Range and Effective Altruism

I wrote this around 4 months ago and have been planning to edit it since. Until that happens...

This post is me making a case for thinking that developing ‘range’ (a wide range of skills, knowledge and experience) is under-incentivised in the EA community.

I’m borrowing the concept of ‘range’ from David’ Epstein’s book ‘‘Range: How generalists triumph in a specialised world’. The title captures the main ideas of the book pretty well, and I think it makes a decent case for developing range being important.

I’m mostly assuming that developing range is good in this post, and focus on the question of whether it’s sufficiently incentivised - if you’re interested in how valuable building range is, I’d recommend reading the book. I thought it was pretty decent and gave it the honour of a five star review on goodreads.

Summary

Developing Range is Good

• Range - having a breadth of knowledge, skills and experience - is good. This is because it enables you to explore opportunities, prepare for opportunities that you can’t predict in advance, and learn.

Developing Range is Hard

• The best way to build range is to have a career which involves testing out lots of different roles and projects.

• But it’s hard to have a career like that, because jobs typically have circumscribed roles which make it hard to explore within them, and because changing jobs is hard.

Why is Developing Range Hard?

• It’s hard to build range in part because it’s not in the interest of organisations to encourage it.

• For example, building range takes a while to pay off. By that point the employee might have already left to do something else, which means the organisation doesn’t get any benefit.

EA Organisations and Range

• Creating an environment where people can build range imposes short-term costs on organisations. In particular, encouraging range trades off against producing legible impact.

• EA organisations should invest more in building the range of their employees compared to non-EA organisations, because EA org employees will probably still do impactful stuff after they leave the organisation (meaning that investments in their long-term development aren’t wasted).

How can EA make building range easier

• It’s hard to work out how exactly to balance the costs against the benefits, but my best guess is creating a more ‘range friendly’ environment would mean a better allocation of people to roles, better skilled people and overall better outcomes for EA.

• Concrete examples of things EA organisations could do to make building range easier: quitting bonuses for employees, sabbaticals, internships.

Building Range is Good

Three reasons why range is good:

• Information value: The more range you have, the more understanding you'll have of what opportunities exist, how valuable they are, and how well you fit them.

• Predictability: The more range you have, the better prepared you will be for unknown or unpredictable opportunities. This is useful because the world is hard to predict.

• Learning: The more diverse your work, the more you will learn, and the more you'll be able to do things well.

Side note - the ‘learning’ reason is the one I find most interesting personally. I’m particularly interested in how true the following is:

Having experience in multiple domains will increase your ability more than having experience in just one domain, even when you know what the ability you want is in advance and can practice it directly. Eg. experience doing event management, communications and people management will increase your event management ability more than if you spent the same amount of time just doing event management.

Anyway, here’s a quick sketch of what a career that optimises for building range might look like

1. Do a liberal arts degree where you can learn about lots of different stuff.

2. Try out many different types of work for 3 months each over the course of 2-3 years. Eg. research, teaching, candlestick making, event management.

3. Focus on a couple of areas which seem the most promising, and try out a bunch of different types of work within that for a year or so at a time. Eg. focus on research as an area, and test out qualitative research, quantitative research, research communications etc.

4. Pick whatever seems best and specialise in that, but still expect to switch roles lots more times.

I don’t have a great sense of exactly how good building range is. But, to me it seems like a person who followed a rangey career path like this would end up contributing more value than a person that followed a more typical career path.

Actually having a career path like that is hard though, and there are lots of obstacles to doing it.

Building Range is Hard

Building range involves trying out lots of different projects, responsibilities, work environments etc. which can be difficult to do.

Range within a job

Jobs come with specific projects and responsibilities. Some jobs have more room for range than others, but in general jobs require you to focus on doing a specific thing. If your job is candlestick maker, trying event management will probably be out of scope.

Range between jobs

You can get range by doing different jobs, but switching jobs is hard.

Applying for jobs takes a bunch of time, which people often don’t have if they are already in a job. You can quit your job and then apply for stuff, but this means you need to back yourself to find something better, and also have enough runway for the search period. Also CV gaps might not look great to prospective employers.

Jobs are big things - typically a full-time, multi-year experience. The typical length of a job doesn’t lend itself to range. Also, despite the size of the commitment you make in starting a job, it’s pretty hard to test them out in advance. In switching jobs, there’s always the risk that it will be bad. Then you either have to have a sub-par multi-year experience, ask for your old job back, or try and get a different one. Also, having lots of short stints on your CV might not look great to prospective employers.

Overall, there's a bunch of uncertainty, risk and effort involved in switching jobs, and no one likes any of those. People are risk averse. If you’re lucky enough to have a ‘good’ job with good money, status and a good work environment etc. switching jobs means you have more to lose.

Range outside of a job

You could build range outside of your job by taking weekend courses, evening classes, doing side projects and the like, but this will be pretty slow compared to building range through your job.

Why is Building Range Hard?

I think part of this can be explained by organisations’ incentives.

Range within a job

Employees in an organisation have a job to do, and so organisations are usually keen for them to do that and not do other stuff. As before, if your job is candlestick maker, your employer probably wants you to make candlesticks and not try your hand at event organisation, or whatever other opportunities could increase your range.

Plausibly, the opportunities for range shrink as organisations grow. In the early days of a candlestick startup, you might actually have to do event organisation as well as candlestick making and a bunch of other stuff. As the organisation gets bigger, it will want employees to specialise in specific roles. This makes sense for an organisation because it gets the gains of people specialising according to their comparative advantage, and also because the leadership needs a way to make the work of employees legible as the distance between the leadership and employees grows.

Range between jobs

Organisations don’t want good people to leave them. Having a good person leave means the organisation has to find someone else and train them up which is lots of hassle. Organisations may want to poach good people from other organisations, but not as much as they want their own good people to stay. So on net, organisations are going to create incentives for people to not switch jobs.

Range within a job again

If no one could switch jobs and organisations knew that they would be stuck with the same employees forever, they would probably want to invest a bunch in them so they could do their jobs really well. They might want to give them lots of different opportunities so they can find their comparative advantage within the organisation, and help them develop generally useful skills so that they can switch between roles as necessary.

But organisations know that people are going to leave, and when they do, all the training spent on them isn’t going to be very useful for the organisation anymore. So they’ll probably want to train them up a bit so they can do their job, but not as much as if they were there forever. They’ll also want to train them up in specific skills which will help them do well in the short-term, but be less bothered about general skills or opportunities for exploration that will pay off in the long-term, as they might have left by then.

EA Orgs should incentivise range

Most organisations have fairly narrow goals, e.g. make profit by selling candlesticks or something. If someone leaves the organisation, they aren’t going to make any more candlesticks, so they aren’t useful for the organisation’s goals. EA organisations have much broader goals like building a flourishing world. If someone leaves an EA organisation, there’s a reasonably good chance they’ll go on to do something useful towards the organisation’s goals. Maybe they go and work at another EA organisation, with a different way of building a flourishing world. Maybe that’s even more useful than what they were doing initially.

Because of this, EA organisations should invest in building the range of their employees more than typical organisations, whether this is letting them explore different roles within the organisation, helping them develop general skills, or encouraging them to explore opportunities for impact elsewhere.

One obstacle for EA organisations is that the impact produced by employees developing range isn’t very legible. Potential funders, stakeholders, community members etc. judge EA organisations based on their impact. EA organisations will also judge themselves based on their impact.

The impact of a staff member leaving for a better role isn't great in terms of legible impact - eg. an impact report which says 'Barbara left to do something better' or ‘Barbara took some time off her main role to try out event management’ might not be a great impact report. Also, Barbara developing range takes away time that she could be spending on achieving the organisation’s main objectives, which are legible.

How much should EA Orgs incentivise range?

It’s hard to tell exactly how much EA organisations should invest. This is going to depend some on:

1. Just how valuable is having range compared to not having range? How long does it take to pay off?

2. How much more valuable is work now vs. later?

3. How likely is it that someone who leaves an EA organisation will continue to work on things with EA aims?

How can EA orgs make building range easier?

Building range is hard because: jobs have restricted scope, jobs are typically for multiple years, leaving a job involves financial risk, finding a new job takes time, building range doesn’t look great on a CV etc.

So the EA community could make building range easier by:

1. Having broad roles which allow for exploration and can be approached from different angles

2. Having more short-term opportunities

3. Reducing the financial cost of exploration and switching jobs

4. Simpler job application processes

5. Making exploration more ‘high status’

Concrete ideas for how organisations can incentivise range:

Leaving bonus - eg. a lump sum for people that decide to quit or are fired

• This ensures that people automatically have financial runway to explore options after leaving an organisation, and makes it more likely that they’ll be able to find a position where they can do a lot of good.

• [Also, this might reduce the need of organisations to fire people. If someone is performing less well/ isn't a good fit for the organisation they might be more likely to quit as a result. It might also make it less hard for organisations to fire people as the cost of being fired for the individual is lower. The harder it is to fire people the higher the threshold for hiring people, and the more difficult it is to for someone to get a job as a result, and so the more difficult it is for people to switch into the job and get more range]

Sabbaticals - e.g. paying people to take time to explore other opportunities/ develop skills

• This means that people can explore different options or build skills without having to leave their job.

• Sabbaticals may mean that: A) Instead of someone staying in their current role, someone can try out a new role, realise that they can have more impact there, and switch to that. B) Instead of someone switching jobs, someone can try out a new role, realise that they can have more impact in their original role, and stay in that.

• Similar ideas: Exploration Days - eg. a set amount of days a year that people can take off to explore other opportunities. Also, that thing where tech companies give employees a day or so a week to let employees work on their own projects

Other Ideas

• Early stage career programmes that are built around exploration (RSP seems like a good example of this)

• Offering internships - I expect these are underprovided

• Extended (eg. 3 month) work trials. It seems like these are somewhat common already, though I don’t know whether this requires people to take a break from their current job or quit entirely.

• Organisations giving input on and helping their employees to achieve their career plans

Range and EA Braindump

• It might be that an individual can maximise their own career success by building range, but that a community can maximise good by having people specialise early. Don’t think this is right, but plausible.

• Football teams buy players from other teams. Plausibly this means that there is a stronger incentive to train them well, as they will be compensated according to the player’s ability.

• What factors lead to an increase in labour mobility?

• An obstacle to organisations adopting policies that encourage range building is the variance of employees’ dedication/ potential outside the organisation.

• EA organisations might have slightly differing goals/ worldviews. The more different they are, the more they are going to not want their employees to leave (like non EA organisations).

• The highest status jobs are in a small number of organisations/ areas and are hard to get. People who get these jobs won’t want to leave, because it will be hard to get another job with as much status. This means the best people will explore less.

• The stronger the organisational focus on having employee performance, the more building range is disincentivised, as building range is a poor way of increasing short-term performance.

• Why do management consulting firms have an up or out model where lots of people leave?

(Appendix) Notes on 'Range: Why Generalists Triumph in a Specialized World'

I recently read 'Range: Why Generalists Triumph in a Specialized World' - by David Epstein. It's one of the most interesting books I've read recently and I've given it the honour of a 5 star rating on goodreads.

Range isn't explicitly defined in the book, but my interpretation of it is having a broad set of skills, experience and understanding. The main thrust of the book is arguing that range is a big factor of being successful, and that people typically over-specialise.

Why is Range valuable?

My breakdown of the value of having range is:

1. Information value: The more range you have, the more understanding you'll have of the value of different work opportunities, and your degree or personal fit for these opportunities.

2. Predictability: The more range you have, the more things you'll be able to do. The harder it is to predict in advance what will be valuable to work on, the more useful having range is.

3. Learning: The more diverse your work, the more you will learn, and the more you'll be able to do things well.

The information value/ predictability stuff didn't seem too novel to me - eg. 80k seems to have talked about this a bunch already in terms of career exploration. The importance of range for learning is something that I'd thought about less beforehand, and I've updated towards thinking this is more important.

I was particularly interested in the idea that the best way to become good at X isn't just to do X, but is to do X, Y and Z. Eg. Even if you know that you are going to be an ML researcher it might be best to focus on learning maths, writing skills and machine learning than just focusing on machine learning.

The Cult of the Head Start

The book starts off by characterising the 'cult of the head start' - the idea that the key to excellence in a given domain is specialising early and amassing (similar to the 10,000 Hours idea). I think I previously kind of implicitly bought into this view. The book gives these examples

1. Sofia, Susan and Judit Polgar - were brought up to become amazing at chess, were homeschooled and played chess all the time. One became an international master, one a grandmaster and the other was ranked 8th in the world.

2. Tiger Woods - was brought up to become amazing at golf. Started playing golf before he could talk, won an under 10s tournament when he was 2 or 3 etc.

3. Harri Besceli - played lots of age of empires as a child.

Easy and Hard Learning Environments

It then goes on to argue that these domains (golf and chess) are pretty unique. They are 'easy' learning environments, with regular feedback, static rules and predictable environments. Most things that we care about are in 'hard domains', where there's little feedback, no rules and the environment is hard to predict. In hard learning environments, range is much more important. One example talked about in the book is a study showing that the best predictor of the success of comic books is the number of different genres the author has previously written in. Years of experience itself isn't a good predictor.

The main ideas for range being more important than narrow experience in hard learning environments:

• Learning in different areas means you can understand the general and fundamental principles that underly multiple domains

• Learning in different areas means you have a bank of problem solving analogies which can be applied to different domains

• Lots of areas require lots of different skills and areas of understanding rather than one big skill.

Career Trajectories in EA and Misc Thoughts

The main way I see this as being relevant to EA is in terms of what kind of career trajectories we should encourage/ incentivise. The book made me update towards thinking that people should be aiming to gain more diverse experience, and work in multiple different roles and career paths before specialising. (The book talks about the 'Dark Horse Project', a study of successful people's career histories, apparently showing that successful people typically have worked in lots of different seemingly unrelated areas and roles before specialising late).

It seems pretty plausible to me that EA over incentivises early specialisation.