I feel the desire to make the largest pro-social contribution I can make and I feel that I have identified some core projects that have a lot of upside. I see this post as an early submission for peer review focused on getting the basics directionally right rather than trying to leap to a refined estimate

Contender #1: Land Value Tax

I think there are two dimensions to rating the projects, being total potential value vs my likely influence. I notice that influence doesn't need to be specifically correct as long as it's relatively well adjusted to the chance of my having influence on the other projects. These two initial estimates for the land value tax should be longer as they establish my estimation process

Value

According to Wikipedia's Excess Burden of Taxation page the excess burden of taxation is between 2.5 and 30%, so I assume 15%. US governmental revenue for '22 is $5 trillion, giving me a loss of $750 billion. In the ACX land value tax series Doucet says Krugman and Friedman agree that it has no deadweight loss, making the avoided loss the theoretical value of the project. I'm ignoring net benefits but it's probably quite optimistic to assume the LVT will be fully substituted for other taxes. Since the LVT also shifts tax from productivity to wealth it must have a marginal benefit multiplier in terms of greater relative benefit for low income people. I guess a trillion sounds good?

Chance of Impact

This is a fairly well known policy which suggests to me that additional marginal advocation for it will experience a liminal multiplier where a small effect may help get the whole thing rolling downhill. Although, it does have a lot of headwind as people tend to recoil from the massive change and the negativity biased sense that they could get boned by this. If I look at a bell curve where the LVT rests half a standard deviation from the peak of neutral acceptance it looks to me like one unit of advocacy at that point will have 50% of the effect of a maximal unit at the peak experiencing the least climb or difficulty so I will give my unit a .5 multiplier. It seems to me that my visual guess of the liminal effect has significant error bars

Since there are 30,000 registered LessWrong users I assume there are 90,000 people with a significant interest in the LVT and 1% doing significant advocacy, for 900 people. I assume I am potentially at the 25th percentile of advocates for this. A power law curve assuming 80% of the effect is accomplished by the top 20% would suggest that I am then making about one unit of advocacy out of the 720 people making the remaining 20% effect. My advocacy value is then (.2/720) x .5 for .00014 out of a 100% successful effort

If the LVT has a 25% chance of being adopted then my final share is .000035 of a trillion for a $35 million value annually once the LVT were fully adopted. I can see that I have the absolute numbers (likely) very wrong so I will just adjust this process by 1000 for a $35,000 dollar value. As long as I keep the relative chance of impact ok the specific value isn't critical within this process. This sounds like a pretty good use of effort

Contender #2: Earned Income Tax Credit

Value

The CBO estimates that a $15 national minimum wage would cost one million jobs persistently. Losing a job results in lost earnings of about one year as I recall (90% confident, may apply more or less to minimum wage jobs), and the marginal effect of earnings at low income levels is very strong on self reported happiness. Involuntarily not being involved in work or education or similar for a year or longer is said by Jordan Peterson to be one of the most detrimental experiences a person can have, making this much more significant than the difference between minimum wage and the nominal welfare benefit

This seems to introduce a lot of guessing to the value estimate. If I assume a tax credit has essentially no loss relative to a high minimum wage then its minimum value should be about 100% of the high minimum wage losses. Notably this ignores positive net benefits which may not be what they are claimed as there seems to be a trend of assuming any kind of economic program has a magic multiple benefit of its spending cost. But this could be a big oversight 

A year of minimum wage is about $16,000 giving me nominally $16 billion annually for the value of a million jobs. Adjusting by feel for the non linear greater marginal effect I'm getting $50 billion. On the international GDP to reported happiness graph I see a rise from 5.5 at the US minimum wage to 6.5 at minimum wage plus $16,000 and most of the charts in the image search look more linear than I recall so this could be an overestimate and I revise to $30 billion annually

Chance of Impact

Considering the potential headwinds of the LVT this seems much more possible. I think it's essentially at the top of the bell curve in the medium term for a liminal multiplier of 1

Support for the tax credit is definitely a lot more broad than the LVT, lets say 10x for 9000 significant advocates. Then my share is 10x less for .000014 units x 1. It seems like there's a lot of sensitivity being given to the share of advocacy effect vs both the liminal and marginal multiplier. I do suspect that the marginal multiplier on this should be much more relative to the LVT as it directly targets people on the margin

If an EITC general substitution for minimum wage raises has a 75% chance then I get a $625,000 annual value adjusted to $625

Contender #3: Dynamic Social Media Regulation

This one is fun because it is by far the most subject to value speculation. As far as I know this is a novel idea also putting it up in the air for liminal effect. Being low on the bell curve lengthens out the horizontal movement of one unit of advocacy again, probably by a very large amount, while the chance of implementation must be much lower

The concept is that rather than prescriptive and reactional policies to reduce the harm of social media algorithms keying on the euphoria of emotional tribalism, there should simply be a large tax shift from the most harmful platforms to the most beneficial. Carbon tax for social media. Carbon taxes are said to be twice as effective as whatever else politicians usually do and I think by also promoting beneficial social media effects this is relatively even better. So I think I could value this as having an additional value equal to the value of successful prescriptive regulation

Value

The printing press caused many civil wars by similar dynamics and this theoretical cost overwhelms any other consideration, perhaps. I have heard that Facebook was influential in at least one civil war. Military spending went to around 50% of GDP for major contenders in the world wars, so that would be like $10 trillion for the US, probably for a short period of time and with a very low(?) likelihood. 1% might be a really high guess but I'll take it for the insurance value of preventing something that definitely costs more than money. Averaged over 100 years, assuming by then we have solved social media in some fashion, I get an annual risk of $5 billion for a 1% chance of one $50 trillion event. Interestingly this is by far the lowest annual value estimate

Chance of Impact

Let's say that roughly speaking I create the existence of this policy. My marginal impact must be high. Let's say the liminal multiplier is 10. As the only original advocate for some amount of time I am 100% of the advocacy, but still with 25th percentile skills, for an effective 20% share, and .2 x 10 = 2. I now have an advocacy share above 1 but it is simply a relative figure based on the liminal multiplier concept. The relative impact might still hold despite accidentally taking 200% credit in absolute terms 

If the policy has a 5% chance of success then I get $500 million annually adjusted to $500,000. This has arrived at a large relative value that is quite sensitive to multiple variables. Based on the next paragraph I will correct it again by a factor of five to $100,000

I can't tell how much this is in error. I think the concept of value that I have landed on is that one year of total advocacy might advance a policy by one year even if the amount of advocacy is that of one person. If the preceding year hadn't existed the next year might not, especially for an original concept. On the other hand the Free State Project observed that 20,000 core activists nearly got Quebec to vote for independence, and that applied to New Hampshire 10,000 people might do the same for libertarianism. Their website says "thousands" have moved, and Bing search sounds relatively upbeat about the significance of their effect. This creates the possibility that advocacy share should be partially normalized to about 10,000 people for low numbers of advocates so that if you are one of 1000 you don't take credit for a 10x greater share than you are really capable of. I do think the concept of marginal effect continues to exist and may be very large at initiation but especially for a 25th percentile advocate there must be a spot where that breaks down. It's possible that the advocacy share should be much lower but the liminal multiplier should be quite a bit higher. As a best guess I will divide the result by five

Results

Dynamic Social Media Regulation - "$100,000", large error bars
Land Value Tax - "$35,000"
Earned Income Tax Credit - "$625"

It's nice to think that maybe 25th percentile advocacy for a year is worth as much or more than I might give (on the order of the EITC number) if I focused on that instead. If I did advocate in a sustained manner I might be motivated to donate more to reduce the sense of prediction error due to feeling like I was doing a certain amount of good but not knowing because the statistics are made up

Thoughts

I estimate a 20% chance I made an order of magnitude math error that I didn't catch. I have mixed feelings about AI risk and therefore my very small share and small delta of advocacy about it rule it out of consideration for me

The sensitivity to advocacy share is interesting to me and I would particularly be interested in opinions on it because I also have a concept of a project that could very significantly increase my advocacy share in whatever I do, although the project itself has a low potential success rate

I notice that advocacy share may be quite sensitive to Scout and ripple effects. Climate policy has a billion advocates but the rapid changes or turnover as people adjust their position for or against means that a Scout advocate who came up with a more convincing and effective set of policy responses could have a very large effect over time even from a seemingly low direct advocacy share standing

Also notable is that advocacy share can change by a multiple if I find one policy much more enjoyable to advocate for and then put in a multiple of my baseline effort. This suggests that I might favour my preferred option within half an order of magnitude spread

Questions

Obviously any alternate figures or comments on my estimates are really interesting as well as alternate methods within the framework or other framework alternatives

I can't picture a way to reliably estimate the absolute value of my effort so some method that might accurately do that could be interesting. For starters I think the liminal multiplier has to be a maximum of 1, making it .1 at the top of the bell curve. Psychologically I know I don't want to hear that there are obvious better uses of my time but that might be the most useful tip

Are there other contenders for high value projects that pencil out in these ranges? What's the total annual value of your big idea?