Hide table of contents

Loneliness Is a Big Problem

On Facebook, my friend Tyler writes:

Lately, I've been having an alarming amount of conversations arise about the burdens of loneliness, alienation, rootlessness, and a lack of belonging that many of my peers feel, especially in the Bay Area. I feel it too. Everyone has a gazillion friends and events to attend. But there's a palpable lack of social fabric. I worry that this atomization is becoming a world-wide phenomenon – that we might be some of the first generations without the sort of community that it's in human nature to rely on.

And that the result is a worsening epidemic of mental illness...

Without the framework of a uniting religion, ethnicity, or purpose, it's hard to get people to truly commit to a given community. Especially when it's so easy to swipe left and opt for things that offer the fleeting feeling of community without being the real thing: the parties, the once-a-month lecture series, the Facebook threads, the workshops, the New Age ceremonies. We often use these as "community porn" – they're easier than the real thing and they satisfy enough of the craving. But they don't make you whole.

I've had some thoughts about experiments to try. But then I think about how hard it is (especially in this geographic area) to get people to show up to something on at least a weekly basis. Even if it's for something really great. I see many great attempts at community slowly peter out.

Young people are lonely. Old people are lonely. Loneliness is bad for your health. It's bad for society's health.

With EA's recent focus on happiness and mental health, maybe loneliness is something we should consider working on.

Having a smartphone that keeps you entertained all day, and enough money to live by yourself, might sound like first world problems. But they are likely contributors to loneliness. And as developing countries get richer, they'll start having first world problems too. So I think addressing loneliness could be very high-leverage for the world.

People are starting businesses to address loneliness: you can pay someone to call you periodically or take you for a walk. But I'd argue these services are a band-aid in the same sense that parties, workshops, and ceremonies are. They don't solve the underlying problem: You're still alone by default instead of together by default.

Roommates Could Be a Great Solution

Sociologists think there are three conditions necessary for making friends: proximity; repeated, unplanned interactions; and a setting that encourages people to let their guard down and confide in each other. These conditions tend to be present during college for many people, but not afterwards.

Why do people find it easier to make friends in college? Maybe it's because college students don't usually live alone.

Going to events doesn't work because (a) you don't typically get repeated interactions with the same person and (b) events take place at a scheduled time. Which may or may not be a time you're feeling lonely.

If you have a lot of roommates, all you have to do is step outside your room and find someone to chat with. No transportation CO2 emissions needed. But more important, you know your roommates are always gonna be around.

But I Already Have Roommates

Even if you already have roommates, I think there's a good chance your roommate situation is under-optimized. Given that you spend so much time with them, there's a lot of value in living with people you really connect with. (Finding great coworkers makes sense for similar reasons.)

The layout of your house and the number of roommates you have can also make a big difference. I used to have friends living in a 4-bedroom place where all the bedrooms opened directly into a single large common area. If anyone else was outside their room, you'd immediately know it and have an opportunity for interaction. Later I lived in an 8-bedroom place which felt far lonelier, even with every room occupied. The house was laid out so it was easy to go about your day without ever running into a fellow roommate. I also lived in a house with over 50 bedrooms for a while, which was wild & a lot of fun.

But I Don't Want Roommates

One reason you might not want roommates is because you're worried you might have conflicting preferences for what living together should be like. For example, my philosophy towards dirty dishes is to let them pile up on the counter and periodically stuff them all in the dishwasher, to be as time-efficient as possible. Surprisingly, some people dislike this approach.

RoomieMatch.com is a website which tries to solve the roommate compatibility problem. You create a profile by answering questions about dishes, food in the fridge, housecleaning, social events, noise, overnight guests, shared household items, walking around in your underwear, TV, etc. In addition, there are questions to help predict how you well you will connect as people.

You Could Make a Lot of Money

RoomieMatch has two search options: free and cheap. Cheap costs $20/year.

The problem with RoomieMatch is they're leaving a massive amount of money on the table.

A few years ago, a friend of mine was jobless & struggling financially. He was living in a 4-bedroom house at the time, and he was the primary contact with the landlord. My friend took responsibility for vetting folks from Craigslist in order to fill the remaining rooms in the house. He found that folks from Craigslist were willing to pay enough rent for the remaining 3 rooms that he was able to live rent-free until he found a job.

I acknowledge this is murky ethical territory, and I'm not condoning my friend's actions. (I don't believe anyone ever found out or got upset, for whatever that's worth.) The point I'm trying to make is that property management is way more lucrative than roommate matching. RoomieMatch makes $20 per user per year at best. My friend was making $100+ per user per month.

What I'm suggesting is that you take the full-stack startup playbook which has been successful in Silicon Valley recently, and apply it to online roommate matching + property management.

The extreme full-stack approach is to own your own properties. Apparently the US has a surplus of big houses right now.

There are already players in this space such as Roam which are proving that people will pay for community. (As if people paying extra to live in hip cities like SF & NYC didn't prove that already. BTW, I found that the awesome community at the Athena Hotel more than made up for the fact that it's in a non-hip city.) Anyway, I think existing players are mostly pursuing the extreme full-stack option. I actually think this is the wrong play. You want to be a marketplace, like Airbnb (valued at over $30 billion). The more people who are using your tool, the finer-grained roommate matching services you can provide. It's hard to achieve massive scale if you have to own every property. You want to be playing matchmaker for individuals with common interests who all happen to be looking for rooms around the same time, plus landlords with empty houses. Maybe you'll want to undercut RoomieMatch, and provide free matching services for people who live in their properties, in order to achieve the necessary scale. (RoomieMatch's existing scale is impressive by the way--I quickly got 100+ active, vetted matches in a midsize US city when I tried the tool. You might want to just buy it if you can.)

So instead of buying properties, maybe you just want to contact people selling large homes & see if you can convince them to let you manage their property.

Note that this is a good company to start if a recession happens, since people who currently live alone will be thinking about how to save on rent.

This Could Be Really Great

Most roommate search tools, like Craigslist, don't make it easy to figure out if a future roommate is someone you'd actually want to live with. Imagine reaching a scale where you could match people based on factors like:

  • They love to play board games, or pool, or Super Smash Bros.

  • They want a garden or compost pile in their backyard.

  • One has a pet, and the other likes animals but isn't yet ready to make a lifetime commitment.

  • They want a squat rack in the basement to save time & money going to the gym.

  • They want to continue partying like college students after graduation.

  • They want to be part of an intentional community devoted to mutual improvement and life optimization, or spirituality, or whatever.

  • They want to share childcare responsibilities.

  • They're all fans of the same sports team.

  • They enjoy reading and discussing the same genre of novels, or watching the same movies.

  • They're musicians looking for people to jam with.

  • They want to live near hiking trails and go on group hikes together.

  • They want to do independent study of the same topic.

  • They're trying to eat a healthier diet.

  • They just moved to a new city and want friends they can explore the city with.

  • They have the same unusual work schedule.

  • One needs a caretaker, the other wants to make extra money.

  • They like the idea of having a couch or two listed on CouchSurfing.

I also see opportunities to reduce friction in the current roommate matching process:

  • Automatically find times when everyone is available for a meet & greet video call.

  • Let people take virtual tours of the houses on offer to minimize driving.

  • No need to worry about breaking a lease if someone moves to a different house in your company's network. Let people try out a few communities & see what works for them. Use machine learning to improve your matching as you gather more data.

  • Provide external mediation in the event of roommate disputes, and have a reputation system to encourage good behavior.

You aren't providing housing as a service (like Airbnb), or companionship as a service (like the people-walking startup). You're providing community as a service. You could even organize mixers across your houses.


Technology has been blamed for the loneliness epidemic, but I think we can use technology to cure the loneliness epidemic as well.

I'm too busy thinking about AI safety to start any company which isn't related to AI. But I think this is a product the world needs, and I want you to build it and donate the money you make to effective charities if it sounds exciting to you.

I apologize if you found the tone of this post overly sales-y. My goal was to light a spark in the right person. (Feel free to steal phrases from this post when pitching investors!)

Some folks in the community might be a little underwhelmed by this idea, if they've already been living together in group houses. The thing is, finding roommates by connecting based on mutual interests via the internet is still kind of weird in the eyes of the general public. As Paul Graham put it: "Live in the future, then build what's missing." The existence of so many lonely people proves that this option is still missing for most people.

Anyway, if you're interested in building/investing in this, please comment below, or send me a private message via my user page with the country you're in and I'll put you in contact with others who message me.

Cross-posted to LessWrong





More posts like this

Sorted by Click to highlight new comments since:

Loneliness may indeed be rising, but we don't yet have good evidence that that's the case:

"It's easy to find claims that loneliness is rising (for example, here's a recent Wall Street Journal article on that theme). But last summer the Social Capital Project run by the Joint Economic Committee of the US Congress published "All the Lonely Americans?" (August 22, 2018) and found little evidence of such an increase. The report cites a broad array of claims and evidence, which you can check out for yourself. But here's a quick overview of some main points (with citations omitted for readability):

There are a few different but related questions that tend to get lumped into one general story about whether loneliness is on the rise in America, in part because of a lack of good data, and occasionally because of a failure to distinguish the two often distinct lines of psychological and sociological research. One question is whether Americans are increasingly alone (that is, have fewer social contacts, or have less social interaction). This question, which sociologists tend to study, is about objectively observable social networks or relationship characteristics. It is distinguishable from the second question, regarding the subjective experience of loneliness. This latter question—whether Americans are increasingly experiencing loneliness (`perceived social isolation')—has typically been the research purview of psychologists.

Correlations are lower than we might expect between the most common measures of loneliness and objective measures of social network characteristics, so these two questions are substantially though not wholly distinct from each other. ... However, it is not at all clear that loneliness has increased over the last several decades.

In his 2011 book, Still Connected: Family and Friends in America Since 1970, sociologist Claude Fischer puts a fine point on this question: `For all the interest in loneliness, there appears to be little national survey data that would permit us to draw trends.'

We looked for the strongest support for the claim that loneliness has risen, and the best we could find comes from polls by FGI. Between 1994 and 2004, the FGI polls indicate that the share of adults saying loneliness was a problem for them rose from roughly 25 percent to 30 percent. It is unclear, however, whether this five-point difference reflects a real shift or arises from chance differences in the people sampled in each year or in survey administration.

The remaining evidence suggests flat trends. ... The claim that loneliness has doubled—or even increased—since the 1980s (let alone the late 1960s) is simply unwarranted. ... It is entirely possible that loneliness has increased over time, but the available evidence does not appear to support that claim. It is just as possible that loneliness has stayed the same or even declined."

It looks like this report is from 2018, and doesn't incorporate the 2019 YouGov research I linked. (I doubt pre-2004 data will give us insight into modern loneliness. Facebook and Twitter didn't exist back then, for instance.) This bit is interesting though:

More recently, some media outlets have misinterpreted the results of a 2018 Cigna survey to argue that loneliness has increased. The survey indicated that loneliness was higher for younger Americans than for older ones. A mistaken interpretation of this finding would be that older Americans were less likely to be lonely when they were younger than today's younger Americans are. This interprets life-course changes in loneliness as reflecting a change over time for Americans whatever their stage in the life course. While USA Today reported the age-based results as "surprising," the research on the relationship between age and loneliness suggests that the "[p]revalence and intensity of lonely feelings are greater in adolescence and young adulthood (i.e., 16-25 years of age)," decline with age, and then increase again in the very old.33 The Cigna survey does not support the claim that loneliness has increased over time, nor is the increased loneliness of adolescents a new revelation.

It's not clear to me how to reconcile this with e.g. the research YouGov cites to attribute loneliness among current youth to social media use. I guess a natural first step would be to see whether the magnitude of historical effects in the Handbook of Individual Differences in Social Behavior can explain what YouGov saw. I think you'd have to analyze data carefully to figure out if it supports the hypothesis "young people just tend to be lonelier" or the hypothesis "social ties get weaker with every passing generation + elderly people get lonely as their friends die".

In any case, I think loneliness could be a problem worth tackling even if it isn't rising. (And you will notice I didn't technically claim it was rising :P) The point is also somewhat moot as only one person expressed interest as a result of me posting here.

Fair enough I haven't looked at the YouGov report.

I responding to the thrust of Tyler's quote at the top.

I doubt pre-2004 data will give us insight into modern loneliness. Facebook and Twitter didn't exist back then, for instance.

That data is especially precious because you need a 'before' measurement to see whether social media coincides with any change or loneliness staying the same as before!

But I agree many problems aren't increasing but are still well worth addressing!

I'm too busy thinking about AI safety to start any company which isn't related to AI. But I think this is a product the world needs, and I want you to build it and donate the money you make to effective charities if it sounds exciting to you.

Probably you'd make a big contribution to AI safety by becoming a billionaire who cares about AI safety?

I have a lot more ideas than I know what to do with. So I try to prioritize ruthlessly. I feel like I've got a comparative advantage working on AI stuff and a comparative disadvantage starting a company like this one. I'm experimenting with posting some of my ideas to the EA Forum to see if they can be useful to other people, e.g. folks who wanted to get a job at an EA organization but weren't successful.

Interesting ideas... the main thing I am struggeling with is the inherent danger of seeing this as for-profit business... You will start to optimize for your metrics and want hockeypuck growth figures to satisfy shareholders in the attempt to become the next WeWork unicorn. Quicker than you can say "Utopia" the whole thing will turn into a creepy shit show tracking your every move trying to upsell you on the next great product.

That's maybe a little harsh but I think you get my idea... there are some interesting aspects in your proposal but without a really well designed incentive system the whole thing will go up in flames. I warned you...

Facebook and Google have an incentive to track their users because they sell targeted advertising. The user isn't the customer, they are the product. This is an atypical business model.

One thing about the real estate business is because so much money is changing hands, there's a big incentive to cut out the middleman. (Winning Through Intimidation is a fascinating book about this.) I would highly recommend you avoid actions which run the slightest risk of pissing your customers off, lest they cut a deal with the property owner directly. Airbnb will ban anyone who exchanges money outside their platform, but that's less of a threat here because people don't change homes frequently. With the amount of money you're making per customer, you should be able to afford an army of customer service people in order to provide a high-touch customer experience.

There are a few reasons I think for-profit is generally preferable to non-profit when possible:

  • It's easier to achieve scale as a for-profit.
  • For-profit businesses are accountable to their customers. They usually only stay in business if customers are satisfied with the service they provide. Non-profits are accountable to their donors. The impressions of donors correlate imperfectly with the extent to which real needs are being served.
  • First worlders usually aren't poor and don't need charity.
  • You can donate the money you make to effective charities.
Facebook and Google have an incentive to track their users because they sell targeted advertising.

Even without ads they would have a very strong reason for tracking: trying to make the product better. Things you do when using Facebook are all fed into a model trying to predict what you like to interact with, so they can prioritize among the enormous number of things they could be showing you.

Thanks for your reply.

I disagree with your statement that

For-profit businesses are accountable to their customers. They usually only stay in business if customers are satisfied with the service they provide. Non-profits are accountable to their donors. The impressions of donors correlate imperfectly with the extent to which real needs are being served.

This is per definition false. For-profit businesses are accountable to their shareholders which can but does not have to mean that they want to act accountable towards their customers for strategic reasons. Strategic reasons can also lead to irresponsible behavior towards customers. You make a good example with facebook and google.

In a similar vein, non-profits are not accountable to their donors but to their charter and members. However, non-profits may want to act accountable towards donors for strategic reasons. For example, if a non-profit is not tax-exempt it can act just as regular company.

Moreover, there are organization types between simple for-profit and standard non-profits, e.g., public benefit corporations [1] or cooperatives [2].

Having said that, I have nothing against well-calibrated for-profit companies but I think my point still stands that anyone who may follow your proposal and has a vested interest in making the world a better place for everyone (from a tentatively impartial and welfarist perspective) should really think about the incentive structure they get themselves into. At least investigate a little bit beyond the standard playbook of neoliberal start up 101.

1:https://en.wikipedia.org/wiki/Public-benefit_corporation 2: https://en.wikipedia.org/wiki/Cooperative

See also: Bungalow, HubHaus, Common.

I wish someone would do shared living for people who want private living units (not just bedrooms) + shared common spaces, e.g. multiple houses on one block and a shared common space on that block. Makes co-living work for families or people who want a bit more privacy.

Nice post. Your friend wrote that

"Lately, I've been having an alarming amount of conversations arise about the burdens of loneliness, alienation, rootlessness, and a lack of belonging that many of my peers feel, especially in the Bay Area",

which made me wonder if there are geographical areas or cities in the western world that have a particularly high level of connection and community? Maybe those cities could be studied and promising cultural characteristics be spread to other cities and communities?

Although I can't comment on the sense of community felt by the local residents, I observed and to some extent experienced this in Spain. I'd say the key was the combination of high urban density and availability of shared spaces. Another factor could be the low price of eating/drinking outside the home - - I'd say this facilitates socializing since it's easier to say "Let's meet at X at 9pm [Spanish people have dinner very late!]" rather than having to prepare your house to host guests. There's a joke that you only go into a Spanish person's flat for a wake (which is an exaggeration, but somewhat based on truth).

Someone also mentioned to me that it is culturally more normal in Europe for people to socialize after work, likely due to some of the factors I mentioned. Cal Newport recently implied that this may have been the case in other countries pre-television. It's also socially acceptable to take children to most events, even late into the evening.

Unfortunately, these aren't really cultural characteristics, as I'd say it's fundamentally based in the high urban density.

There are already at least three companies in this space: RoomieMatch, Roomi, and Roomster. I wonder why nobody I know uses them, but dating apps are very popular?

It seems to me that the triangulation, trust, and transfer problems in roommate matching that go beyond what OKCupid has to deal with:

  • There are more than two people involved, and the difficulty of finding communal compatibility complexifies geometrically with the number of roommates.
  • By the same token, people moving in and out happens more frequently with larger numbers of room mates, often with short notice, making it hard to keep a stable equilibrium of preferences.
  • Imagine if it was easy to "date your future housemates," perhaps by living together for a month. It's already emotionally painful for people to deal with or inflict rejection in one-on-one dating. Imagine being the "odd man out" in this situation. That sounds like a recipe for really uncomfortable social dynamics.
  • People who rent because they can't afford their own place probably can't afford a high-touch service. People who have more money could buy their own place and interview enough room mates to make sure everyone is a good fit with them personally.
  • Land lords often influence or even entirely control the process of finding new room mates. There are also laws around evictions that make it very difficult to kick somebody out if its not working for others, whereas there are no legal barriers to breaking up with someone you're dating if there's no marriage and no kids.
  • There's a much higher effort and commitment barrier required to move than to go on a date.
  • This is speculative, but OKCupid's success may stem from capitalizing on a cultural institution that makes romantic love feel of vast importance. By contrast, finding an ideal group of room mates doesn't have the same cultural importance: we still dream of having our own place by ourselves or with our own biological family. To have comparable success, such a service would need to create a new dream. Even if that's your dream, is it the dream of your housemates?
  • Similarly, the service OKCupid provides may be less in matching people with compatible characteristics, and more in identifying an abundance of single people and getting them hyped to go on a date. The purpose of the "matching" is to trick you into building up anticipation, not to ensure a really good fit (after all, if it did that too well, people wouldn't come back for more!). Instinct, hormones, and love do most of the work of making people stick together in the end.
  • When people do try and start intentional group houses, they're often organized around a shared social movement, which already have word-of-mouth and social media channels where people can learn about these opportunities for free.

I think a company would do better to work on solving one or more of these problems.

Crossposted from the LW forum

I had high hopes for this post...and was disappointed. I don't think getting roommates or changing roommates is a cure for loneliness for the majority of adults. If I wrote a similar post, instead, I'd discuss various forms of **meditation**. Additionally, I'd mention how to find friends in new places (not necessarily roommates as I don't base my room/housing on the who--but the where).

Even then, since most people are looking for a life partner, the better way to reduce loneliness is to meet potential suitors. So then one more prospective route is how to meet, generally, single people. Colloquially, at least in the US, this means going to bars or "day gaming."

I believe that the feeling of loneliness probably is one big contributor to mental health issues and I like your idea of tackling it pragmatically/for-profit.

My gut feeling is that this won‘t be used and people are happy enough with the Craigslist solution. Anectdotally, my roommates (intelligent people) thought it was a joke when I proposed that I would design a questionnaire for people applying to live with us. Another platform, OkCupid, tries to offer meaningful matching scores for romantic partners and it seems to be rather fringe, at least in Germany.

OKCupid was huge before Tinder came along in the US. And as I mentioned, RoomieMatch is already pretty big. That said, it's possible there wouldn't be as much of a market for this in Germany. One approach is to start in a city with lots of early adopters who like trying weird new stuff (San Francisco is traditional) and gradually expand as the product concept is normalized. But sometimes things don't go much beyond early adopters.

Well, how about starting "Tinder for sparerooms"?

It could work. However, Tinder works well because people can quickly guess whether they want to date someone based on physical attraction. I don't think there is a single easy-to-evaluate factor which predicts roommate compatibility. Also, moving in with someone is a bigger commitment than going on a date with them.

Curated and popular this week
Relevant opportunities