An adapted, somewhat updated version of this article posted to the EA Forum a few months ago: https://forum.effectivealtruism.org/posts/i2Q3DTsQq9THhFEgR/introducing-effective-self-help
Effective Self-Help (ESH) is a pilot initiative researching the most effective ways people can improve their wellbeing and productivity. The project is currently funded by the EA Infrastructure Fund.
Improving your ability to function happily and productively should help you to be more effective in trying to solve the most pressing global problems. Effective Self-Help is an attempt to synthesise the best advice in service of this goal, slowly building a single, highest-quality source of information for anything related to wellbeing and/or productivity.
This post explains in greater depth what Effective Self-Help is, what I hope the project can achieve, and why I believe this work may be highly impactful. As this is a long post, I suggest reading the Summary section below and then skipping to whatever sections seem of greatest interest.
Effective Self-Help is currently an ongoing series of reports and articles researching key wellbeing and productivity topics. So far, we have published reports on sleep, stress, and habit formation interventions. We intend to develop the project into a home for the best guidance on how individuals can improve all significant aspects of their wellbeing and productivity.
In the What is Effective Self-Help? section, I discuss how I think ESH can be unique as a self-help resource by combining several key components:
Under Why Effective Self-Help could be highly impactful, I envisage three main routes to impact which are discussed in greater depth in the ‘Theory of Change’ sections. These are:
I created Guesstimate models for the first two of these routes to impact, producing best-guess estimates that ESH could add 180 hours of productive work and 4.2 QALYs per article. However, these results are highly uncertain and best interpreted as an indication that ESH has the potential to be highly impactful if things go well, rather than a strong endorsement that it will be.
Downside risks and arguments for ESH being ineffective discusses a range of ways in which Effective Self-Help could be of limited value or of active harm. The main downside risks discussed are that ESH could:
The main ways listed for how that Effective Self-Help could be of limited effectiveness are:
In the Final notes section, I discuss what we're currently working on and how you can help by filling out our evaluation survey (which is useful to us even if this is the first time you've heard of Effective Self-Help).
An early-stage research organisation studying effective wellbeing and productivity interventions.
Currently, ESH is an ongoing series of articles researching the most effective actions anyone can take to improve their wellbeing and productivity. All of our research up to now can be found on our website.
Effective Self-Help aims to be comprehensive, gradually covering all aspects of improving wellbeing and productivity that are highly effective. Future articles are planned and/or underway for a wide range of topics, including:
You can suggest topics that we should cover in the comments below, at the end of our feedback survey, or by getting in touch with us directly.
A one-stop shop for learning how to live better.
The big idea of Effective Self-Help is to provide comprehensive guidance on how best to improve your wellbeing, researching interventions across the full spectrum of wellbeing and productivity topics. My primary aim is to create a one-stop shop for self-help advice, helping people and organisations maximise the effectiveness of efforts to improve their wellbeing and/or productivity.
Topics will expand over time to include any area of wellbeing and productivity that appears important, from improving relationships to the relative effectiveness of different supplements. With a backbone of research articles, ESH will aim to present this information in the most useful and engaging format possible, such as through videos, podcasts, or interactive programmes.
Self-help is a huge industry. Additionally, even just within the EA community, there are already good projects and articles on self-help. What makes ESH different and valuable?
Below are the 7 key components that I believe make this project new and worthwhile. While there are existing projects[1] that meet some of these criteria, I am unaware of any that meet most or all. On that basis, I think Effective Self-Help can offer something distinctive and highly valuable, slowly outcompeting lower-quality projects by providing better, more trustworthy guidance in a superior format for free.
Many self-help articles provide lists of actions to take to improve a certain aspect of mental health and wellbeing. However, these articles tend to lack any transparent methodology for how actions have been selected, which might be the most effective, and by what margin.
In line with one of the cornerstone findings of effective altruism, that some charities are >100x as effective than others, it seems likely to assume that some self-help interventions are >10x as effective than others for a given individual.
This makes the prioritisation and ranking of actions essential for directing people to the most effective improvements to their wellbeing and productivity. Considering the time constraints of many people doing high-impact work, prioritisation appears even more vital.
Effective Self-Help aims to produce better recommendations than are currently available through deeper reviews of the available scientific literature, a more consistent and rigorous research framework, an emphasis on cost-effectiveness, and the use of a Bayesian approach to evidence and value.
In particular, current self-help literature often focuses on the statistical significance of study results when the focus should be directed to effect sizes. Much like hits-based giving, it is potentially far more valuable to pursue an intervention that could be very high impact but with uncertain evidence in its favour than something that we are near certain has only a small impact.
Effective Self-Help will rank interventions based on their average effect size, directing attention to how much something will help rather than just that it is of some help, and highlight the most effective interventions.
Though I am still uncertain of the best method for doing so, I intend for ESH’s recommendations to also account for the costs (e.g. time, money, risks) involved in making different changes and for ESH to use a wider basis of evidence (with appropriate weighting) than just available scientific studies to determine what may be most effective.
A substantial proportion of popular self-help guidance is in written form, either in books or on blogs. While much of this is useful, I think there is a general lack of curation. Recommendations are often buried partway through the text and clear, actionable advice can be difficult to find.
By curating recommendations and presenting them prominently, ESH could cause a much higher percentage of its readers to make changes based on the recommendations presented than the average self-help resource. The articles on sleep and stress provide good examples of trying to present key takeaways in this kind of quick and obvious way.
While written advice can be effective in presenting a clear argument, I think much greater benefit would come from the clear presentation of key conclusions and arguments in more engaging formats. I envisage the greatest benefit from presenting the research findings in non-text formats.
Videos, podcasts, online courses, and interactive programmes all could plausibly make the changes we recommend more engaging to learn about, easier to understand, and therefore easier to implement. Organisations like Clearer Thinking, The School of Life, and Waking Up provide good examples of efforts to provide valuable content in more engaging formats.
While there is a staggering range of self-help literature, there are very few projects looking to cover all aspects of useful life advice/ guidance. I have started Effective Self-Help as a relatively limited research project to test its effectiveness, only covering a few relevant areas, but the long-term aim is to provide comprehensive self-help research, covering all plausibly impactful topics and interventions.
A singular location for high-quality guidance could save people substantial time, ending the need to trawl through resources and establish what is actually useful. It could also plausibly help direct people to the most effective recommendations based on their circumstances, with people finding the website through an interest in a certain topic but then discovering interventions that are also or even more useful for them.
Having a quick, screening quiz for prioritising recommendations prominently placed on the website could achieve this quite efficiently.
As discussed above, a lot of good self-help advice is presented in a blog format. While this can be good for valuably changing the way we think about a topic, I think the greatest benefit comes from providing people with specific actions and changes they can make.
A focus on practical action helps to reduce the percentage of readers who will finish an article and agree with its conclusions but fail to make any beneficial changes in behaviour based on its ideas.
As part of this, particular emphasis will be given to providing support for implementing new behaviours and researching how to maximise people’s success in adopting new habits. Knowing what to do to improve something is not enough; ESH must ensure that people are best supported to follow through with effective action.
Existing mental health projects often primarily target people who currently suffer from poor mental health. While this is undoubtedly important, there is also an opportunity to provide advice to people simply looking to optimise their wellbeing and productivity, whether through eliminating more minor issues or experimenting with potential enhancements (improved diet; nootropics; effective time management; etc.). This would look to build on the kind of work presented in many previous articles here on LessWrong, and by writers like Gwern and Scott Alexander elsewhere.
In this way, we hope to produce small changes in wellbeing and productivity that add up to a large effect for people who may already have relatively high wellbeing and productivity. In this way, the project roughly emulates the philosophy of marginal gains in elite sport, made famous by Dave Brailsford’s work for British Cycling and Team Sky.
Current self-help resources are often influenced by a desire to sell a certain product, course, or book. The best advice, presented in the best way, is saved for those willing to pay for it as its producers depend on these revenue streams for their income. Even with the best of intentions, I think this financial incentive can compromise the quality of advice given, at least to an extent, such as when:
As a moonshot goal, the project would become the default location for wellbeing and productivity advice, continually updating guidance on the most effective interventions based on new research.
Self-help is something we all do all the time. Many of our everyday decisions - from what we eat, read and watch, to how we communicate and why we maintain certain habits - are motivated by a belief that these choices will improve our quality of life.
But if the general quality of information available to us is low, we are likely making sub-optimal decisions. With better knowledge, we could improve the choices we already make regarding how to maximise our wellbeing. Great self-help advice highlights actions we take that have a significant impact on our wellbeing but, with conscious thought and small changes, could be made substantially more beneficial.
The following sub-sections discuss three theories of change for Effective Self-Help. As the project is still at an early stage and I have limited information upon which to weigh their relative value, I have not yet prioritised between these paths.
Members of the EA community and adjacent groups/ organisations do high-impact work on the world’s most pressing problems. Improving their productivity increases the amount and quality of high-impact work on these problems by increasing the expected quality and quantity of work completed. Improving wellbeing could further increase productivity and produce a range of indirect benefits to work quality and quantity.
I envisage two main routes to impact in this way:
1. Incremental improvements for lots of people (marginal gains)
2. Profound improvements for a small % of people
As an aside, ‘high-impact’ work is an unhelpfully vague term that I am using to avoid getting into the weeds of the relative impact of different jobs. Roughly speaking, I am defining high-impact work as either earning to give, working for an EA organisation, or working on a high-priority cause (X-risks; animal welfare; etc.), though this likely has significant flaws as a framing.
To gain a better impression of the potential value of ESH in increasing the number and quality of productive hours for EA community members, I made a rough Fermi calculation through Guesstimate. This allows the user to input a value range rather than a fixed estimate for a given property, better accounting for the high uncertainty inherent in rough models like this one.
It is also worth noting that Guesstimate runs a new Monte Carlo simulation each time the page is refreshed, meaning the numbers below may be slightly higher or lower than what you see on the model.
According to this model, the average Effective Self-Help article will produce between 28 and 580 hours of additional high-impact work, with a mean estimate of 180 hours of productive work added. Please take a look at the model and the comments attached to each value to see how I produced these figures.
Looking at the sensitivity analysis for the above estimate, some key (mean average) figures from the model include:
As is evident from the wide range in hours of high-impact work added by each article, there is a lot of uncertainty in these estimates. This is a rough model and I have generally only briefly reviewed relevant evidence to produce a relatively informed range of estimates.
With more time and detailed consideration, these numbers could change significantly. The numbers are also based on my initial model of publishing articles and so do not account for how different and combined methods of presentation may affect the project’s success.
Given these points, I believe the most useful takeaway here is simply that the Guesstimate model demonstrates that ESH could be very high-impact if things go well with the project. At the most optimistic end of the estimates, each article could produce an additional hour of high-impact work for $4.5, thereby producing roughly six or more hours of high-impact work for every hour that could have been paid for directly.
Better self-help advice may also be a cheap way to incrementally improve global wellbeing if it were to reach a sufficiently large audience. This would plausibly occur through the same rough mechanisms as discussed above, combining incremental improvements for lots of people with large improvements for a small minority.
As with the previous theory of change, I have made a Guesstimate model to roughly quantify the potential value Effective Self-Help could produce. This model estimates the Quality-Adjusted Life Years (QALYs) that the average ESH article may add by cultivating a large audience among the general public and is therefore based on a larger readership than ESH currently has.
Some parameters in this model are the same as the previous one but with values adjusted to reflect potential differences in engagement between an EA/ rationalist audience and one comprised of the general public. Again, please take a look at the model and the comments attached to it to see the reasoning/ evidence behind the different estimates included.
Based on the model, the average Effective Self-Help article will add anywhere from 0.14 to 25 QALYs, with a mean score of 4.2 QALYs added per article. At the most optimistic end, this would result in a cost per QALY of $190[2]. Similarly to the previous model, the very large range in potential QALYs added makes it foolish to draw firm conclusions from these numbers.
At this stage, it remains plausible that with a large audience ESH could either be disappointingly ineffective at producing significant wellbeing improvement or impressively effective. Much of this current variance appears rooted in the eventual size of the audience (e.g. 5k vs. 50k) and the QALYs added per recommended change (which appeared highly uncertain from my very brief evidence review).
As a final route to impact, a large and successful self-help website based on the principles of effective altruism and rationality could plausibly attract many good new people to these communities. Self-help is a very popular and widely-applicable topic, giving Effective Self-Help a huge potential audience.
80,000 Hours provides a useful example of a project that offers advice that is relevant and useful to the general public, through which people then develop long-standing engagement in a community. Outside of personal contacts, 80,000 Hours has been the most common way people have first heard about EA since 2018. This suggests that providing people with a service that is beneficial to them for non-EA reasons, but underpinned by key ideas of effective altruism, may be a particularly effective way of engaging them with EA ideas.
As discussed previously, ESH would place a high bar on evidence quality, finding the most effective solutions, and using numerical estimates when uncertain - all key parts of the quantitative approach that underpins much of effective altruism. In this way, the project could serve to sell the mindset of EA to people while they learn about topics they find more engaging and relevant.
With smart links and nudges, Effective Self-Help can direct its audience towards the effective altruism rest of EA (such as by frequently linking to LessWrong and EA Forum articles).
Needless to say, I am enthusiastic about the potential impact of this project. However, this would not be a reasonable evaluation without at least a reasonable attempt at exploring why ESH might fail. The following section is split between downside risks (ways in which ESH could be counterfactually harmful) and arguments for ineffectiveness (reasons why ESH may be of low impact/ value).
Some of the arguments below appear important and could significantly compromise the value of the project. However, by taking reasonable steps to mitigate against them, these issues also seem solvable. Better arguments against the value of Effective Self-Help, whether in the comments or privately, are encouraged!
Without sufficient care, ESH could provide advice that produces more harm than good. Assuming a reasonable minimum standard of research, I think it’s unlikely that ESH would provide advice that is directly harmful. More realistically, ESH could be of net harm in one of two ways:
Mitigating downside risks:
One way ESH could be impactful is by introducing new people to EA ideas, some of whom may then become actively involved in the EA community/ work in high-impact areas. However, ESH could be counterfactually harmful if it serves as a significantly worse introduction to EA than people may have otherwise received.
Specifically, Effective Self-Help could either:
Mitigating downside risks:
In researching the first two articles, I have found this to be somewhat, but not fatally, true. There are often only a handful of studies on the effectiveness of certain practices that meet basic thresholds (e.g. the inclusion of a control group). That said, there has generally been enough information to make a reasonable estimate of effectiveness so far.
This problem can also likely be further mitigated by continuing to improve methods of research and comparison, particularly by finding a consistent, and ideally relatively objective, way of including more general forms of evidence (e.g. expert opinion; survey data; etc.).
This seems quite plausible to me but I haven’t yet tested this. Engagement for our first post was higher than expected and suggestive of a large potential audience, but what appeals to LessWrong and EA Forum readers is likely to be quite different to what appeals to a more general audience. Producing content in more engaging formats (audio, video, etc.) may also help to counter this.
If this were a significant issue, it’s unlikely that ESH would still be cost-effective as an intervention targeted at improving the wellbeing of the general public (Theory of Change #2) as it would struggle to generate a sufficiently large audience. However, this would be unlikely to affect the project’s benefits to people in EA and similar communities, given the focus on practicality and evidence already apparent in these groups (Theory of Change #1).
The project could also still be highly valuable if the subset of the general public who do not prefer narrative-based articles are also those most likely to engage with EA, which seems a reasonable assumption. In this way, ESH may only cultivate a modest audience amongst the general public but still drive significant new engagement with EA through this smaller audience (Theory of Change #3).
Put another way, prioritising interventions would be of little value because the large majority of effectiveness is determined by the circumstances, background, and behaviours of each reader. This seems relatively plausible in a milder form (individual differences have a significant, but not overwhelming effect) but my intuition is that it is unlikely that individual differences are large enough to make generic differences in effectiveness unimportant.
Providing multiple top recommendations and encouraging experimentation so people try multiple solutions both seem like somewhat effective solutions to this. Some form of a screening quiz for prioritising recommendations based on the individual could also be valuable in addressing this.
This seems likely to me but also not much of a concern. With good, evidence-based recommendations to provide, Effective Self-Help could pivot in multiple directions in finding the best way to disseminate this information. Classes, podcast episodes, ebooks, and Youtube videos all seem like possibly useful and achievable formats to repackage information into.
This appears quite likely but only applies to Effective Self-Help as a means of improving global wellbeing (Theory of Change #2).
Self-help advice targeted at the general public is likely to have an audience that mostly comes from developed countries. As a general principle, interventions targeted at larger and more neglected groups (e.g. the global poor, farmed animals, or future generations) are likely to be more cost-effective, often by a significant margin. This could make ESH counterfactually harmful if it diverts work and/or funds away from more effective causes.
However, Effective Self-Help could still be highly worthwhile and cost-effective by focusing on improving productivity in the EA community (Theory of Change #1) and attracting a significant number of new people to EA (Theory of Change #3). In doing so, ESH could increase the amount of valuable work done to serve larger, more neglected groups/ causes and the number/quality of people working in these areas.
I will be publishing an article on the most effective ways to purchase productivity next week, collating the recommendations made in many articles posted to LessWrong, the EA Forum and elsewhere.
The rest of the ESH team are currently conducting research into:
If you'd like to stay up to date with our research, you can sign up for our monthly newsletter here.
We're currently trying to assess the value of our research up to now, and to canvass for feedback on how this work could be most useful going forward. If you could fill out this 5-minute survey about the project, this would be of great help.
Thanks for reading!