Resigning from West Point

Recently, I decided to resign my appointment to the United States Military Academy at West Point. This was not an easy decision, but I’m convinced it was the right one given the sum of evidence I had available. The bottom line is that this was a careful, dispassionate, utilitarian decision that I had to make in order to maximize my expected impact on the world. The primary supportive reason was the comparative advantage argument.

Comparative Advantage

In any organizations with a finite number of jobs available and more demand for the jobs than supply (e.g. the United States Army), the impact of one’s actions that one can take credit for is not merely all the good that one does through one’s role, but the marginal difference of the impact of one’s actions over the impact of the actions of the person who would have taken the job otherwise. Merely doing a good job, even if doing that job is very taxing on oneself, does not mean one is actually causing that much of an impact. Rather, one’s labor is only especially useful in a given job if the person they are replacing (the next best person who would get the job otherwise) would not do as good of a job.

The equation to calculate one’s impact of taking a job is a bit more complicated than this because one must also consider the marginal impact of the displaced person who doesn’t get your job over the impact that you would have had should you have not taken the job. In essence, although I was not quantifying my intuitive sense of impact with ‘utils’, this was the situation I was facing:

Evan Bob
Army 55 utils 50 utils
Non-Army careers X utils 30 utils

Bob is the person wanting to commission through OCS or ROTC that the Army would permit to commission if I resigned. As you can see, even if I would have done a slightly better job than Bob (which is not guaranteed), it still makes sense for me to resign so long as my impact outside of the Army is more than ‘5 utils’ greater than Bob’s would have been.

I do not wish to share a number for how relatively impactful a career I think I can have outside the Army is. However, I believe that I can likely have an impact greater than ‘5 utils’ more than Bob’s impact outside the Army and frankly, an impact greater than what I would have had in the Army if I use my talents to work hard in especially important, solvable, and neglected cause areas.

That’s essentially it. It wasn’t an emotionally easy decision–I have read my share of military books and had many dreams of leading troops and positively shaping my future units and the Army as a whole. I fully expected to come back to the big Army after I left my position as an enlisted Army engineer Diver to attend West Point.

I could go into more depth about all the details I considered. However, seeing that you know I want to work in impactful and neglected cause areas outside of the Army, I think you get the point that my decision to resign was optimal, or at least, respectable.

Computational Intractability as an Argument for Entropy-Driven Decision Making

BLUF: I have another argument in favor of choosing courses of action via recently produced, quantum-generated random numbers.

Recently, I wrote a long post about a decision criterion for choosing Courses of Action (CoA) under moral uncertainty given the Many-Worlds Interpretation of quantum mechanics. While this has been shot down on LessWrong and I am still working on formalizing my intuition regarding relative choice-worthiness (a combination of Bayesian moral uncertainty and traditional expected utility), as well as figuring out exactly what it means to live in a Many-Worlds universe, I tentatively believe I have another argument in favor of entropy-driven CoA selection: computational intractability.

Traditional decision theory has not focused a ton, to my knowledge, on the process of agents actually computing real world expected-utility estimates. I think the simplest models basically assume agents have infinite computations available. What decision is an agent to make when they are far from being done computing the expected-utility of different CoA? Of course, this depends on the algorithm they use, but in general, what decision should they make when the time to make a decision comes early?

In a Many-Worlds universe, I am inclined to think agents should deliberately throw entropy into their decisions. If they have explored the optimization space to the point where they are 60% sure they have found the optimal decision, they should literally seek out a quantum mechanics generated random number–in this case between 1 and 5–and if the number is 1, 2, or 3, they should choose the course of action they are confident in; otherwise, they should choose a different promising course of action. This ensures that child worlds are appropriately diversifying so “all of our eggs are not in one basket”.

If the fundamental processes in the universe–from statistical mechanics to the strong economic forces present today in local worlds based on human evolutionary psychology–lean in favor of well-being over suffering, then I argue that this diversification is anti-fragile.

A loose analogy (there are slightly different principles at play) is investing in a financial portfolio. If you really don’t know which stock is going to take off, you probably don’t want to throw all your money into one stock. And choosing courses of action based on quantum random number generation is *the only* way to reasonably diversify one’s portfolio; even if one feels very uncertain about one’s decision, in the majority of child worlds, one will have made that very same decision. The high-level processes of the human brain are generally robust against any single truly random quantum mechanics event.

I am still working on understanding what the generic distribution of child worlds looks like under Many-Worlds, so I am far from completely certain that this decision-making principle is ideal. However, because it does seem promising, I am seeking to obtain a hardware true random number generator to experiment with this principle–I won’t learn the actual outcomes, which have to be predicted from first-principles, but I can learn how it feels psychologically to implement this protocol. At this point, it looks like I am going to have to make one. I’ll add to this post when I do.

A Decision Theory for Many-Worlds Living

Here, I describe a decision theory that I believe applies to Many-Worlds living that combines principles of quantum mechanical randomness, evolutionary theory, and choice-worthiness. Until someone comes up with a better term for it, I will refer to it as Random Evolutionary Choice-worthy Many-worlds Decisions Theory, or RECMDT.


If the Many World’s Interpretation (MWI) of quantum mechanics is true, does that have any ethical implications? Should we behave any differently in order to maximize ethical outcomes? This is an extremely important question that I’m not aware has been satisfactorily answered. If MWI is true and if we can affect the distribution of other worlds through our actions, it means that our actions have super-exponentially more impact on ethically relevant phenomena. I take ethically relevant phenomena to be certain fundamental physics operations responsible for the suffering and well-being associated with the minds of conscious creatures.

My Proposal

We ought to make decisions probabilistically based on sources of entropy which correspond with the splitting of worlds (e.g. particle decay) and the comparative choice-worthiness of different courses of action (CoA). By choice-worthiness, I mean a combination of the subjective degree of normative uncertainty and expected utility of a CoA. I will go into determining choice-worthiness in another post.

If one CoA is twice as choice-worthy as another, then I argue that we should commit to doing that CoA with 2:1 odds or 66% of the time based on radioactive particle decay.


Under a single unfolding of history, the traditional view is that we should choose whichever CoA available to us which has the highest choice-worthiness. When presented with a binary decision, the thought is that we should choose the most choice-worthy option given the sum of evidence every single time. However, the fact that a decision is subjectively choice-worthy does not mean it is guaranteed to actually be the right decision—it could actually move us towards worse possible worlds. If we think we are living in a single unfolding of history but are actually living under MWI, then a significant subset of the trillions↑↑↑ (but a finite number) of existing worlds end up converging on similar futures, which are by no means destined to be good.

However, if we are living in a reality of constantly splitting worlds, I assert that it is in everyone’s best interest to increase the variance of outcomes in order to more quickly move towards either a utopia or extinction. This essentially increases evolutionary selection pressure that child worlds experience so that they either more quickly become devoid of conscious life or more quickly converge on worlds that are utopian.

As a rough analogy, imagine having a planet covered with trillions of identical, simple microbes. You want them to evolve towards intelligent life that experiences much more well-being. You could leave these trillions of microbes alone and allow them to slowly incur gene edits so that some of their descendants drift towards more intelligent/evolved creatures. However, if you had the option, why not just increase the rate of the gene edits, by say, UV exposure? This will surely push up the timeline for intelligence and well-being and allow a greater magnitude of well-being to take place. Each world under MWI is like a microbe, and we might as well increase the variance, and thus, evolutionary selection pressure in order to help utopias happen as soon and as abundantly as possible.

What this Theory Isn’t

A key component of this decision heuristic is not maximizing chaos and treating different CoAs equally, but choosing CoAs relative to their choice-worthiness. For example, in a utopian world with, somehow, 99% of the proper CoAs figured out, only in 1 out of 100 child worlds must a less choice worthy course of action be taken. In other words, once we get confident in particular CoA, we can take that action the majority of the time. After all, the goal isn’t for 1 world to end up hyper-utopian, but to maximize utility over all worlds.

If we wanted just a single world to end up hyper utopian, then we want to act in as many possible ways based on the results of true sources of entropy. It would be ideal to come up with any cource of action and flip a (quantum) coin and go off its results like Two-Face. Again, the goal is to maximize utility over all worlds, so we only want to explore paths in proportion to the odds that we think a particular path is optimal.

Is it Incrementally Useful?

A key component of most useful decision theories is that they are useful insofar as they are followed. As long as MWI is true, each time RECMDT is deliberately adhered to, it is supposed to increase the variance of child worlds. Following this rule just once, depending on the likelihood of worlds becoming utopian relative to the probability of them being full of suffering, likely ensures many future utopias will exist.

Crucial Considerations

While RECMDT should increase the variance and selection pressure on any child worlds of worlds that implement it, we do not know enough about the likelihood and magnitude of suffering at an astronomical level to guarantee that the worlds that remain full of life will overwhelmingly tend to be net-positive in subjective well-being. It could be possible that worlds with net-suffering are very stable and do not tend to approach extinction. The merit of RECMDT may largely rest on the landscape of energy-efficiency of suffering as opposed to well-being. If suffering is very energy inefficient compared to well-being, then that is good evidence in favor of this theory. I will write more about the implications of the energy-efficiency of suffering soon.

Is RECMDT Safer if Applied Only with Particular Mindsets?

One way to hedge against astronomically bad outcomes may be to only employ RECMDT when one fully understands and is committed to ensuring that survivability remains dependent on well-being. This works because following this decision theory essentially increases the variance of child worlds like using birdshot instead of a slug. If one employs this heuristic only while having a firm belief and commitment to a strong heuristic to reduce the probability of net-suffering worlds, then it seems that yourself in child worlds will also have this belief and be prepared to act on it. You can also only employ RECMDT while you believe in your ability to take massive-action on behalf of your belief that survivability should remain dependant on well-being. Whenever you feel unable to carry out this value, you should perhaps not act to increase the variance of child worlds because you will not be prepared to deal with the worst-case scenarios in those child worlds.

Evidence against applying RECMDT only when one holds certain values strongly, however, is all the Nth-order effects of our actions. For decisions that have extremely localized effects where one’s beliefs dominate the ultimate outcome, the plausible value of RECMDT over not applying it is rather small.

For decision with many Nth order effects, such as deciding which job to take (which, for example, has many unpredictable effects on the economy), it seems that one cannot control for the majority of the effects of one’s actions after an initial decision is made. The ultimate effects likely rest on features of our universe (e.g. the nature of human market economies in our local group of many-worlds) that one’s particular belief has little influence over. In other words, for many decisions, one can affect the world once, but they cannot control the Nth order effects through acting a second time. Thus, while certain mindsets are useful to hold dearly regardless of whether one employs RECMDT, it seems that it is not generally useful for one to not employ RECMDT if they are not holding any particular mindsets.

Converting Radioactive Decay to Random Bit Strings

In order to implement this decision theory, agents much require access to a true source of entropy—pseudo-random number generators will NOT work. There are a variety of ways to implement this, such as by having an array of Geiger counters surrounding a radioactive isotope and looking at which groups of sensors get triggered first in order to yield a decision. However, I suspect one of the cheapest and most reliably random sensors would be built to implement the following algorithm from HotBits:

Since the time of any given decay is random, then the interval between two consecutive decays is also random. What we do, then, is measure a pair of these intervals, and emit a zero or one bit based on the relative length of the two intervals. If we measure the same interval for the two decays, we discard the measurement and try again, to avoid the risk of inducing bias due to the resolution of our clock.

John Walker
from HotBits

Converting Random Bit Strings to Choices

We have a means above to generate truly random bit strings that should differ between child worlds. The next question is how do we convert these bit strings to choices regarding which CoA we will execute? This depends on the number of CoAs we were considering and the specific ratios that we arrived at for comparative choice-worthiness. We simply need to determine the least common multiple of all the individual odds of each CoA, and acquire a bit string that is long enough that its representation as a binary number is higher than the least common multiple. From there, we can use a simple preconceived encoding scheme to have the base 2 number encoded in the bit string select for a particular course of action.

For example, in a scenario where one CoA is 4x as choice-worthy as another, we need a random number that represents the digits 0 to 4 equally. Drawing the number 4 can mean we must do the less-choice worthy CoA, and drawing 0-3 can mean we do the more choice-worth CoA. We need at least 3 random bits in order to do this. Since 2^3 is 8 and there is no way to divide the states 5, 6, 7 equally to the states 0, 1, 2, 3, and 4, we cannot use this bit string if it is over 4, and must acquire another one until we acquire a number under 4. Once we select a bitstring with a number below our least-common-multiple, we can use the value of the bit string to select our course of action.

The above selection method prevents us from having to make any rounding errors, and it shouldn’t take that many bits to implement as any given bit string of the proper length always has over a 50% chance of working out. Other encoding schemes introduce rounding errors, which only detract from the uncertainty of our choice-worthiness calculations.

What Does Application Look Like?

I think everyone with solid choice-worthy calibrating ability should have access to truly random bits to choose courses of action from.

Importantly, the time of the production of these random bits is relevant. A one-year-old random bitstring captured from radioactivity is just as random as one captured 5 seconds ago, but employing the latter is key for ensuring the maximum number of recent sister universes make different decisions.

Thus, people need access to recently created bit strings. These could be from a portable, personal Gieger counter, but it could also be from a centralized Gieger counter in say, the middle of the United States. The location does not matter as much as the recency of bit production. Importantly, however, bit strings should not ever be reused as this is not as random as using new bit strings as whatever made you decide to reuse them is non-random information.

Can We Really Affect the Distribution of Other Worlds through Our Actions?

One may think that since everything is quantum mechanics including our brains, can we really affect the distribution of child worlds from our intentions and decisions? This raises the classic problem of free will and our place in a deterministic universe. I think the simplest question to ask is: do our choices have an effect on ethically-relevant phenomena? If the answer is no, then why should we care about decision theory in general? I think it’s useful to think of the answer as yes.

What if Many Worlds Isn’t True?

If MWI isn’t true, then RECMDT optimizes for worlds that will not exist at the potential cost to our own. This may seem to be incredibly dangerous and costly. However, as long as people make accurate choice-worthiness comparisons between different CoAs, then I will actually argue that adhering to RECMDT is not that risky. After all, choice-worthiness is distinct from expected-utility.

It would be a waste to have people, in a binary choice of actions with one having 9x more expected-utility than the other, choose the action with less expected-utility even 10% of the time. However, it seems best, even in a single unfolding of history, that where we are morally uncertain, we should actually cycle through actions based on our moral uncertainty via relative choice-worthiness.

By always acting to maximize choice-worthiness, we risk not capturing any value at all through our actions. While I agree that we should maximize expected-utility in both one shot and iterative scenarios alike and be risk neutral assuming we adequately defined our utility function, I think that given the fundamental uncertainty at play in a normative uncertainty assessment, it is risk neutral to probabilistically decide to implement different CoAs relative to their comparative choice-worthiness. Importantly, this is only the ideal method if the CoAs are mutually exclusive–if they are not, one might as well optimize for both moral frameworks.

Hence, while I think RECMDT is true, I also think that even if MWI is proven false, a decision theory exists which combines randomness and relative choice-worthiness. Perhaps we can call this Random Choice-worthy Decision Theory, or RCDT.

I am still actively working on this post, but I am excited enough about this idea enough that I didn’t want to wait to post it. Let me know what you think of this!

The Exponential Impact of Socially-Contagious Philanthropy

If doing any significant amount of good was basically intractable, it would be more permissible for individuals to ignore the utilitarian imperative to do the most good. However, doing incredible amounts of good is in the reach of many of us. We don’t necessarily have to research and contribute to Multiverse-wide Cooperation via Correlated Decision Making in order to do our part; doing good can be as simple as donating 10% of one’s salary to EA Funds, which, if used for causes as effective as the Against Malaria Foundation, can avert a year of lost health (a DALY) for $29. One may be able to do far more good than just this though. Consider the power of exponential growth:

If you commit to convincing two other people per year to donate 10% of their income to the EA Funds, and convince them to convince two people to do the same themselves, etc., you can expect to have 27 people donating 10% of their income within three years. Considering a simple model based on a mean US income of $72,000, one can expect to be responsible for averting 814,097 DALYs within 7 years. This assumes that these people would not do anything productive with 10% of their money if they did not donate it, that none of these people would have discovered effective giving over this time period, and that the $29 per averted DALY rate would hold. Even accounting for more realistic estimates of these factors, it is likely that one could still claim responsibility for averting over 500,000 DALYs over a 7 year period.

This is a substantial amount of good. Frankly, I struggle to imagine how these 2187 people’s discretionary income could be better spent. To say it would be better for people to not deliberately spend part of their discretionary money on charity and research via the EA Funds is to suggest either that the EA Funds managers are ineffective at choosing organizations and causes to give to, or that each person could get about 247 years worth of pleasure by spending that 10% of their income on themselves. I think both of these are highly unlikely to be the case.

If this inspired you, I encourage you to take a giving pledge and share your reasons for taking the pledge on social media! Like all habits, giving is contagious 🙂

Suspend Moral Self-Judgement for Higher-Quality Reasoning

I think most people have an intense psychological need to feel they are ‘good’. After all, if we are not ‘good’, we probably have extra work ahead of ourselves to set ourselves straight and at the very least, preserve our social standing. Some of us do moral calculus all the time in order to stave off guilt and justify our current course of action. The mature among us value intellectual honesty when doing this, and try to avoid jumping to convenient conclusions.

With all this being said, I think a lot of us too often fall short of being intellectually honest because we really value perceiving ourselves as being ‘good’. For example, just consider the most common argument against moral philosopher Peter Singer’s main point in his famous essay “Famine, Affluence, and Morality“. Many people reject his argument because it’s too demanding–not because its clauses are flawed or the logic tying them together is faulty, but because the conclusion implies just about everyone is currently not as good as they think the are.

If people could better suspend their moral self-judgment, they wouldn’t fall into this sort of trap. There is a time and a place to deal with moral guilt (hopefully by altering our behavior), but it shouldn’t be while we are trying to determine moral truth.

If this sounds trivially obvious, when is the last time you felt you were a moral monster? When did you last feel heavy guilt for spending resources on yourself that could be better allocated to reliably avert a lot of others’ suffering? If you’ve never felt that guilt, you may be putting the cart before the horse in your moral reasoning.

Ethics Demands Speed

BLUF: For problems that we are destined to solve, one’s impact can be thought of as how quickly their efforts expedite a solution times the utility of each moment of that expedition.

We are already well on the path to solve a few great ethical challenges that face us. A clear example, in my mind, is factory farming. The science is already here that these animals are very likely to be experiencing immense suffering, we are quickly coming up with replacements for meat (to satisfy stubborn consumers) via food technology, and moral advocacy efforts are leading people to become sentientists and reducetarians or vegans. It’s only a matter of time until the will of the populace leads politicians to pass legislation to drastically tax animal products and improve the conditions of factory farms, or outright shut them down.

There is a lot to be done between now and the end of factory farming, so I don’t mean to detract from the efforts still required at all. In my mind, this is still a very neglected cause area with room for many people to spend a life devoted to it. That being said, I would bet a lot that factory farming will be done away with in the western world by 2100, and hopefully much sooner. Perhaps the simplest moral question then for one to ask is “when?”

There are a variety of ways to measure one’s impact in a cause area. How much does one increase the probability of a solution coming to fruition times the marginal utility of said solution? The other clearest metric to me is how much sooner will a solution be enacted via one’s efforts? If my actions cause a tax spike on animal products to arrive two weeks early and in those first two weeks, we see a reduction in demand of animals by 50 million, which leads to a reduction in supply by 40 million, I am responsible for preventing the suffering of these 40 million animals. That’s a huge impact. On the contrary, if I didn’t do anything for these factory farms when I could have statistically sped up tax-hikes by 2 weeks, then the suffering of those 40 million animals is on my hands.

Ethics demands speed.

[Link] It’s Supposed To Feel Like This: 8 emotional challenges of altruism – Holly Morgan

You should try to do more good in the world because it’s the right thing to do.

Unfortunately, this isn’t always sufficient motivation for people to act accordingly, so altruists[1] often tell non-altruists about the personal benefits of altruism — the evidence that giving makes you happier, the friendly and supportive communities of altruists you can join, the sense of meaning altruism can bring to fill the void left by an increasingly secular world, and so on.

But leading an altruistic life is not always plain sailing, and I think it’s important to acknowledge that from time to time. And not only to acknowledge that times get rough, but the specific ways in which they get rough.

-Holly Morgan, It’s Supposed To Feel Like This: 8 emotional challenges of altruism

Update: An additional useful link about a related topic, scrupulosity

Subjective Probability

BLUF: I found an essay by Nick Bostrom that perfectly coincides with my ideals towards Bayesian probability and how I aspire to consciously hold degrees-of-belief and continually update on evidence.

For me, belief is not an all-or-nothing thing—believe or disbelieve, accept or reject. Instead, I have degrees of belief, a subjective probability distribution over different possible ways the world could be. This means I am constantly changing my mind about all sorts of things, as I reflect or gain more evidence. While I don’t always think explicitly in terms of probabilities, I often do so when I give careful consideration to some matter. And when I reflect on my own cognitive processes, I must acknowledge the graduated nature of my beliefs.

-Nick Bostrom, 2008, Response to the 2008 EDGE Question: “What Have You Changed Your Mind About?”

Unpopular Ideas: Theory and Links

I was inspired by some of Julia Galef’s posts and am going to collect some unpopular ideas here as it’s paramount we consider ideas far outside the status-quo on a continual basis.

Just as many ideas we hold as true today were counter-intuitive in the past, we will probably eventually shift towards accepting a number of currently counter-intuitive beliefs in the future. This seems almost inevitable unless you think we have reached the pinnacle of human progress, but that is probably (not to mention hopefully) not the case. Since an eventual change in many of our beliefs is almost inevitable, we should hasten the progress and let “that which can be destroyed by the truth, be destroyed by the truth.”

One way this can happen quickly is by taking unconventional and unpopular ideas seriously and trying to make them work. Not only will we find hidden truths sometimes, but even when we do disprove these unconventional ideas, we still benefit from decoupling a bit from the status-quo and thinking in a new way.

A counterargument to considering unconventional ideas are that some of them are information hazards and that merely by considering them, one will inevitably be brought astray. I acknowledge that this can be the case in some people, and even in myself if I somehow forget my priors and to be intellectually rigorous in my explorations. That being said, if some of our most rational people cannot safely consider unpopular ideas, humanity probably doesn’t have the intelligence to survive long-term anyways.

I am not endorsing these ideas! I disagree with many of them and am merely collecting them for the sake of intellectual thoroughness.

Julia Galef’s Lists of Unpopular Ideas about:

My List: Coming soon once I get through Julia’s.

A Concrete Way to do Good

Theories abound around doing good. There remains much meta-level thinking to be done in effective altruism and cause prioritization. However, not everyone can or should focus on this research. So for someone looking to have a greater impact without undergoing too drastic of a life change, what solutions do we have for them?

One solution, I believe, is taking the Giving What We Can Pledge:

“I recognise that I can use part of my income to do a significant amount of good. Since I can live well enough on a smaller income, I pledge that for the rest of my life or until the day I retire, I shall give at least ten percent of what I earn to whichever organisations can most effectively use it to improve the lives of others, now and in the years to come. I make this pledge freely, openly, and sincerely.”

If one donates 10% of their income to GiveWell’s recommended charities or the Effective Altruism Funds, they can reliably do a ton more good with their life. Over a lifetime, one can reliably avert thousands of years worth of human suffering if one donates to the most effective (known) charities. Unless one has a particularly high impact job, this 10% donation will likely surpass the direct impact one has with their job and are responsible for.

If the lifetime 10% pledge is too daunting, one may pledge to donate a different chosen percentage of one’s income over a specific duration.

Necessary EA Institutions Part 1: Daily Impact Journaling

Here, I present the case that one aspiring to do as much good as possible with their lives should experiment with keeping a daily log of their impact.

Journaling about impact forces one to reflect on what is, by definition, the most important thing in life. This helps hold one accountable for how they spend their time and helps one pin down their most impactful actions in order to learn from them and replicate them more. Further, where one perceives themselves as failing, it should be easier to muster the motivation to brainstorm how one can be more impactful with their time in the future.

Logging serves as an anchor which can keep even the most aspiring effective altruist grounded.  Even if one thinks about having an impact throughout the day, logging the concrete expected effects of one’s various actions can keep one grounded with the actual impact they are making.

It may be scary to actually write out the expected effects of one’s concrete actions. One can no longer rely on long-term future ideals to satisfy their superego and make one feel like they are living in line with their values.  If it is indeed painful to log one’s impact, this will likely serve as a powerful, necessary motive for change. In the long-run, this active self-accountability should ultimately help one avoid future regrets and it will almost certainly lead to one having a bigger impact on the world.

We Should Have Prevented Other Countries from Obtaining Nukes

Ever since nuclear weapons fell into the hands of the USSR and other countries beyond the United States, human civilization has been under tremendous risk of extinction. For decades now, the Doomsday clock has been perilously close to midnight; we continue to flirt with disaster which could strike once any nuclear leader falls into a suicidal mindset, which breaks the calculus of Mutually Assured Destruction. There is no solution in sight: we will only continue to avoid the destruction of all that we care about insofar as a handful of world leaders value living more than winning or being right. Perhaps down the road, some institution will emerge which will lead denuclearization to non-extinction levels, but even navigating this transition will be risky.

Given the dilemma of this current state of affairs, we messed up. We should have had the strategic foresight to prevent this from happening, and done nearly everything in our power to prevent it from happening. We should have negotiated more fiercely with the Soviet Union to make them stand down their nuclear development, and we should have backed up our words with the threat of bombs. Further, moral philosophy messed up by not laying the groundwork for this to happen at the time: as undesirable as it would have been to target a research lab in Siberia or even a populated city, this pales in comparison to the hundreds of millions, billions, or even all future people (we are talking trillions+) who remain under significant, perpetual risk in the current nuclear environment we created.

We should have never allowed more than one state to develop the Bomb. “But this one state might abuse their power and try to dominate the world” one might counter. This could be the case, but I would venture that one state enforcing its values on another would probably not have been as bad as extinction. Further, this one nuclear state would have an incentive to be good stewards of their power to discourage others’ pursuit of nuclear development; insofar as non-nuclear states are not desperately unsatisfied with their lives, it does not make sense to pursue nuclear development under the threat of annihilation should the one nuclear state find out before they had amassed an arsenal big enough for Mutually Assured Destruction.

My Current Career Path

Update (19 April 2019): My thinking has evolved since this post! Thus, I have submitted my resignation memorandum and plan to leave the Army as I believe this is the personal route that best allows me to maximize expected utility.  This was motivated primarily by the replaceability argument (in the Army, I can only claim credit for the marginal impact I would have over the officer who would otherwise fill my shoes).

I also came to the conclusion that I overemphasized the career capital I would gain from graduating from West Point and serving as an officer (I am still a veteran after all), and underestimated the career capital I could gain from forging my own route should I resign.  I also realized that the skills I would gain in the Army would not translate well to what I want to do afterward and that whatever I started to do while 29 years old, I would be significantly behind my peers.  I have a lot more reasons for leaving the Army which I might post about later.

I am fairly certain that utilitarianism is true, and that I have an obligation to live my life in a way so as to maximize expected utility (I consider the well-being/suffering of all expected future creatures to be the one utility to be maximized).

Obviously, it’s very difficult to know and predict the precise Nth-order effects of our actions; I’m very uncertain about what actions are necessary to maximize net expected utility. Fortunately, I have a few solid options available to me:

  • Drop out of West Point to attend Lambda School (or another coding school), which would reliably lead me to begin earning $80k+ annually within the next year. This might allow me to maximize earning to give in the short term.
  • Drop out of West Point and work full time on effective altruism research, either inside or outside of an organization. This could allow me to maximize short-term intellectual contributions to the world.
  • Finish West Point, serve in the Army for 5 years, and probably maximize my career capital. Don’t get me wrong, I believe that serving as an officer is good in itself, but the impact that I can claim credit for is the marginal benefit I would cause to the Army, the mission, and my Soldiers compared to the officer that I am replacing.

Despite having 120+ credit hours and probably being developed enough at this point to move past college (bold claim, I know), I don’t actually have a Bachelors degree yet (West Point doesn’t take transfer credits). As we all know, this piece of paper carries immense signaling benefits, and I’d probably tremendously limit the jobs I could interview for by failing to get a degree.

Of course, I don’t have to get a degree from West Point, but I do not believe my academic GPA at West Point is good enough to get me into a comparably prestigious school. Thus, my current plan is to graduate from West Point in 2.25 years, serve in the Army as an officer for 5 years (which I am excited about out and have wanted to do since middle school, and then move on and do something different.

I am currently planning on attending grad school for computer science, economics, business school, or something else. Then, the options are endless. I want to do direct effective altruism (EA) research, but I think I would prefer to hold a high-income job and fund the EA research that I want to get done.  I also am thinking about working in startups as it has the potential for great reward and benefit to society, and I will be able to bear the statistically-expected repeated failures for a while due to the support of my future wife, who will be a physicians assistant.

Thanks for reading! If you want to talk careers, feel free to DM me on Twitter (@EvanWard97).

The Need for and Viability of an Effective Altruism Academy

Update: I posted this idea on the EA Forums where it was well-received but deservingly shot down.  While it could be a desirable experiment down the road, there are probably more cost-effective and reliable ways for EA to invest in its own right now, such as ensuring the EA hotel stays well-funded.


Effective Altruism may be able to do a lot more good by investing in growing itself more. Here, I lay down a plausible way for EA to do that.

Why do we need an EA Academy?

Effective Altruism encapsulates some of the most powerful, impactful ideas in the world. It seems that when people focus single-mindedly on answering this question with no preconceived conclusions to justify, remarkable ideas are birthed and great work takes place. The bottom line is: the world probably needs a lot more effective altruists.

The problem, in my mind, is that EA is not growing fast enough. Surely, many people like myself are exposed to EA’s ideas and want to pursue them further, but are too bound up in their everyday lives to dive deeply into them. They want to find a way in which they can drastically increase their impact, but between school, work, traditional obligations, and just not being around other effective altruists, they struggle to figure out how to fully realize their impact.

I believe it is time for EA to begin experimenting with investing more heavily in growing its own. It’s time we start doing more to help people pro-actively become effective altruists. We need an EA Academy where people of all walks of life can come together, study ideas essential for making a high impact, build life-long friendships with other EAs, and become fully integrated with the movement.

What would the school be like?

I am thinking the school could last 1-12 months, but perhaps 3-months for people who want to fully dive into the ideas surrounding EA and who want a much deeper knowledge of its principles. I’m thinking the first 60 days should include an introduction to EA, Bayesian inferencing, cognitive biases, moral uncertainty, utilitarianism, cause prioritization, charity evaluation, progress made so far, and more. Then, students will spend 30 days on application to a cause area such as hastening ethical human genetic engineering.

Admission Criteria

I think all ages (16-40+) should, in principle, be admittable which will allow a lot of diversity in the classroom. Admission criteria should involve: SAT scores/ standardized tests predictive of g-factor and performance, an essay on their desire to come to the school, and perhaps an interview over Skype.


I believe one could get an initial grant from the Good Ventures or the Open Philanthropy Project. The school could then be sustained by a Lambda school-style deal (Income Share Agreement): If you get employed outside EA, you owe 10% of your salary for 2 years once you make over $50,000. If you work for a think tank or inside EA, the cost is waived.

A million dollar grant could fund this for the first 2 years with salary for 2 instructors and food+housing for 8 rotations of 18 students. That’s 144 new EA’s ready to take on the world. I think the movement could get the money back in impact by the ideas and impact generated by 144 new members, who will then have the knowledge and ability to influence others to take up EA.

Cost Break-Down

Two instructors at $80k each per year= $320k.
$500 stipend per month for a total of 144 students: $500 * 3 * 144= $216k.
The rest of the money shall go towards paying off the student housing mortgage. If the project fails, the house can be sold and the profit given back to the granter.

Funding beyond 2 years: Suppose half (72) students go into working in the private sector. Suppose 60 of them make over $50k. Suppose the average income of these 60 is $65,000. $65k * 10% * 2= $13k * 60= $780k. This is enough fund the school considering we put down massive funds early on the mortgage for student housing.


I think there are enough people out there interested in finding a cause and a community that we could get hundreds of applications within a few months to fill the first 144 slots. I could imagine many university students being willing to take a gap semester for this, as well as promising students out of high school who could class up in the spring. If the Effective Altruism community was inclined to hire people who graduate from the Academy, that may only bolster applications.

Ideal Results

18 new effective altruism-minded people ready to take on the world every 3 months. Additionally, we get 144 person months of research done. 12+ years of research from people new to the field and with fresh minds is not a bad deal itself for $1 million. Additionally, these people are now well-equipped to spread the ideas of effective altruism; they are ready to convince their friends to adjust where they donate to and to convince their friends and family to take up higher-impact careers. Additionally, in each class of 18 people, there will undoubtedly be a development of several long-term friendships where people continue to spread high-impact ideas and hold each other accountable for doing good.

Worst Case Scenario

The worst case scenario is that the project gets approved, the project manager ends up hiring ineffective, uninspiring instructors who teach an uninspiring curriculum, and we don’t measure results and student feedback enough to change instructors and/or shut down the project early. This can easily be prevented by extensively interviewing those instructors who are interested, and only running the project once one finds the right people.

Worst Realistic Case Scenario

18n people end up studying high impact ideas for 2 months each and attempt research for 1 month, and decide EA isn’t for them. Even if this is the case, the time studying important ideas should still have been time well spent that ultimately increases their economic output. Additionally, due to the Lambda-style contract, most of the investment can be recovered.

Highly Plausible Scenario

The grant is approved. The person in charge of the project becomes an instructor and finds another. 200+ people interested in effective altruism apply for this program which is set to begin ~6 months out. The two instructors begin sorting through applications and decide on the first 36-72 people. They also begin shopping for housing for the Academy and build up an extensive, detailed curriculum in their free time. 45 days out, they begin getting paid by the grant so they can move (the academy probably needs to be in a relatively cheap city/suburb) and prep full time. The students fly or drive in the day(s) before classes start, settle in, and then the instruction begins. The days are full of discussion, reading and writing assignments, and are focused around molding one into a high-impact person. After 1-7 days, individuals sign an Income Share Agreement if they would like to remain in the course. About 6-12 hours each day are demanded of people’s time, with the weekends less structured and demanding. The instructors ensure students are able to get one on one time with them in order to clarify their ideas. After 2 months of work under a focused curriculum, the students begin working on their own projects and doing their own research. The instructors are there to guide them. Towards the end, students could even fill out a broad application and the instructors could work with the EA community to match interested, high performing students with EA jobs.

What do you think of this idea? Is it necessary and viable?

Let’s Not Decouple Suffering from Survivability

In the interest of avoiding the worst possible futures, let’s not decouple suffering from survivability.

What are some unique elements of the worst possible futures? What features can be found more often in these than other, better futures? I propose that a key feature of the worst possible futures is a strong decoupling of well-being from survival and reproduction. Further, I propose that a practical way for us to reduce the risk of realizing a terrible possible future is to avoid genetically engineering creatures to be more resilient, and instead, focus on increasing their well-being.

Why do we have the levels of well-being that we do today?
Our levels of well-being could be so much worse or so much better, but for me, it seems to generally range (minus short-term extremes) from slightly-below-worth-living to pretty-worth-living. It seems interesting that our well-being hangs pretty close, relative to the range of possible well-being states, to that minimum level which is worth living. It’s easy to imagine how natural selection could have caused this: individuals with lives not worth living either committed suicide before they had kids, didn’t have kids, or just were not as evolutionary successful because of a loss of will. People that were too happy tended to be less careful and less concerned about worldly pursuits, which caused them to reproduce less. They may have also been selected against (whose first instinct is to like happy, go-lucky people when you are hungry and life isn’t going your way?)

Anyways, there seems to be this connection between well-being and evolutionary success. This goes pretty deep: people (and animals) that lose the desire to live can often die within days of giving up, despite no apparent (non-psychologically-induced) physiological cause.[1]

I propose this coupling has helped prevent exorbitant amounts of suffering from happening. The most unhappy people didn’t reproduce as much, so we are left today with this decent level of well-being. Additionally, people and animals in hopeless situations have often not had to endure suffering endlessly (until their physiology failed them)- there has been a way out for them.

This coupling, at least conceptually, did not have to be the case. Why must there be this connection between well-being-space and physical-survival-space? It seems conceptually possible for well-being to have no effect on survivability. No matter how miserable a creature is, its wiring could force it to eat, survive, and reproduce. On a side note, how do we know that is not the case with animals today?

I propose that the worst possible futures are not devoid of sentient life- these are merely neutral futures- but that the worst possible futures are full of self-replicating, miserable creatures. Insofar as a strong connection exists between well-being and resilience (or suffering and evolutionary-failure), we have a pretty strong form of protection from these possible futures. But if we uncouple resilience and well-being, we open the door to a lot more of these horrible futures.

What are the practical implications for this idea today? One thing I propose is that as we begin to do more genetic engineering, let’s not make creatures more resilient before making them subjectively better off. Let’s focus more on ensuring that life is great for however long we live it, rather than designing super resilient creatures who will never give up the fight, no matter how bad a post-apocalyptic future is.

“Getting the Hands Dirty for a Good Cause”- a Short Story

In my Plebe (freshman) year literature class, about 7 months ago, I wrote an adventurous short story (~2600 words) that combines effective altruism, the tech world, fake news, and a moral lesson about the possible tragic consequences of naive unilateral action.  Let me know what you think!

Getting the Hands Dirty for a Good Cause

“I know. You’re disappointed that I didn’t turn out to be like you,” boomed out of the surround-sound system connected to the 68 inch plasma TV, which blocked the view of the Pacific from San Francisco.

“No. No, nonono. I was disappointed… that you tried.”


Morgan clicked off the television, intent to use this temporary wave of positive emotion to do something good with his time. He refused to let himself fall victim to this modern societal “curse” of ennui and complacency that seemed to naturally follow from having his basic evolutionary needs so easily met. He was intent on escaping the boredom this summer day.


He pulled out an index card with 3 things he set for himself last night to do today. He knew that if he didn’t write out goals and accomplish them early in the day, he would probably wind up with little to show for his time. But it was summer, so he certainly didn’t feel like going too crazy. But 3 things was reasonable, and his list for today was:

Create and distribute a flyer for teaching piano lessons.

Get in a sprint workout at the track.

Figure out something really good to do for society.


He had created the flyer in the morning and hung up a few around his neighborhood community center at the beginning of his run to the track, where he had gotten in his track workout. He checked both those off.

He had just finished his lunch of tofu and vegetable stir fry, which was eaten while lounging on the couch and watching Inception, which he found playing on TV.

As it was summer, he had virtually no external obligations, which was awesome at first, but he had found his that life was void of the usual meaning he found when he was busy during the school year.


He had brought up this boredom to his parents and they suggested he volunteer, and he was bored enough that he took their advice: he went online, found a local food bank, and went in for about 6 hours. It helped his boredom a little, but by the end of the day, he felt like he did more good for himself just moving heavy boxes of food around than he did for the hungry people of his area. He wasn’t really redirecting resources to places they wouldn’t go otherwise, he figured; he was just processing them in a job that probably would have been filled by someone else had he not committed to volunteering the hours he did.


So he got interested in doing something good this summer that was a better use of his time.

The only problem was that any ideas he could come up with were far too unappealing. They would constitute a lot of work, and just not have an impact that really made it worth it to him.


But he kept thinking. “How can I do the most good?”, he punched into his Google search bar on his laptop.

The Most Good You Can Do -Wikipedia was his 2nd hit, and he clicked on the link.


Chapter 2

“Mate, toss me another Red Bull” commanded Arthur, who never looked away from his code. He had 4 minutes left out of an hour for this problem in the hackathon, and he was not about to lose this early on the competition. He was almost done 10 minutes ago, but his computer restarted without his permission and he ended up with having to retype everything. You might think a Google software engineer with a Ph.D. in computer science would have automated his computer saving by now, but unfortunately, you would be wrong.


Anyways, there was money on the line- $50,000 for 1st place- and equally important, his pride. He would never hear the end of his coworker’s jokes if he failed to complete this problem. Hiding Moby Dick in a picture with fewer pixels than characters in the text didn’t require learning anything particularly new, but it did require some techniques he hadn’t practiced since his undergrad years at CalTech.


2:59, 2:58, 2:57 the clock went down.


He wasn’t going to make it. He was going to have to have to play one of his strongest cards now. He was going to freeze Gmail. Yes, disable all Gmail transfers. It was a trick he set up while working in the G-suite department and one that only would work once after the patch would be emplaced within probably 2 hours. He was hoping he would end up using it for something bigger, like shorting Alphabet stock, but he was also concerned the Feds and Google’s investigation team would trace it back. This seemed petty enough that no one would trace it back to him. He texted “stop” from his burner phone to his trigger mechanism he implanted in Google’s server farm. He tried to send an email from an account not connected to him to his Gmail account. Nothing showed up in his inbox. It had worked!

1:59, 1:58. 1:57

He got up and took his laptop to the bathroom and got back to his coding and hoped enough people haven’t sent their submissions yet so that competition regulators would extend the time. Obviously, something was wrong if half the people didn’t submit anything this early on.

Arthur’s adrenaline was pumping. He would get blasted by the courts- civil and criminal- if his trick was ever traced back to him. He was fine though, he knew he was. He finished up his text compression algorithm, compressed Moby Dick to about a third of its original size, and hid it into a picture that was not even 2 megabytes. It was already 3 minutes after the clock had run down, but he submitted his work to the competition officials, knowing that he was still in the competition. He checked Gmail’s Twitter, and sure enough, they were already apologizing for technical errors. He felt like how the Inception team must have felt when they woke up on the 747 after implanting an idea in Fischer’s mind. He felt amazing.


Chapter 3

1 week later.


Morgan was typing away passionately. His life had changed so much in the past week. He had discovered the effective altruism movement, which was about using science and rationality to do the most good possible, which generally meant finding charities in developing countries that added the most years of quality human life per dollar. Apparently, some charities were 1000x better than others, and some could reliably add a year of human life for about $100.


But soon later, he had discovered the existential risk reduction movement, whose basic idea was: it would be great to help out some of the 7.5 billion people alive today, but even better to ensure that trillions of people got to exist in the future, which could only happen if humanity does not become extinct. He felt like there was a pretty good chance of happening in the next century, considering the Tweets being exchanged by Trump and “The Little Rocket Man” Kim Jung Un.


Anyways, he found the ideas so compelling that he turned down requests from his friends to hang out. He was just reading the major works in the field and taking detailed notes while he did so. Mainly, he was frustrated by the lack of concern policymakers, or even scientists, seemed to make about this. “There is more scholarly work on the life-habits of the dung fly than on existential risks” he read from Nick Bostrom. Ensuring humanity’s survival was obviously a more critical problem than almost anything else, but it was just not salient enough for the public to vote on. Politicians would be accountable for spending billions on preventing something that over the coming century, would most likely not happen.


While he was doing so, an insane idea came to him: what if he could spread enough fake news about a biological superbug that would get society to actually care about the threat and invest in protecting against it? That would certainly raise the salience of the issue to the American people and get the extra billions he thought was needed for the CDC.


Morgan decided he needed a machine learning expert to help him optimally spread fake news, and was posting to all sorts of websites like StackExchange looking for one, saying how he would pay top dollar for a short-term altruistic problem. By the evening, Morgan had got several Gmails with resumes attached, which he just thought was insane. One of them, in particular, stood out- a Google AI researcher with a Ph.D. from UC Berkeley named Arthur. He figured he was getting trolled by some scam artist, but sure enough, he found the guy’s name on a Google website and his thesis published online.


He emailed Arthur back with his phone number and insisted they call to explain his proposal. Sure enough, Morgan’s phone number rang within 5 minutes, which might have sounded needy except if he did not offer $20,000 for the gig.

“Good evening, this is Arthur Oppenheimer. Is this Jack Rutherford?”

“Yes, it is” Morgan replied in a professional voice, relieved that Arthur sounded just like he did on a video he found on YouTube of him lecturing. And yes, he wasn’t about to give up his name this early on.

“I saw your ad posted for an AI gig, what is it that I can help you with?”

“Yeah, I am not comfortable talking on this unsecured line, do you have Signal?” Morgan asked, writing about an encrypted internet phone app.

“Yes, of course,” Morgan replied, knowing that many entrepreneurs in Silicon Valley were paranoid about getting their ideas stolen. “Same number?”


“I’ll call you right back then.” Arthur hung up and called back “Jack” on Signal.


“Hey Jack, what can I help you with?”

“I am looking to spread a message.”

“Like marketing?”

“No, well yes, but I need an AI researcher to help me do it because I need the message to be adaptable and look like news.”

Arthur, although alone in his apartment, raised an eyebrow, concerned this was some kind of trap. “What kind of news?” he asked.


“Do you believe the ends justify the means?” Morgan asked as calmly as possible, his palms sweating, as he realized the implication of what he was doing.


“I am very concerned organizations like the CDC and the NIH are not being funded enough to do research into preventing abuse of CRISPR and the creation of superbugs. I think there is a significant risk of a catastrophic outbreak, and do to the interconnectedness of everyone, a chance that 99% of people could become infected with a bug that could be made in a lab or inadvertently in a factory farm.”

“I hear you. What are you asking of me?”

“I want you to help me create a fake pandemic and get as many people to believe it as possible.” Morgan was sweating profusely now.

Arthur was shocked by the question, and while he hated what Cambridge Analytica did, this didn’t sound bad to him. He thought about some of the algorithms he could write to propagate this “news” to himself, but he was getting nervous too. What if this was a trap, or even if it wasn’t, what if this idiot would mess something up and get him caught?


“Nah mate, not interested. You really ought to go about the proper channel to make the change you want to see in the world.” Arthur hung up, glad that he had made the right call. He had a great job, and he just got the $50,000 in his bank account from the hackathon. He didn’t need to risk what he had built for himself over $20,000.


Morgan felt relieved in the ensuing silence, but he also felt ashamed that he wasn’t going through the proper channels. No, he told himself, the proper channels don’t work, and you have to take this into your own hands. He knew that existential risk movement was doing all they could, petitioning Congress, writing books. No one was getting their hands dirty, so he had to do it. If he were caught, the world would eventually thank him, although he wasn’t sure he wouldn’t end up in jail.


Arthur had trouble sleeping that night, mad at himself for failing to exercise at all. His mind wandered back to the strange conversation he had earlier, and he considered the offer again. It would be an excellent opportunity to apply one of his algorithms he had developed in grad school, but he was concerned about getting caught. Other people were so careless about their digital footprint. He himself narrowly dodged a bullet from today, when he almost considered calling Jack with his burner phone from the hackathon, which was still off and with the battery out. The idea came to him: What if he did it himself? He could anonymously reach out to Jack with his Monero (anonymous cryptocurrency) public key, and offer to do it, so long that he got, say $5,000 after the first bit of news surrounding the article appeared on Twitter, and $15,000 after he got mention of it on the local news.


He got up then and emailed Jack anonymously and inquired about the job offer, pretending to be a new person.


Morgan had been up, responding to more emails, generally focusing on ensuring the quality of the applicants more than he was explaining his job. He did not want to get hung up on again. He needed to find the right person.


“Send your CV for further consideration, please” Morgan replied.

“How many people have you interviewed for the job?” asked Arthur.

“Plenty” lied Morgan, unsure of the real usefulness of the question.

Arthur sent back, “I know your job, and I’ll do it. I’ll need you to look for the hashtag #biothreat on Twitter when I tell you to, and then you will pay me, and then I will do more, and then you will pay me more, and we can do this. But I am going to do this myself.” Arthur also attached his Monero public key along with instructions on how to pay.

“Ok, I trust you,” replied Morgan. What else could he do?

“Save the Monero key, and delete these emails, and I’ll contact you in a couple weeks.


Chapter 4

Two weeks later, after the two exchanged their emails, the #biothreat hashtag started trending on Twitter, but more significant than that, reports started trending about a nuclear bomb detonating over Delhi, India. Initial fingers naturally pointed at Pakistan, and situation rooms all around the world were rapidly filled with their leaders and staff. However, the Indian’s claimed responsibility- they said they had made the extremely difficult call to neutralize a rapidly spreading virus that was released from a biological warfare lab and killing people within the hour. India’s president reported that scientist said the virus had likely already spread to 100,000 people, and they believed it was being dispersed by the wind.


Little did they know, except for Morgan and Arthur at the time, that Arthur’s black box AI algorithm, designed to maximize reshares and retweets, had acted a horribly wrong. Except it hadn’t. It had acted right, per the goals programmed into it. Way too right.