You should join an EA organization with too many employees

By Ben_West🔸, JP Addison🔸, Clifford @ 2022-05-21T19:38 (+42)

I sometimes hear prospective employees feel like they are "overqualified" for EA positions by trying to compare relevant metrics. E.g. "you don't need a product manager if you have less than 5,000 page views per day." I think this is misunderstanding a key difference between altruistic and profit-motivated organizations.
Many companies only capture a small fraction of the value they produce. Nordhaus 2004 estimates that companies capture about 2% of the value they create through innovation.
This means that companies underinvest their products (relative to the social optimum), because they can't capture the value from improved products.
Therefore, altruistic organizations should invest substantially more into their products than comparable for-profit organizations do.
As a concrete example: Reddit has approximately one employee per 600,000 users, whereas this Forum has approximately one employee per 13,000.^[1] Interestingly, this is almost exactly the 50x multiple Nordhaus would predict.^[2]
- I would further argue that it's multiple orders of magnitude more valuable to attract a user to the EA Forum than to Reddit, so the 50x multiple should be even higher, though obviously I have a bias.
I think this adjustment for externalities is still not actually the right way to make career decisions – you should think more about the impact you have, not how some arbitrary metric compares – but if you are going to consider arbitrary metrics, I think you should consider adjusting by a large amount.

Some of these ideas are also referenced in the tech entrepreneurship 80 K article.

CEA Online, the team supporting this forum, has current openings for for UI, UX, or Graphic Designer, Product Manager, Full-Stack Engineer.

^{^}
Reddit data taken from https://backlinko.com/reddit-users. There are four people who work full-time on the Forum (JP, Sarah, Clifford, Lizka); without evidence I will claim that the portion of operational staff who support them add up to a 5th FTE. Forum MAU count from Google analytics, trailing 28 days ending May 17, 2022.
^{^}
Of course, this is oversimplified. For example, presumably there are diminishing returns to labor, so naïvely multiplying employees by 50 is too simplistic.

Linch @ 2022-05-22T15:06 (+31)

I think I very likely misunderstood your post, so some quick thoughts to check for my potential misunderstanding:

I haven't read the Nordhaus paper, but I assume innovation is a relatively small segment of most companies' activities.
Whereas this post is treating all of Reddit/this Forum's employees as contributing to innovation?
- No offense to Lizka, JP et.al, but I would not consider most of their work to be innovation
  - A major caveat here is that iirc economists' definition of innovation, technology, etc. is very different from the commonsensical definition, so I can totally imagine me being wrong here.
Certainly I would not guess that the typical company has total consumer surplus equivalent to that of 50x their total capital or labor investments.
- As a sanity check, I don't think most users would pay 50x their Netflix subscription fees to access Netflix (even if the counterfactual entails being cut from Netflix's competitors as well), and I think this generalizes to most companies.
  - (main exception I can think of is maybe smartphones, maybe internet overall)
I get the impression that you don't believe this argument too, and are using it as a way to persuade people who are used to working for a much larger audience, without getting into inside-view calculations of impact.
- If this is the case, I feel like a more persuasive argument would be to look at differences between consumer and internal or enterprise software, and argue that the value of meta-EA software work is closer to internal company software than consumer software

Charles He @ 2022-05-22T16:43 (+19)

Note: I literally have a PhD in this specific topic (but my alma mater is Devry Technical College of Alabama. This is different from the canonical EA schools of { Harvard, Yale, Oxford } , so I guess it cancels out the value of this signal).

I think Ben West is just coloring in his argument with the 50x stat from the canonical Nordhaus paper, instead of literally declaring it as modifier. I guess you can view this as an instance of a bad information cascade, but I really think it is OK as it is used. He's just making the idea of accounting for the externalities more concrete.

I haven't read the Nordhaus paper, but I assume innovation is a relatively small segment of most companies' activities.
No offense to Lizka, JP et.al, but I would not consider most of their work to be innovation

It's a bit of a worldview/ideological thing, but I honestly disagree with this. I think that things commonly used to measure "innovation", like patents, having a "hot new app", album release, etc. are just outputs, and are different than the actual work to produce innovation.

The work of the forum team can be genuinely very innovative, even for activity or inputs that can seem mundane at first.

Maybe way of seeing this worldview, is that it's extremely difficult to find the true original inventors of many important ideas (who actually invented Rock and Roll or Hip Hop? What were the series of steps contributing to powered flight or rocketry?).

As a sanity check, I don't think most users would pay 50x their Netflix subscription fees to access Netflix (even if the counterfactual entails being cut from Netflix's competitors as well), and I think this generalizes to most companies.

Note that in the Nordhaus paper, the 50x multiple is comparing company profit to consumer surplus.

We agree that Netflix isn't getting $9.95 a month (or whatever it is now) in profit. I think they are getting maybe 20%? So with that 0.2X, things work out to $1,000 a year in consumer surplus, I think that's plausible (maybe a big factor is tail values as with most things, this might be driven by outliers, e.g. the value of Holden Karnofsky examining Beach Boy innovation using streaming media).

Note that the claim of Nordhaus is that companies can't capture surplus. (So in that framework) it isn't surprising that Netflix can't raise it's prices many times over, which I think is close to what you claim.

Linch @ 2022-05-23T16:17 (+8)

Thanks, your reply and expertise is helpful.

Note that the claim of Nordhaus is that companies can't capture surplus

One minor point: What's the causal mechanism here? My naive guess would be that companies can't capture surplus because the surplus will be competed away, which is why I specified "even if the counterfactual entails being cut from Netflix's competitors as well" as a sanity check.

Charles He @ 2022-05-23T17:05 (+4)

Note that my comments risks overstating my expertise. My guess is that for the relevant literature involved, you or a RP employee can easily match it with moderate time investment.

One minor point: What's the causal mechanism here? My naive guess would be that companies can't capture surplus because the surplus will be competed away, which is why I specified "even if the counterfactual entails being cut from Netflix's competitors as well" as a sanity check.

Yes, pretty much. So one answer is just quoting from Nordhaus:

(Screenshotting because it's hard to copy text due to the PDF)

Charles He @ 2022-05-23T17:12 (+4)

There's probably many more ideas that are riffs on this, I'm typing this up pretty much as shower thoughts:

Social design: One example is patents, which only have a finite lifespan. So if you actually invented something world changing, it's not unexpected that you would only harvest a portion of the value, by design.
Private Information: Price discrimination is hard (people have heterogenous demand, it's not like you can read off their willingness to pay off their forehead, and charge them that)
Investment/Uncertainty/Holdup: A lot of the value might involve risky endeavors, e.g. a new company or project. If an entrepreneur is starting something based on your idea, it's hard to to charge them 50% or even 20% of their profit due to your invention, without collapsing the whole thing
Assignment of contribution: Ideas and their impact are probably really complicated, maybe involving combinations of existing ideas, so it's hard to even attribute, much less collect payments on each component.

I think one point of writing up these shower thoughts, is pointing out that it is hard to know.

It's better to have these ideas out there, than just deferring to Nordhaus.

The work of an economist or even Nordhaus, risks glorifying what are complex, ultimately ideological tinged views of the world. These are hard to be certain of, or even communicate to others with different worldviews.

Ben_West @ 2022-05-23T19:13 (+3)

On the sanity check: Reddit makes about four cents in revenue per user per month. It doesn't seem crazy to me that the average user gets two dollars of value per month, but a lot of this would depend on things like how many of their users are diehard versus casual users.

Linch @ 2022-05-23T20:35 (+2)

Ah yeah that seems pretty reasonable, sure.

James Lin @ 2022-05-21T21:17 (+14)

I think this is an interesting consideration!

I'm still curious about the opportunity cost for those people's time, though. On first pass, it seems likely that if you feel overqualified for a position, it probably tracks something important and you should look into alternative ways of maximizing your impact. But it's good to keep in mind that even if you feel overqualified, there are legitimate reasons to still choose that position besides looking for a job.

Jack Ryan @ 2022-05-21T21:57 (+8)

Could someone show the economic line of reasoning one would use to predict ex ante from the Nordhaus research that the Forum would have 50x more employees per user? (FYI, I might end up working it out myself.)

Ben_West @ 2022-05-22T16:45 (+3)

My model was:

It's kind of unclear what a marginal unit is for a software company like Reddit, but let's just say it's one user of their software
Profit maximizing firms produce until marginal cost = marginal benefit
Marginal benefit is 50 times higher
Therefore firms will accept 50 times higher marginal cost
Let's suppose that the percentage of this additional marginal cost which goes to labor remains unchanged, resulting in 50 times more employees per user

30 seconds of thought can identify a bunch of problems with this model, but I think the underlying insight that firms would hire more labor is broadly correct. I would be interested to hear if people think this is wildly off though!

Jack Ryan @ 2022-05-22T19:52 (+1)

Thanks!

Is it correct that this assumes that the marginal cost of supporting a user doesn’t change depending on the firm’s scale? It seems like some amount of the 50x difference between EAF and reddit could be explained by the EAF having fewer benefits of scale since it is a smaller forum (though should this be counter balanced by it being a higher quality forum?)

Continuing the discussion since I am pretty curious how significant the 50x is, in case there is a powerful predictive model here

Thomas Kwa @ 2022-05-22T03:09 (+3)

You can get it from log returns to labor. If impact is k*log(labor) for for-profit firms and 50*k*log(labor) for altruistic firms, the altruistic firms will buy 50x the labor before returns diminish to the same level. I'm not sure this is the right model for companies though.

Jack Ryan @ 2022-05-22T19:57 (+1)

Ah I see — thanks!

Larks @ 2022-05-21T22:38 (+5)

Interesting idea, but (based on about 20 seconds thought) I'm not convinced it holds in equilibrium. If for-profit firms under-invest by 50x, and charities adapt to this by ramping up their investment massively, aren't they then taking resources (e.g. labour) away from the for-profit sector and thus making for-profit under-investment even worse? (Assuming the total consumption/investment tradeoff is unchanged).

Thomas Kwa @ 2022-05-22T01:04 (+12)

Yeah, I think you also have to assume that charities want something other than to create surplus value in the way for-profit companies do. Suppose not-- say there are altruistic and for-profit firms, and altruistic and selfish labor, and the altruists want to maximize total surplus created. Not an economist, but I think the equilibrium looks like:

altruistic firms are better buyers of selfish labor; therefore the price of labor goes up
altruistic employees work for the most altruistically efficient firms, whether they're altruistic or for-profit. For-profit firms are less altruistically efficient because they're optimizing their profit, not total surplus. So they shift to partially optimizing for altruistic value to attract better talent in the face of higher labor prices.

Mati_Roy @ 2022-07-30T21:01 (+4)

read quickly, but basically: value that is harder to capture by the market is more neglected, so actually, there's a lot of opportunities of helping more people per employee in altruistic sectors, so not doing that is an opportunity cost

Kevin Lacker @ 2022-05-21T21:00 (+3)

Is this forum looking to hire more people?

There is also a “startup” aspect to EA activity - it’s possible EA will be much more influential in the future, and in many cases that is the goal, so helping now can make that happen.

I feel like the net value to the world of an incremental Reddit user might be negative, even….

Ben_West @ 2022-05-22T17:11 (+2)

We are looking to hire, thanks! I put a link to our open positions at the bottom of the post.

Charles He @ 2022-05-23T20:59 (+2)

No one has said it, but the main critique is:

The post doesn't make an argument for large EA organizations, it makes an argument that EAs^[1] are just broadly more impactful than they appear.

The fact that EAs might be 50x more impactful, does not imply that we should "add more people than is standard" to an EA organization, because exactly the logic applies if those people joined a smaller organization, created a new one, or started a personal project.

Another way of saying this: Ben West is saying that EAs are worth (50x) more than they appear at CEA. But they are worth 50x to any EA org.

(I don't think this is true but 🏈Something something motivated reasoning^[2]🏈.)

^{^}
This argument probably requires EA to be especially focused on internalizing externalities, otherwise it applies to for-profit orgs, e.g. making Reddit larger too.
This key condition is missing in the OP, but I think EAs do internalize externalities much much more, so it works.
^{^}

Charles He @ 2022-05-23T21:03 (+3)

My guess is that a reply to my comment above would be:

"Ok, fine, but this rationalizes more effective, better managed, and larger teams for higher impact, since we're earlier on the "curve of returns to labor" than it appears and can accept more labor and capital.

Furthermore, I can, and have been, provisioning this management and setting these conditions, so the argument applies and strong EAs should join my awesome team.

Also, hasn't it been a few months since your last forum ban?"

This is true. One reply to this reply, is that this implies EA organizations should be systemically larger than other organizations (due to the OP's reasoning of internalizing externalities and other conditions, see footnote 1 in above comment).

This is interesting, but doesn't immediately doesn't match our intuition or the general size of EA orgs, which are small (but maybe this is masked by the limited supply of high quality EA talent).

If Ben West's logic in the OP holds, this might have other implications that are important or interesting (again this is relying on EAs differentially supporting externalities).