Relevant calibration game that was recently posted -  I found it surprisingly addictive - maybe they'd be interested in implementing your ideas.

Meta-level: Great comment- I think we should be starting more of a discussion around theoretical high-level mechanisms of why charities would be effective in the first place - I think there's too much emphasis on evidence of 'do they work'.

I think the main driver of the effectiveness of infectious disease prevention charities like AMF and deworming might be that they solve coordination/ public goods problems, because if everyone in a certain region uses a health intervention it is much more effective in driving down overall disease incidence. Because of the tragedy of the commons, people are less likely to buy bed nets themselves. 

For micronutrient charities it is lack of information and education - most people  don't know about and don't understand micronutrients.

Lack of information / markets

Flagging that that there were charities - DMI and Living Goods - which address these issues, and so, if these turn out to explain most of the variance in differences in cost-effectiveness you highlight then these need to be scaled up. I never quite understood why a DMI-like charity with ~zero marginal cost-per-user couldn't be scaled up more until it's much more cost-effective than all other charities.

There is a good Cold Takes blog on the 'Bayesian mindset' - which gets at something related to this as '~20-minute read rather than the >1000 pages of Rationality: A-Z (aka The Sequences).'

Summary:

This piece is about the in-practice pros and cons of trying to think in terms of probabilities and expected value for real-world decisions, including decisions that don’t obviously lend themselves to this kind of approach.

The mindset examined here is fairly common in the “effective altruist” and “rationalist” communities, and there’s quite a bit of overlap between this mindset and that of Rationality: A-Z (aka The Sequences), although there are some differing points of emphasis.¹ If you’d like to learn more about this kind of thinking, this piece presents a ~20-minute read rather than the >1000 pages of Rationality: A-Z.

This piece is a rough attempt to capture the heart of the ideas behind rationalism, and I think a lot of the ideas and habits of these communities will make more sense if you’ve read it, though I of course wouldn’t expect everyone in those communities to think I’ve successfully done this.

If you’re already deeply familiar with this way of thinking and just want my take on the pros and cons, you might skip to Pros and Cons. If you want to know why I'm using the term "Bayesian mindset" despite not mentioning Bayes's rule much, see footnote 3.

 

This piece is about the “Bayesian mindset,” my term for a particular way of making decisions. In a nutshell, the Bayesian mindset is trying to approximate an (unrealistic) ideal of making every decision based entirely on probabilities and values, like this:

Should I buy travel insurance for $10? I think there's about a 1% chance I'll use it (probability - blue), in which case it will get me a $500 airfare refund (value - red). Since 1% * $500 = $5, I should not buy it for $10.

(Two more examples below in case that’s helpful.)

The ideal here is called expected utility maximization (EUM): making decisions that get you the highest possible expected value of what you care about.² (I’ve put clarification of when I’m using “EUM” and when I’m using “Bayesian mindset” in a footnote, as well as notes on what "Bayesian" refers to in this context, but it isn’t ultimately that important.³)

It’s rarely practical to literally spell out all the numbers and probabilities like this. But some people think you should do so when you can, and when you can’t, use this kind of framework as a “North Star” - an ideal that can guide many decisions even when you don’t do the whole exercise.

Others see the whole idea as much less promising.

I think it's very useful to understand the pros and cons, and I think it's good to have the Bayesian Mindset as one option for thinking through decisions. I think it's especially useful for decisions that are (a) important; (b) altruistic (trying to help others, rather than yourself); (c) “unguided,” in the sense that normal rules of thumb aren’t all that helpful.

In the rest of this piece, I'm going to walk through:

• The "dream" behind the Bayesian mindset.
  • If we could put the practical difficulties aside and make every decision this way, we'd be able to understand disagreements and debates much better - including debates one has with oneself. In particular, we'd know which parts of these disagreements and debates are debates about how the world is (probabilities) vs. disagreements in what we care about (values).
  • When debating probabilities, we could make our debates impersonal, accountable, and focused on finding the truth. Being right just means you have put the right probabilities on your predictions. Over time, it should be possible to see who has and has not made good predictions. Among other things, this would put us in a world where bad analysis had consequences.
  • When disagreeing over values, by contrast, we could all have transparency about this. If someone wanted you to make a certain decision for their personal benefit, or otherwise for values you didn’t agree with, they wouldn’t get very far asking you to trust them.
• The "how" of the Bayesian mindset - what kinds of practices one can use to assign reasonable probabilities and values, and (hopefully) come out with reasonable decisions.
• The pros and cons of approaching decisions this way.

I really don't understand how distributing nets can keep people in poverty. 

There is one  paper from 2009 suggesting that, in the short run, eradicating malaria can lower income per capita slightly but only by a few percentage points and in the longer run it raises income. This is because it doesn't affect prime age workers so much.

Question for economists: Would this drive up the prize of the remaining bad debt in expectation, so that so that the marginal utility created by this go down significantly?

Agreed.. a good way to think about this is that since you get ~5% annual returns on stocks, annual rent equivalent is ~5% of the property value, and so the opportunity cost is spending ~$750k/y or $62.5k per month on conference accommodation.

My opinionated and annotated summary / distillation of the SBF’s account of the FTX crisis based on recent articles and interviews (particularly this Bloomberg article).

Over the past year, the macroeconomy changed and central banks raised their interest rates which led to crypto losing value. Then, after a crypto crash in May, Alameda needed billions, fast, to repay its nervous lenders or would go bust.

According to sources, Alameda’s CEO Ellison said that she, SBF, Gary Wang and Nishad Singh had a meeting re: the shortfall and decided to loan Alameda FTX user funds. If true, they knowingly committed fraud.

SBF’s account is different:

Generally, he didn’t know what was going on at Alameda anymore, despite owing 90% of it. He disengaged because he was busy running FTX and for 'conflict of interest reasons'.^[1] 

He didn’t pay much attention during the meeting and it didn’t seem like a crisis, but just a matter of extending a bit more credit to Alameda (from $4B by $6B^[2] to ~$10B^[3]). Alameda already traded on margin and still had collateral worth way more than enough to cover the loan, and, despite having been the liquidity provider historically, seemed to be less important over time, as they made up an ever smaller fraction of all trades.

Yet they still had larger limits than other users, who’d get auto-liquidated if their positions got too big and risky. He didn’t realize that Alameda’s position on FTX got much more leveraged, and thought the risk was much smaller. Also, a lot of Alameda’s collatoral was FTT, ~FTX stock, which rapidly lost value.

If FTX had liquidated, Alameda and maybe even their lenders, would’ve gone bust. And even if FTX didn’t take direct losses, users would’ve lost confidence, causing a hard-to-predict cascade of events.

If FTX hadn’t margin-called there was ~70% chance everything would be OK, but even if not, downside and risk would have been much smaller, and the hole more manageable.

SBF thought FTX and Alameda’s combined accounts were:

1. Debt: $8.9B
2. Assets:
3. Cash: $9B
4. 'Less liquid': $15.4B
5. 'Illiquid': $3.2B

Naively, despite some big liabilities, they should be able to cover it.

But crucially, they actually had $8B less cash, since FTX didn’t have a bank account when they first started, users sent >$5B^[4] to Alameda, and then their bad accounting double-counted by crediting both. Many users’ funds never moved from Alameda, and FTX users' accounts were credited with a notional balance that did not represent underlying assets held by FTX—users traded with crypto that did not actually exist.

This is why Alameda invested so much, while FTX didn’t have enough money when users tried to withdraw.^[5]

They spent $10.75B on:^[6]

1. $4B for VC investments
2. $2.5B to Binance buy out its investment in FTX (another figure is $3B)
3. $1.5B for expenses
4. $1.5B for acquisitions
5. $1B labeled 'fuckups’^[7]
6. $0.25B for real estate

Even after FTX/Alameda profits (at least $10B^[8]) and the VC money they raised ($2B^[9] - aside: after raising $400M in Jan, they tried to raise money again in July^[10] and then again in Sept.^[11])—all this adds to minus $6.5B. The FT says FTX is short of $8B^[12] of ~1M users’^[13] money. In sum, this was because he didn’t realize that they spent way more than they made, paid very little attention to expenses, was really lazy about mental math, and there was a diffusion of responsibility amongst leadership.

While FTX.US was more like a bank and highly regulated and had as much reserves as users put in, FTX int’l was an exchange. Legally, exchanges don’t lend out users' funds, but users themselves lend out their funds to other users (of which Alameda was just one of). FTX just facilitated this. An analogy: file-sharing platforms like Napster never upload music themselves illegally, but just facilitate peer-to-peer sharing.

Much more than $1B (SBF ‘~$8B-$10B at its peak’^[14]) of user funds opted into peer-to-peer lending / order book margin trading (others say that this was less than $4B^[15]; all user deposits were $16B^[16]). Also, while parts of the terms of service say that FTX never lends out users' assets, those are overridden by other parts of the terms of service and he isn’t aware that FTX violated the terms of use (see FTX Terms of Service).

—

For me, the key remaining questions are:

1. Did many users legally agree to their crypto being lent out without meaning to, by accepting the terms of service, even if they didn’t opt into the lending program? If so, it might be hard to hold FTX legally accountable, especially since they’re in the Bahamas.  
2. If they did effectively lend out customer funds, did they do it multiple times (perhaps repeatedly since the start of FTX), or just once?
3. Did FTX make it look like users' money were very secure like a highly regulated bank and that their money wasn’t at risk e.g. by partnering with Visa for crypto debit cards^[17] or by blurring the line between FTX.us (‘A safe and easy way to get into crypto’) and FTX.com?
4. Did FTX sweep users to opt into peer-to-peer lending?

1. ^{^}

    FTX Founder Sam Bankman-Fried Says He Can’t Account for Billions Sent to Alameda  

2. ^{^}

    ‘We kind of lost track’: how Sam Bankman-Fried blurred lines between FTX and Alameda | Financial Times 

3. ^{^}

    Sam Bankman-Fried’s trading shop was given special treatment on FTX for years | Financial Times 

4. ^{^}

    FTX Founder Sam Bankman-Fried Says He Can’t Account for Billions Sent to Alameda 

5. ^{^}

    SBF Reveals FTX Was Selling Assets That Didn't Exist - The Defiant

6. ^{^}

    ht/ to Ryan Carey: ‘notably some of this could be consistent with macro conditions crushing their financial position, especially the VC investments in crypto.’

7. ^{^}

    I think he might refer to this: archive.ph/ATPHq#selection-1981.172-1981.301 

8. ^{^}

    Milky Eggs » Blog Archive » What happened at Alameda Research.

9. ^{^}

    Investors Who Put $2 Billion Into FTX Face Scrutiny, Too - The New York Times 

10. ^{^}

     Crypto Brokerage Genesis Tries to Raise Funds and Eyes Bankruptcy 

11. ^{^}

     FTX in talks to raise up to $1 billion at valuation of about $32 billion, in-line with prior round 

12. ^{^}

     ‘We kind of lost track’: how Sam Bankman-Fried blurred lines between FTX and Alameda | Financial Times 

13. ^{^}

     Sam Bankman-Fried’s trading shop was given special treatment on FTX for years | Financial Times 

14. ^{^}

     See video interview here: FTX Founder Sam Bankman-Fried Says He Can’t Account for Billions Sent to Alameda  

15. ^{^}

     https://twitter.com/adamscochran/status/1593020920695660546 

16. ^{^}

     FTX Tapped Into Customer Accounts to Fund Risky Bets, Setting Up Its Downfall - WSJ 

17. ^{^}

     Crypto Exchange FTX's Token Surges 7% After Visa Partnership Report 

Protect our Future PAC spent unprecedented levels on Carrick's campaign, and they seem to have spent $1.75M on attack ads against Salinas, which maybe biggest 'within party' attack ad budget in a primary. Seems understandable this can be seen as a norm violation (attack ads are more sticky) and perhaps it's poor 'cooperation with other value systems'.

On the other hand, SBF donated to the House Majority PAC, which financed John Fetterman's campaign.

It received the honorable mention prize and the winner of the contest had a similar proposal and also commented in this thread. So it's on Openphil's radar.

One recent paper  suggests that an estimated additional $200–328 billion per year is required for the various measures of primary care and public health interventions from 2020 to 2030 in 67 low-income and middle-income countries and this will save 60 million lives. But if you look at just the amount needed in low-income countries for health care - $396B - and divide by the total 16.2 million deaths averted by that, it suggests an average cost-effectiveness of ~$25k/death averted. 

Other global health interventions can be similarly or more effective: a 2014 Lancet article estimates that, in low-income countries, it costs $4,205 to avert a death through extra spending on health^[22]. Another analysis suggests that this trend will continue and from 2015-2030 additional spending in low-income countries will avert a death for $4,000-11,000^[23].

For comparison, in high-income countries, the governments spend $6.4 million to prevent a death (a measure called “value of a statistical life”)^[24]. This is not surprising given the poorest countries spend less than $100 per person per year on health on average, while high-income countries almost spend $10,000 per person per year^[25].

GiveDirectly is a charity that can productively absorb very large amounts of donations at scale, because they give unconditional cash transfers to extremely poor people in low-income countries. A Cochrane review suggests that such unconditional cash-transfers “probably or may improve some health outcomes.^[21]  One analysis suggests that cash-transfers are roughly equivalent as effective as averting a death on the order of $10k .

So essentially  cost-effectiveness doesn't drop off sharply after Givewell's top charities are 'fully funded', and one could spend billions and billions at similar cost-effectiveness, Gates only has ~$100B and only spends~$5B a year.