Tag Archives: Cognito Mentoring meta

Increasing the pool of people with outstanding accomplishments

Cross-posted from Less Wrong

In How can Cognito Mentoring do the most good? I included a section on our potential social value. I want to flesh out what we hope to achieve.

Consider the following people:

  • Scott Alexander. One of Less Wrong’s major contributors, and writes Slate Star Codex. His articles regularly get hundreds of Facebook shares.
  • Bryan Caplan. GMU economist, blogger at EconLog, inspired the creation of Open Borders.
  • Alex K Chen. Quora celebrity who has asked 20k+ questions and ~2k answers. One of our adviseesreported to changing his major based on some of Alex’s Quora answers.
  • Paul Christiano. Shifted away from doing pure math and theoretical computer science exclusively, and has played a major role in the effective altruist community, doing research for GiveWell, MIRI and 80,000 Hours, and is involved in a miscellany of other related projects.
  • Anton Geraschenko, Scott Morrison and David Zureick-Brown. They started MathOverflow, one of the first and largest online communities for academic researchers to discuss research-level questions.
  • Elie Hassenfeld and Holden Karnofsky. Started GiveWell, which moved ~$17 million to its recommended charities last year, with the money moved growing exponentially over time.
  • David Jay. Founded the Asexual Visibility and Education Network which has 70k registered members.
  • Paul Niehaus. An UCSD economist who foundedGiveDirectly. GiveDirectly appears to be much more cost-effective than most charities that work in the developing world. It’s one of GiveWell’s top recommended charities, and is perhaps the first organization of its kind.
  • Richard Rusczyk. Founded Art of Problem Solving, which offers ~50 classes at a given time for high potential math students, and which hosts forums that have 144k registered users.
  • Peter Singer. A philosopher who may be the most pivotal figure behind the animal welfare movement and theeffective altruism movement.
  • Ted Suzman. Cofounded Graffiti Labs while in college, the first product of which of has over 1.5 million monthly users, and is earning to give.
  • Eliezer Yudkowsky. He started MIRI, which could appreciably reduce the risk of extinction from artificial intelligence. He started Less Wrong, which has 8+ million page views per year, has given rise to communities around the world, and which people have described as substantially improving their rationality. He’s writing Harry Potter and the Methods of Rationality (which has high consumptive value and which has raised awareness of MIRI and principles of rationality).

Some of these accomplishments are more impressive than others, but all of them are impressive, and most of the people listed are quite young, and will plausible do more impressive things along similar lines as they get older.
Some common threads that I see in these people are:

  • They’ve all taken an unconventional path, doing something that they would not have done if they were making their decisions by mirroring the behavior of the people around them.
  • They’ve contributed much more than have people of their ability level who take more conventional paths. In some cases, the factor by which the value of their activities has increased is perhaps ~2x, in others it may be more like ~100x.

Unconventionality isn’t necessarily a path to success, and there are plenty of people who adopt unconventional paths and don’t get much done at all, but when executed well, it’s possible to pursue an unconventional path with relatively little risk and high potential upside.

We think that we can enable more people to engage in activities like the ones above. Many of those who are well-suited to them are already engaged in them. But there are others who have most of the relevant traits for whom there are only one or two limiting factors. Some ways in which we think that we can remove the limiting factors are as follows

  • By connecting people with better learning resources and raising awareness of the benefits of learning particular subjects, we can help them pick up relevant subject matter knowledge.
  • By encouraging meta-cognition, we can help people become more goal-oriented, recognizing when their life aspirations may be better served by departing from conventional routes.
  • By raising awareness of the sorts of impactful side projects and entrepreneurial efforts that others can engaged in, we can help them get into the mindset of seeing opportunities, and judging when they’re promising.
  • Through networking, we can help our advisees find collaborators.

According to student feedback we’ve had some success on the first two fronts. We’re continuing such efforts, and are in the process of working on the latter two.

By moving people in the directions suggested above, we hope to tip more people into the high achieving pool that has the above as representative members. We expect that we can enable an average of one additional person per year to get into this achievement range, with the benefits accruing throughout their lives.

Concerning the feasibility of this: The number of people with the requisite traits is not very small. As above, the people on the list have in some cases achieved far out of proportion with their ability, so there are a fair number of people of the same ability level who don’t. So far we’ve had a number of advisees who probably have similar characteristics to people on the list above at the same age. So it’s not necessary to influence a huge number of people to succeed (though we’re casting as wide a net as possible.)

Assuming the estimate here is correct, we get a lower bound on the social value generated by Cognito Mentoring. We have other sources of social value, which we touched on in our earlier post and might elaborate in later posts.

How can Cognito Mentoring do the most good?

By Jonah Sinick

Cross-posted from Less Wrong

In late December 2013, I announced that Vipul Naik and I had launched Cognito Mentoring, an advising service for intellectually curious young people.

Vipul Naik and I are aspiring effective altruists, and we started Cognito with a view toward doing the most good. We’ve learned a lot over the past 3 months, and are working on planning what to do next. We’d be very grateful for any feedback on current thinking, which I’ve described below.

Our Mission

Human capital is one of society’s most valuable resources, and school years (ages 5 through 22) are a crucial time period for building human capital. Education is a ~$1 trillion dollar sector, but schools are often dysfunctional institutions, and very little effort goes into helping young people develop as much as possible and to allocate their human capital as well as possible. We want to help optimize young people’s life trajectories. For the time being, we’ve chosen to focus on helping highly intellectually capable young people. Some reasons for this are:

  • Intellectually capable people contribute disproportionate social value (e.g. Bill Gates solved an unsolved mathematics research problem as a sophomore in college; the Google co-founders were computer science graduate students at Stanford), and helping them develop is correspondingly more leveraged.
  • We have deep knowledge of the population.
  • The educational infrastructure is designed for the average student, and the gap between how things are and what would be optimal is greatest for the outliers.
  • By focusing on a subpopulation, we can offer more targeted recommendations.

Some ways in which we aim to help them improve their life trajectories are:

  • Encouraging reflective decision making and meta-cognition: we get them thinking about what they want out of life, and how best to attain it. In this respect, we overlap with CFAR.
  • Highlighting the advantages of learning different subjects to help them decide which ones are most important to learn.
  • Pointing them to the best learning resources available.
  • Helping them find high value extracurricular activities to engage in.
  • Informing them of the advantages and disadvantages of different career choice. In this respect, we overlap with 80,000 Hours (while differing in that our focus is on people who won’t be entering the job market for several years).
  • Connecting them with people who have subject matter knowledge in their academic or professional areas of interest.


We’re primarily targeting high school and college students within the range of intellectual ability of Less Wrongers.

About 75% of respondents to the 2013 Less Wrong Survey who reported SAT scores out of 2400 gave a score of 2130+: this is at the 98th percentile of SAT takers. There are ~40,000 people per grade in that score range in the United States nationwide, so ~320,000 Americans. When one accounts for people at lower percentiles who would benefit, as well as students from other countries, the relevant population is ~1 million.

We’re also well equipped to serve people of younger ages who are highly gifted, and are at a developmental stage where they’re capable of engaging in metacognition and learning high school and college level material. There are perhaps ~200,000 such people worldwide.

Our activities

At the time when we posted in December 2013, we were thinking of focusing on personalized advising, perhaps with a view toward becoming a franchise. Since then, we’ve shifted in the direction of focusing on producing written content. There are two reasons for this:

  • Our advisees have derived most of the value from our generic written content.
  • While  of our advisees have benefited very substantially, the average benefits per person don’t seem to be outsized.

Based on the first point and the size of the target population, if we can produce high quality written content and disseminate it widely, in principle 100k+ people could get a large fraction of the benefit of personalized advising for free.

So far ~70 people have contacted us, including ~40 from Less Wrong (c.f. What we learned about Less Wrong from Cognito Mentoring advising). We corresponded at length with a substantial fraction of them. We’ve taken the advice that we’ve generated and converted it into dozens of articles on our advice wiki, at our Quora blogon Less Wrong and at the Davidson Institute Gifted Issues Discussion Forum. (We’ll be consolidating everything into the wiki eventually: the reason that we’re posting to multiple forums is for outreach purposes and to get feedback.)



Our front page has been getting ~400 page views per week, and our wiki has been getting ~400 page views a week. Our Quora blog has 27 followers. We would like our visibility to increase by 1000x.

We’ve struggled to find avenues by which to disseminate our advice. There seem to be few forums where smart high school students congregate. Those forums and mailing lists that do exist often have strict guidelines against posters promoting their own blogs. We’re grateful that Less Wrong has been welcoming.

We’d appreciate any suggestions for how we might be able to reach more people.

Where will the social value come from?

The main avenues through which people generate social value and disvalue are

  1. Career
  2. Side projects / volunteering
  3. Donating to charity
  4. Enjoying recreational activity, health and wealth
  5. Relations with family and friends
  6. Having children

We have to offer our advisees advice that improves their lives for them to find it worthwhile, but we think that our social impact will be mediated primarily through the impacts of #1 and #2 on others.

It may be surprising that we highlight #2. One reason that we highlight it is that high school and college students tend to have free time outside of school, that they can spend more productively on side projects than on the relatively low-skilled part time jobs that are available to them without the credential of a college degree. Another is that it can be hard to find funding to work on something of high social value full time. Some examples of successful side projects created by members of the effective altruist / Less Wrong communities are:

Why don’t we expect our impact to be through #3 (donating to charity)?

  • A lot of the people well-suited to making money already do it by default: while there are individuals who would do more good taking a higher paying job and earning to give, we wouldn’t expect to be able to boost people’s salaries a lot on average, given the constraints that they operate under, both with respect to skills and with respect to the sorts of work they’d be willing to do.
  • Our advisees won’t be making a lot of money for a long time — by the time they do, they may have had a lot of exposure to the ideas of effective altruism through other channels (whether through existing organizations such as GiveWell and 80,000 Hours, or through future organizations).
  • For effective altruist types, 80,000 Hours Executive Director Benjamin Todd has said that he doesn’t think that it’s plausible that earning to give is likely to be the path toward doing the most good. I gave more points against earning to give as optimal effective altruism in Earning to Give vs. Altruistic Career Choice Revisited.
  • We’re in a different cultural sphere from people in finance and business / consulting, and better suited to help people who are engaged in more intellectual endeavors.

Concerning #4, the benefits would not be leveraged; concerning #5, one would expect the benefits would be ~1x the benefits to the individual, which isn’t a large multiplier; concerning #6, we wouldn’t expect to have much impact on people’s decision to have children, the sign of the effect would be ambiguous, and our advisees are far from the point of actually raising children.

Here are some examples of channels through which we expect to have a positive impact on #1 and #2:

  • There’s a widespread misconception amongst high school students that they have to engage in particular extracurricular activities (or many extracurricular activities) to get into good colleges. By raising awareness that this is not the case, we can free students up to engage in substantive side projects such as contributing to open source software projects and writing Wikipedia articles on important topics — things that both have direct social benefit and that build skills that are useful for future activities.
  • We’re disseminating information about the benefits of computer programming, pointing people to programming learning resources, and pointing people to information about how to learn programming. By reaching high school students, we can help people get a head start, preparing them for the option of becoming software engineers, which will (in expectation) move people into the tech sector, which has unusually great positive externalities.
  • A moderately large fraction of intellectually capable people go to graduate school and end up not using their degrees (e.g. because they’re unable to get jobs in the academic market), or end up doing research of little practical relevance. By disseminating information on academia as a career option and promoting an unbiased view of the value of theoretical research, we can divert people into careers where they can make a difference.
  • By educating people about the unconventional path of entrepreneurship as a career option (for example, by pointing them to entrepreneurship learning resources and connecting them with entrepreneurs who we know) we can enable more people to innovate more than they otherwise would.
We’re very interested in further ideas along these lines, as well as suggestions for how we can realize them.


We originally thought in terms of supporting the operation by charging for personalized advising. This could still be an option, but:

  • High school and college students generally don’t have much money.
  • Most of the students who we advised said that knowing what they know now, they would have sought advising from us only if it were free. This is true even of those who reported to benefiting substantially, suggesting that we can’t resolve the issue by improving the quality of our advice.
  • Students only need ~5 hours of advising from us at a given time, so even to the extent that people are willing to pay, there’s substantial overhead involved per paid hour.
  • While personalized advising does feed into our public content at the current margin, if we had to focus on it heavily, it would distract from producing the more valuable public content.

At this point, we’re seeking philanthropic funding, and would appreciate any ideas as to how to secure it.