomplish multiple things like search/apply for jobs, join topic threads, and chat with each other, complete surveys and more, they may be more likely to login and engage. This can help on the backend with things like surveys, community health team work (ticket system), and maybe down the road a common app or work trials. It also helps draw a boundary around community conversations that don’t need to be as public (which I think is a huge deterrent for engagement right now).
I’d be happy to discuss more in-depth and share some examples of what I’m visualizing here but I’ve long thought the Forum was heading in the “intranet” direction and could provide a lot of benefit for both members of the movement and those running infrastructure for it.
Hi David, I’m working on a project that might be relevant to this—sending you a DM.
Yes it was!
EA in the wild: I’m having trouble adding a screenshot but I recently made an online purchase and at the bottom of the checkout page was a “give 1% of your purchase to a high-impact cause”—and it was featuring Giving What We Can’s funds!
Always fun to see EA in unexpected places. :)
In case you haven’t seen it, CEA has redone their website. I like the new look and the content makes it much easier to understand the scope of their work. Bravo to whomever worked on this!
EDIT: just confirmed that FHI shut down as of April 16, 2024
It sounds like the Future of Humanity Institute may be permanently shut down.
Background: FHI went on a hiring freeze/pause back in 2021 with the majority of staff leaving (many left with the spin-off of the Centre for the Governance of AI) and moved to other EA organizations. Since then there has been no public communication regarding its future return, until now…
The Director, Nick Bostrom, updated the bio section on his website with the following commentary [bolding mine]:
″...Those were heady years. FHI was a unique place—extremely intellectually alive and creative—and remarkable progress was made. FHI was also quite fertile, spawning a number of academic offshoots, nonprofits, and foundations. It helped incubate the AI safety research field, the existential risk and rationalist communities, and the effective altruism movement. Ideas and concepts born within this small research center have since spread far and wide, and many of its alumni have gone on to important positions in other institutions.
Today, there is a much broader base of support for the kind of work that FHI was set up to enable, and it has basically served out its purpose. (The local faculty administrative bureaucracy has also become increasingly stifling.) I think those who were there during its heyday will remember it fondly. I feel privileged to have been a part of it and to have worked with the many remarkable individuals who flocked around it.”
This language suggests that FHI has officially closed. Can anyone at Trajan/Oxford confirm?
Also curious if there is any project in place to conduct a post mortem on the impact FHI has had on the many different fields and movements? I think it’s important to ensure that FHI is remembered as a significant nexus point for many influential ideas and people who may impact the long term.
In other news, Bostrom’s new book “Deep Utopia” is available for pre-order (coming March 27th).
On a related note to my other comment on this post:
A lot of organizations are acknowledging the impact of FTX on their work which is important but I would also like to see an EA organization try to evaluate the positive or negative impact switching to longtermism has had on their ability to attract talent, donors etc.
You say you want to diversify assets but Open Phil still holds a commanding 80% of your bottom line—and both orgs have become much more longtermist in recent years. If OP is going to just prop up 80k for the next several years because it serves their aims, why would I fund 80k when there is now a considerable gap in the landscape for a new career service org that caters to the other cause priorities of EA?
Equally curious about the push to grow the team if not seeing significant increase in impact, especially given the $2M marketing push this past year.
In 80K’s 2021-2022 Review it mentioned:
(1) “we seem to be hitting diminishing returns in outreach encouraging more people to apply to advising...” (page 7 under current challenges)
and again
(2) “overall, we’d guess that 80,000 Hours continued to see diminishing returns to its impact per staff member per year.” (on page 10 under impact evaluation)
What is the strategy/argument for “expand the team” being the best intervention for increasing organizational outreach and subsequent impact? Is it really just a capacity issue or could it be a scope issue?
The Effective Ventures Foundation UK’s Full Accounts for Fiscal Year 2022 has been released via the UK companies house filings (August 30 2023 entry—it won’t let me direct link the PDF).
Important to note that as of June 2022 “EV UK is no longer the sole member of EV US and now operate as separate organizations but coordinate per an affiliation agreement (p11).”
It’s noted that Open Philanthropy was, for the 2021/2022 fiscal year, the primary funder for the organization (p8).
EVF (UK&US) had consolidated income of just over £138 million (as of June 2022). That’s a ~£95 million increase from 2021.
Consolidated expenses for 2022 were ~ £79 million—an increase of £56 million from 2021 (still p8).
By end of fiscal year consolidated net funds were just over £87 million of which £45.7 million were unrestricted.
(p10) outlines EVF’s approach to risk management and mentions FTX collapse.
A lot of boiler plate in this document so you may want to skip ahead to page 26 for more specific breakdowns
EVF made grants totaling ~£50 million (to institutions and 826 individuals) an almost £42 million increase in one year (p27)
A list of grant breakdowns (p28) ; a lot of recognizable organizations listed from AMF to BERI and ACE
also a handful of orgs I do not recognize or vague groupings like “other EA organizations” for almost £3 million
Expenses details (p30) main programs are (1) Core Activities (2) 80,000 Hours (3) Forethought and (4) Grant-making
Expenses totaled £79 million for 2022 (a £65 million increase from 2021) which seems like a huge jump for just one year
further expense details are on (p31-33) and tentatively show a £23.3 million jump between 2021 and 2022 [but the table line items are NOT the same across 2021/2022 so it’s hard to tell—if anyone can break this down better please do in the comments]
We may now have a more accurate number of £1.6 million spent on marketing for What We Owe The Future (which isn’t the $10 million I’ve heard floating around, but still seems like a lot for one book) (p31)
staffing details mentioned (p35)
cryptocurrencies on (p37)
real estate (p39) <£15 million went to Wytham Abbey and <£1.5 million went to Lakeside (another relatively new EA property)
More details on Specific Organizations under EVFs purview begin (p43) specifically beginning fiscal year balances, income, expenses and end of fiscal year balance
The only thing that stood out to me on this page—Forethought overspent by ~£2 million?? was it the marketing?
(p46) FHI/GPI also overspent (on office costs) but Forethought doesn’t list their reasons only that they got a grant to cover it (from whom, Open Phil or a private donor?)
Restricted funds analysis on p47
p53 (last one!) notes DGB has earned <£100 on royalties for 2021/2022 and the Precipice has earned £37,293 for both years
p53 mentions EVF UK received donations of < £1.5 million from the FTX Foundation AND that the CEO of FTX was on the Board
p53 mentions the charity commission investigation, name change from CEA and separation of EVF UK from EVF US
* A reminder that both EVF UK and US annual reports can be found on their website
EVF US has also released their 2022 990 tax form
* Note the fiscal year 2021-2022 (June) preceded the FTX collapse and leadership changes.
If I messed up any of these numbers please let me know and I’ll update. Thanks!
Wytham Abbey soft-launched earlier this year with it’s own team, but has now formally been added to EV’s list of projects and is accepting workshop applications https://www.wythamabbey.org
Atlas Fellowship has announced it’s shutting down its program—see full letter on their site: https://www.atlasfellowship.org Reasons listed for the decision are 1) funding landscape has changed 2) the programs were less impactful than expected and 3) some staff think they’ll have more impact pursuing careers in AI safety.
barkbellowroar’s Quick takes
Sounds good, thanks for responding Nick!
I think that’s referring to something else (see the comment above his response). I completely understand the reasons for not releasing all submitted proposals to the public.
I’m asking for the final results of the round—simply, which projects were funded? This doesn’t seem unreasonable to ask, as most foundations usually do release some information about who and what they fund. Also FTX feels like it’s modeled a bit on YCombinator and they definitely release a list of startups funded each cycle. It’s good for transparency, but even more so for helping future applicants get a sense of what a particular funding body wants to invest in.
[Question] Will FTX Fund publish results from their first round?
(short answer) more security, more features and the consolidation of a lot of existing but disconnected infrastructure tools… which could strengthen movement coordination, increase collaboration and calibration and sustain longterm engagement with the community.
Just like you can’t catch rain with a sieve, you can miss a lot of value with a fragmented ecosystem.
(longer answer)
An intranet would subsume under one platform a lot of current tools like… event sign-ons, the forum, EA hub’s directory, facebook groups, job/internship boards, the Wiki, various communication channels (twitter, discords, slacks, email etc), surveys and polls, chapter sites, separate application forms, the librarian project and organization newsletters.
An intranet can also provide a greater array of features that do not currently exist in the ecosystem including (but not limited to) spaces for sub-group discussions, tiered engagement levels, guided on-boarding for new members, greater analytics and much more.
I think the biggest benefit of all is concentrating the online activity of the movement in one place versus the present state of having to check a disorganized collection of websites, blogs, sign-ons and social accounts in order to keep up with what is going on with the community. The majority of our time should be spent on our work and collaboration—not trying to track down important or relevant information, trying to figure out how to get involved and meet people in the movement, and figuring out how to learn, grow and develop as an effective altruist.
Given the recent sunsetting of the EA Hub—and their comments that implied CEA may be attempting to develop a larger platform—this idea may be in progress. However, I still wanted to share and spark more discussion on the need for an intranet because I believe it would greatly improve movement coordination and strengthen the sense of community while significantly reducing the workload for meta organizations so they can invest more time and energy into their high impact programs.
Given the EA movement’s desire to grow more, and the inconceivable amounts of money currently floating around, it may make sense to invest in a pre-packaged intranet for now while also funding a team to begin building an in-house intranet platform that can be fully customized to the needs of the movement as it grows.
If you are interested in learning more about what a unifed platform for EA could look like here are some of the more popular intranets on the market: Sharepoint, Interact, GreenOrbit, Guru or Mangoapp (p.s. my favorites so far are Sharepoint and Interact).
[As someone personally interested in information architecture and digital taxonomy I started looking into this idea a while back and began drafting a proposal on how an EA intranet would operate and what benefits it could have for different roles within the movement. Let me know if you would be interested in reading a forum post on it—I have lots of articles in draft stage and it’s hard to prioritize which ones to work on, so an expression of interest in this particular piece would definitely push it to the top of my list!]
If anyone is interested here is a quick breakdown on differences in intranets, extranets and the internet and the value they provide.
It’s been a minute since I’ve been on the forum so my recollection of Karma voting is rusty. Does the agree-disagree function follow the same rules as strong voting? For example, the first question has 35 agree / 56 disagree—are those votes saying 56 individual users or could it be a few users who strongly disagreed with it? Just trying to figure out the actual number of people answering the poll. Thanks!