Comment Permalink

To answer your specific question, there are a bunch of potential alternatives.

You can use a Roth IRA to have access at least to your contributions without penalty, and have tax deferral and tax free earnings.

You should probably put enough to get the full match into your 401k no matter what, as long as you expect to become vested for the company contribution, since taking that out early and paying penalties is still a win versus forgoing the free money your employer is offering.

You can invest in plain old retail investment accounts. You will pay tax every year, but it's possible to minimize the tax hit with tax-efficient investing strategies (low turnover, paying attention to specific gains and losses -- or use tax efficient mutual funds).

You can get some of the tax advantages of retirement savings without the restriction on when you can use the money using permanent life insurance as a savings vehicle. This is very popular with wealthy clients who want to save more than roth and 401k's will allow and have high marginal tax rates. Drawback: To the extent you don't otherwise need life insurance, or fit well with this kind of plan, it may not make sense versus eating the taxes in a retail (non-qualified) investment account. Look at your individual case carefully. (Disclaimer: I make money from selling life insurance.).

The possible implied alternative of not saving at all seems foolish unless you have a very good use for the money now. if living cheap is so easy, why not do it now? you can always spend your money later. If you want to start startups, having some money saved for that is a good idea. If you want to give away a lot, the world will not lose very much by you giving it away plus reasonable investment earnings later versus doing it right now. Saving money retains flexibility. There is no reason it must be permanently earmarked for retirement.

As a planner, I would say that most people I run into, if they save enough, are saving too much in retirement focused vehicles, and not enough elsewhere. I see a lot of people in their 40s and 50s who have 5-15x salary in their 401k/403b, but a barely sufficient (or not even) emergency fund, and essentially zero other assets they can use without penalty before age 59.5. My general recommendation is to have a good portion of your savings outside the retirement 59.5 gate if possible without losing matches or giving up too much tax efficiency.

See in context

12 Saving for the long term

by Adam Zerner

24th Feb 2015

5 min read

12

I'm 22 years old, just got a job, and have the option of putting money in a 401k. More generally, I just started making money and need to think about how I'm going to invest and save it.

As far as long-term/retirement savings goes, the way I see it is that my goal is to ensure that I have a sufficient standard of living when I'm "old" (70-80). I see a few ways that this can happen:

There is enough wealth creation and distribution by then such that I pretty much won't have to do anything. One way this could happen is if there was a singularity. I'm no expert on this topic, but the experts seem to be pretty confident that it'll happen by the time I retire.

Median optimistic year (10% likelihood): 2022
Median realistic year (50% likelihood): 2040
Median pessimistic year (90% likelihood): 2075
- http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-2.html

And even if they're wrong and there's no singularity, it still seems to be very likely that there will be immense wealth creation in the next 60 or so years, and I'm sure that there'll be a fair amount of distribution as well, such that the poorest people will probably have reasonably comfortable lives. I'm a believer in Kurweil's Law of Accelerating Returns, but even if you project linear growth, there'd still be immense growth.

Given all of this, I find thinking that "wealth creation + distribution over the next 60 years -> sufficient standard of living for everyone" is a rather likely scenario. But my logic here is very "outside view-y" - I don't "really understand" the component steps and their associated likelihoods, so my confidence is limited.
I start a startup, make a lot of money, and it lasts until retirement. I think that starting a startup and using the money to do good is the way for me to maximize the positive impact I have on the world, as well as my own happiness, and so I plan on working relentlessly until that happens. Ie. I'm going to continue to try, no matter how many times I fail. I may need to take some time to work in order to save up money and/or develop skills though.

Anyway, I think that there is a pretty good chance that I succeed, in, say the next 20 years. I never thought hard enough about it to put a number on it, but I'll try it here.

Say that I get 10 tries to start a startup in the next 20 years (I know that some take longer than 2 years to fail, but 2 years is the average, and it often takes shorter than 2 years to fail). At a 50% chance of success, that's a >99.9% chance that at least one of them succeeds (1-.5^10). I know 50% might seem high, but I think that my rationality skills, domain knowledge (eventually) and experience (eventually) give me an edge. Even at a 10% chance of success, I have about a 65% (1-.9^10) chance at succeeding in one of those 10 tries, and I think that 10% chance of success is very conservative.

Things I may be underestimating: the chances that I judge something else (earning to give? AI research? less altruistic? a girl/family?) to be a better use of my time. Changes in the economy that make success a lot less likely.

Anyway, there seems to be a high likelihood that I continue to start startups until I succeed, and there seems to be a high likelihood that I will succeed by the time I retire, in which case I should have enough money to ensure that I have a sufficient standard of living for the rest of my life.
I spend my life trying and failing at startups, not saving any money, but I develop enough marketable skills along the way and I continue to work well past normal retirement age (assuming I keep myself in good physical and mental condition, and assuming that 1. hasn't happened). I'm not one who wants to stop working.
I work a normal-ish job, have a normal retirement plan, and save enough to retire at a normal age.

The point I want to make in this article is that 1, 2, 3 seem way more likely than 4. Which makes me think that long-term saving might not actually be such a good idea.

The real question is "what are my alternatives to retirement saving and why are they better than retirement saving?". The main alternative is to live off of my savings while starting startups. Essentially to treat my money as runway, and use it to maximize the amount of time I spend working towards my (instrumental) goal of starting a successful startup. Ie. money that I would otherwise put towards retirement could be used to increase the amount of time I spend working on startups.

For the record:

I'm frugal and conservative (hard to believe... I know).
I know that these are unpopular thoughts. It's what my intuition says (a part of my intuition anyway), but I'm not too confident. I need to achieve a higher level of confidence before doing anything drastic, so I'm working to obtain more information and think it through some more.
I don't plan on starting a startup any time too soon. I probably need to spend at least a few years developing my skills first. So right now I'm just learning and saving money.
The craziest thing I would do is a) put my money in an index fund instead of some sort of retirement account, forgoing the tax benefits of a retirement account. And b) keeping a rather short runway. I'd probably work towards the goal of starting a startup as long as I have, say 6 months living expenses saved up.
I know this is a bit of a weird thing to post on LW, but these aren't the kinds of arguments that normal people will take seriously ("I'm not going to save for retirement because there'll be a singularity. Instead I'm going to work towards reducing existential risk." That might be the kind of thing that actually get's you thrown into a mental hospital. I'm only partially joking). And I really need other people's perspectives. I judge that the benefits that other perspectives will bring me will outweigh the weirdness of posting this and any costs that come with people tracing this article to me.

Thoughts?

Personal Blog

12

New Comment

38 comments, sorted by

top scoring

Click to highlight new comments since: Today at 6:55 AM

[-]Viliam_Bur10y150

Getting some experience and developing your skills before you start your first startup seems like a good idea.

How difficult it will be in practice, that will depend on your savings, on a support you can receive from e.g. your family, whether you will be able to make more money than you need to survive so you can save some, and whether you will be able to learn new things in your free time.

In my experience, it works like this: When you get a new job, at the beginning you learn a lot. After a few months it gets repetitive, and after a year you mostly do the old stuff over and over again. You probably get more work, but it is the same type of work. Only rarely (in my experience) is there an opportunity to move within the company. If there is a new project, the company will usually hire new people for the new project, and you stay with the stuff you were hired for. It may gradually become more easy, but you will not progress. (If you are strategic, it may be good to keep doing the easy work for the safe income, and learn new stuff in your free time. Or maybe even during your work time, if no one is watching you.)

After more years, you will become responsible for maintenance of all systems you have ever touched while working in the company. You will have to remember thousand little details about these systems, which is mostly an untransferable knowledge that has zero value outside of the current job. Just because you have more work now, it does not mean that your salary will necessarily increase. When you are at such dead end, if you want to optimize for skills, you should change jobs.

(Meanwhile, there are new younger programmers coming from universities. All the new stuff that you would like to know but have no time to learn, they have learned at school.)

Here comes the question of risk aversion, and how good is your safety net; how much would you lose if you quit this job, but can't find a new one soon enough. In theory, you should find the new job first, then quit the old one. In practice, it may be difficult to do job interviews, to prepare for them, and to study new technologies, while having a full-time job. Also, the new job may fail for unpredictable reasons, or may turn out to be very different from what they told you at the interview.

Poor people can easily get stuck at a job they hate, that is exhausting and pays relatively little. Because of the small pay, they do not have safety net. Because the job takes all their energy, they do not have the energy and time to learn new stuff and do job interviews. If the job requires them to learn stuff that has zero market value, the longer they stay, the harder it will be for them to get another job. -- In theory, programmers should not be that poor. In practice, sometimes your expenses grow faster than your income; e.g. at the moment you move away from your parents. (Or when you want to start a family, and suddenly there are three people supposed to live from one income for a few months; and later it becomes one and half person per income.) Or some unexpected expense happens.

The work that is available right now may not require the skill you want to develop now. Maybe you choose an exotic skill that no one needs. Maybe all companies that need people with some skill insist on employing only people who already have a few years of experience with that skill. Some skills you may have to develop at your free time, at least to the level where you are able to make a working prototype.

Here is some advice, but it may strongly depend on specific country and area of specialization:

After a year, unless you are still learning something new (something that has a value outside your current company), quit the job. Or maybe, right before you quit, ask the company whether you could become a lead developer of at least a small team: that would look impressive on your CV, and would teach you skills you will later need for your startup.
Try to keep a safety net; enough money to survive for six months, preferably a year. Then you will have more freedom changing jobs. When doing something expensive, such as buying a new appartment or starting a family, try to arrange it so that some financial reserve remains, instead of spending all you currently have. Always imagine that tomorrow your job could become insufferable, and you would have to leave to preserve your sanity.
Avoid overtime. Your sleep and free time are critical for your health and professional development. Spend some free time learning new stuff. Learning something new at home, making a prototype and bringing it to the interview, may allow you to get a job with a new technology that wasn't available to you at university and at previous jobs. Spend some free time with your friends, meeting new people, networking. Networking is great for getting new ideas, motivation, and even good jobs.

[-][anonymous]10y100

My experience (3 European countries):

Even when not learning anything specific about a job, future employers value X years of experience at doing Y metrics. So far it does not seem to have any diminishing marginal utility in their eyes, despite that it does have in real life. You learn everything you could ever learn about Microsoft CRM in 3 years but they still prefer someone with 7. OTOH somewhere between 10 and 20 years it may get weird for future employers if you did not do anything different nor get promoted.
There is often a lot of undertime i.e. screwing around on the Internet at work, because you were hired to bring a specific skillset inside a company which they did not have before and they don't always need that. Largely, Firm Theory suggests that hiring as opposed to subcontracting is a way to reduce transaction costs, like you don't have to negotiate about the price before every single task you give to an employee and we know it from Fukuyama that DISTRUST is a huge transaction cost. For this reason, I see constantly that businesspeople rather hire someone 40 hours a week than to have external consultants do it taking a random amount of time but often just averaging to 5-10 hours, because they don't trust externals.

I don't either. I know the shenanigans I was forced to do when I was a consultant and I constantly call them on this shit when I am now the customer and negotiate with them :) For example if they are ERP consultants like SAP, Oracle, NAV, they often do unpaid work like fixing bugs in the standard software because they cannot tell a customer that please pay €2000 for fixing a fault in a product I sold you. The customer will say do it for free and try to bill it to the software "manufacturer" of whom you are a reseller of. Customers often see software having the same kind of warranties as manufactured products, not legally, but ethically. So consultants work 2 days for free to fix a stupid bug, and then the next time you want a bunch of new fields and reports they will sell 2 days for "analyzing requirements". That is the shit I call them on :) Because I used to do the same.

Anyway, this INSOURCING due to distrust in subcontractors, consultants results in undertime i.e. bored and Redditing at work.
In a high-context culture, employers will often not set up clear rules to that. They will just wink wink let's pretend you are working here when I can guess you are not. The price is, that when you are overutilized, you do overtime without extra pay as a way to make up partially for the days when you just warmed a seat in the office. Nobody says this openly, but if you complain about overtime, they may be looking into monitoring your internet traffic the next time you are underutilized. And you don't want that probably.
Obviously, using undertime at work to learn new things and making prototypes is a good solution. Assume that they own it, but if it is mainly a learning project, let them own it, then the value is not in the product but in the experience.

[-]JoachimSchipper10y30

Although I've heard the advice to leave after a year, my experience has been different - after three years, I'm still learning a lot and I'm beginning to tackle the really hard problems. Basically, I find myself agreeing with Yossi Kreinin's reply to Patrick McKenzie's advice, at least so far. (Both links are very much worth reading.)

Of course, you do need to push for interesting assignments and space to learn. Also, be sure to pick a company that actually does something interesting in the first place - I work on embedded crypto devices for the government market, in a company that's young enough that there's still plenty of flexibility.

[-]Viliam_Bur10y20

I was thinking about the programmers unable to do the FizzBuzz test, and the interesting insight that Eliezer linked long ago -- that maybe it's not so many programmers who are extremely stupid; maybe it's just the fact that the non-extremely-stupid ones sooner or later get a job, while the extremely stupid ones remain in circulation for a long time, thus more recruiters have the misfortune of meeting them.

Maybe it also works the other way round. When a great IT company is hiring, the employees are happy to tell their friends, so the positions are filled quickly. When a crappy IT company is hiring, it takes them much longer to find someone, and meanwhile other programmers are quitting, so they have to keep hiring to replace their positions. Which means: if you apply for a random IT company that is hiring at the moment, you are not going to see an average company; you are most likely to see a crappy one. Of course, you can be lucky. But most people are not.

In a good company you can keep learning even after the one year. But those are rare.

A higher-level advice is to be strategic and find the good company. For this you want to maintain many contacts with programmers who share your values (who recognize the type of job you would like to have). But how do you get these contacts? Keep in touch with your university classmates (I failed to do this, because at university I had no idea how much will I care about this one day), or take a few crappy jobs and keep in touch with the former colleagues (who may change their jobs later and find something better). Alternatively, do something to become famous (e.g. contribute to open-source software, or write a popular programming blog) and then the people will contact you.

[-]Adam Zerner10y00

Very nice write-up! I've suspected that a lot of those things are true, but wasn't too sure, so your saying them increases my confidence. Especially the thing about avoiding overtime. I've brought it up before but most people seem to have the opinion that "you learn at work, so you may as well work late and keep learning (not taking into account the opportunity cost)".

What do you think about the chances that a good programmer with a few years' experience could easily find a new job? It seems pretty high to me. a) I here so much about the demand for developers. And b) in my job search it seemed that there was a huge demand for developers with a few years' experience. But again, this is outside view-y and I don't understand the components too well.

[-]Viliam_Bur10y70

most people seem to have the opinion that "you learn at work, so you may as well work late and keep learning

Depends. Now that you wrote that, I remembered the time when I was freshly out of university; it did work that way. Thanks for reminding me! Everything was new. I didn't have internet at home. The computer I had at work was more powerful than the computer I had at home. I even had a few books about programming at work. So when I stayed overtime, I learned a lot. I was a 100% nerd back then, zero social life, no hobbies beyond programming. Thus staying at work late was like staying at Disneyland.

The difference is, back then I was not supposed to do any specific work when I stayed overtime. As a total noob I was not expected to be highly productive; the time I had to do my work was measured in days, not hours. It was understood that learning is a part of my work, and that learning takes time. That is why I was free to learn.

That is completely different from when you are told to stay overtime. It usually means there are specific problems you are supposed to work on, which usually means: the same stuff as always, only more of it. Or that someone in management messed up something, so you have to do something again, just differently. Or that there is a bug in the application that no one can find and the customer is angry about it, so you have to find the bug. There is no time to learn.

Exploration / Exploitation -- it is okay to do overtime for the former, but when the overtime is required, it is the latter. Also, make sure the "exploration" is in the direction you want grow, instead of some dead end. (I have spent a few months studying a stupid product that was discontinued a year later.)

Also, being a 100% nerd is not the optimal strategy. For example, I learned a lot back then, but I still had laughably low salary, because I had no friends, didn't communicate with people, so I didn't know how much could I expect, and I didn't know how to sell myself. At that moment, picking the lowest hanging fruit of the social skills (such as calling a few programmers I knew and asking then whether there is a job opportunity where they work, and how much approximately could I ask) was financially more profitable than any increase of my technical skills. Skills and salary are only weakly correlated.

What do you think about the chances that a good programmer with a few years' experience could easily find a new job?

I think you can always get a programming job; the question is what are your salary expectations. I cannot tell you how high salary is realistic where you live.

At the very beginning, people will ask you how many years of experience you have, which technologies you know, and how many years of experience with their preferred technologies you have. After maybe five years, they will also ask you if you have an experience leading a team. Being able to show your own project that you did in your free time is probably a big advantage: it shows you are able to complete the project (as opposed to many people who only do the beginning that is most funny, and then abandon the project). I guess most people do not have such projects.

[-]michael_b10y100

(4) I work a normal-ish job, have a normal retirement plan, and save enough to retire at a normal age.

The point I want to make in this article is that 1, 2, 3 seem way more likely than 4. Which makes me think that long-term saving might not actually be such a good idea.

Are you sure? Take a look at http://firecalc.com/

You plug in your nest egg size and expected cost of living and it will trial it against historical market performance. That is, if you retired the day before the Great Depression and similar events, would your portfolio keep you alive long enough?

It's clear if you plan to live a reasonably frugal lifestyle working a normal job you can save up a substantial nest-egg in 10-20 years to float you indefinitely if you're only a little bit lucky, probably requiring a lot less luck than hoping you strike gold on a startup.

[-]Adam Zerner10y00

It's clear if you plan to live a reasonably frugal lifestyle working a normal job you can save up a substantial nest-egg in 10-20 years to float you indefinitely if you're only a little bit lucky, probably requiring a lot less luck than hoping you strike gold on a startup.

True. The way I see it, the benefit to that is the marginal increase in retirement safety. And the cost is the forgone opportunity of doing something important in those 10-20 years. What do you think of those trade offs? In particular, I'm currently thinking that there's already a really high chance that I can retire comfortably because of 1, 2 and 3.

[-]michael_b10y20

Permit me to ramble a bit.

So, I think I have a bias against startups for getting rich. I think it might be reasonable.

I'll relate my personal experience as well as the experiences of three close friends. Around eight years ago, four of us each joined different startups. I quit mine about 3 1/2 years in. They've remained at theirs. What's the current status?

me: I've learned that my equity in my startup, which I took at a substantial discount to my salary, is now worth low six figures.

friend 1: his startup has almost been acquired but the deal sunk in due dilligence. Not in danger of going out of business but not exactly striking gold either. At least not yet.

friend 2: recently had his startup acquired and he received a low six figure bonus on the condition he stay another two years to ease the transition. It seems like he was cut out of the upside somehow.

friend 3: nothing close to an exit in sight; the company isn't failing, but it doesn't seem like it's going to strike gold either. It's slowly growing.

The low six figures of equity I (now) have might be seen as a positive, but 8 years on and I still can't sell it, and I had to take low pay and do a ton of work to get it. This is better than most startups, but still not a success.

In the meantime after I quit my startup I joined an established company and made a high-normal salary + bonuses and am way ahead of the rest of my friends on net worth (we all basically started at zero).

Two of my friends are only finally now making normal job salaries, but they're basically burned out at their respective companies and can't bring themselves to quit because they signed deals where their equity evaporates if they walk away. This is probably the largest danger to joining a startup: what if it drags on forever, and it seems like you could get your payday if you keep hanging on for six more months and this goes on for years and years? Do you think you could just walk away? I think this is loss aversion bias at work.

Should people work at startups? Absolutely: the experience is outstanding. You will learn a lot, the energy is amazing, and there are seldom better opportunities to feel that you're making a real difference in the world (at least for awhile). Should people work at startups to get rich? I couldn't recommend it. It's probably going to fail, but worse, it might nearly fail and stumble along for years never amounting to anything but tricking you into riding it for longer than you should. Even worse still, it might succeed and you might not have been ninja enough about your contract to ensure you get a fair deal.

I'm glad I worked at one startup, but I definitely couldn't have stomached working for, say, 5 startups in a row. The emotional rollercoaster of just one startup was enough.

[-]Adam Zerner10y00

Do you think starting a startup in pursuit of a large exit makes sense from an expected value point of view? 80000 hours has some thoughts on this that I agree with.

[-]michaelsullivan10y-10

In a relatively healthy economy, to a first approximation in the medium and long term, the amount of money you make approximates how much good you are doing. As a liberal, I'll be the first to say that this has a lot of flaws as a benchmark, but in general, if you cannot find people willing to pay for, or donate to support what you are doing without you having to live on ramen forever, there has to be some question about whether what you are doing is providing value to the world comparable to a standard job in tech, finance, sales or a professional discipline, as long as you are moderately careful about who you work for in the latter cases.

[-]gjm10y100

Regarding the third career trajectory you mention, note that not wanting now to stop working now doesn't imply that when you get to, say, 50 or 60 you will still want to keep working for ever. You have only just got your first job. You are not in a strong position to make an inside-view judgement of whether you will ever want to stop working. It turns out that most people retire. Make of all that what you will.

[-]Adam Zerner10y00

Fair point. I've taken that into account and still stand by my predictions, but it's definitely something to be weary of.

[-]Fluttershy10y80

Say that I get 10 tries to start a startup in the next 20 years (I know that some take longer than 2 years to fail, but 2 years is the average, and it often takes shorter than 2 years to fail). At a 50% chance of success, that's a >99.9% chance that at least one of them succeeds (1-.5^10).

Failing to successfully found a startup would be evidence for your chance of success being less than you had previously imagined, so the bit about

a >99.9% chance that at least one of them succeeds (1-.5^10).

doesn't seem right, even if we posit that your chance of being able to successfully found a startup is 50% the first time around.

Good luck, though!

[-]gjm10y20

Allow me to say approximately the same thing in a different way. Adam, I guess you would agree that there are some people who will almost certainly never found a really successful startup no matter how often they try. Do you really think Pr(you are such a person) is less than 0.1%?

Further:

I think it's more like 75% of startups that fail.
If you have no savings to speak of, then after a startup fails you will need to go back to "ordinary" work for a bit to raise enough money to start the next one.
You may find it harder to raise investments if you have a few failures and no successes under your belt. (Again, if you have no savings to speak of you will need investment quickly.)
Mean time to failure may be only 2 years, but (a) I suspect that the most likely path to startup success goes through startup near-success, and (b) nearly-successful startups probably take longer to fail, and (c) actually-successful startups typically take well over 2 years before they start delivering serious money.
A startup that doesn't fail doesn't necessarily make you rich. You probably need quite a big success for that.

Let's suppose you don't care about (c) above because once you've got a successful startup you don't mind waiting, so your 20-year window indicates only how long you have to start an eventually-successful startup. Then in view of the numbers above I suggest an optimistic timeline looks like: some number of iterations of (1 year working to raise money, then 2 years for a startup to fail), then one of (1 year working to raise money, then 5 years for a startup to only-just-fail), then one of (1 year working to raise money, then startup succeeds). Then you have time for about 3 "quick" failures and one "slow" failure before succeeding. At a 75% failure rate, that's a lot less than a 99.9% chance of success. And if you fail on this path, you're now fortysomething, you have a long track record of failures, and you have no retirement savings at all. Good luck!

(When I say "an optimistic timeline", of course I'm aware that some people get there much quicker. But you have no reason to think you're in the lucky not-very-many-percent who will succeed first or second time around.)

[-]Adam Zerner10y00

Failing to successfully found a startup would be evidence for your chance of success being less than you had previously imagined

Hm. I agree that it might be evidence that you're not as good as you thought you were, but you also gain experience in failing that increases your chances of subsequent success. The relative strengths are obviously case-by-case.

[-]michaelsullivan10y70

Outside view of your 1 2, 3 and 4: most people end up in trajectory number 4, so thinking this is the least likely scenario needs some really good evidence.

In particular let's look at 1: How do you plan on an event that has a reasonable probability of not happening in your lifetime, and about which you know relatively nothing (if we could well predict what will happen on the other side, it won't be anything like a singularity).

Who is to say that a singularity results in happiness for everyone -- even for a positive one? From the standpoint of someone sitting in the 18th century, flipping into life today would be like a singularity -- even poor people have luxuries not dreamed of 200 years ago. That said, it's a lot more pleasant to have money or marketable skills than not, even in the cushy rich world. Try being actually poor in the US to see what it's like. See how well you can live on $800/month (a typical very small social security benefit) with no savings and no family support.

For 3: from your perspective right now, there doesn't seem to be any reason to stop working. Past a certain relatively young age, however, anyone who is not good at selling themselves, developing a network, or established as a well known expert in their field will find themselves at a huge disadvantage in the job market, and may no longer be able to get interesting jobs for good pay. At that point, someone who followed the moustache plan in their 20s and 30s doesn't really care -- they can try as many new startups as they want and give the finger to the "norma"l job market, never worrying about whether they will have enough to eat and live in a comfortable home.

If you have never saved any money, you may end up forced into plan 4 in your 40s often at jobs that are uninteresting and do not pay very well. I've seen it happen to a lot of smart people.

Do I think it's worth trying startups young? Yes. And realistically a few failures in a row will probably sap your will to start a new one in any case, so there isn't that much risk of getting past easily hirable age, so I recommend looking to do it early on.

That said, whenever you are making good money, some long term savings is a good idea -- money gives you future flexibility, and the hedonic hit you take to save 15-20% of your money when you are making a solid middle class income is barely noticeable.

[-]Adam Zerner10y00

Outside view of your 1 2, 3 and 4: most people end up in trajectory number 4

I'm not sure how to word this well, but "most people" haven't lived in the year 2080. Because of a) the possibility of a singularity and b) the law of accelerating returns, the "most people" that you refer to doesn't seem like an appropriate reference class to me.

In particular let's look at 1

I think you're downplaying the chances that a singularity does happen in my lifetime. 90% of experts seem to think it will.

Interesting point abut 200 years ago though. I'm having a hard time imagining the standard of living for the poor in that time period being insufficient for me, but I've noted this as something to examine further.

[-]Lumifer10y110

I think you're downplaying the chances that a singularity does happen in my lifetime. 90% of experts seem to think it will.

The experts are biased.

Consider two competent AI researchers, Alice and Bob. They both investigate the possibility of a singularity. Alice comes to the conclusion that it might happen in a few centuries or never. Bob comes to the conclusion that it will be possible in a few decades. What happens next?

Alice isn't interested in the singularity any more and goes off to work at, I don't now, image recognition. Bob as a consequence of his views still is interested in the singularity and continues to work on it. At this point Alice is not an "expert" on singularity, but Bob is. A survey would ask for Bob's opinion but will not ask Alice what she thinks.

[-]michaelsullivan10y40

I think you're downplaying the chances that a singularity does happen in my lifetime. 90% of experts seem to think it will.

I don't. (Edit: I meant this as "I don't think I am downplaying the chances", not "I don't think the singularity will happen")

It's true that I disagree with your experts here, and Lumifer speaks to some of my reasons. I even disagree with the LW consensus which is much more conservative than the one you quote.

That said, even taking your predictions for granted, there are still two huge concerns with the singularity retirement plan:

Even given that it will occur in your/my lifetime, how do you know what it will look like and that it will lead to a retirement you are happy with even if you have no capital?
If there is even a 5-10% chance that it doesn't happen, or doesn't provide what you want -- that is a fail when I am doing a retirement plan for most of my clients. I'm generally aiming for a 0+epsilon or at least <1% chance of failure if the client is able to follow the plan[*]. The only clients where building in a 10% chance of bust is ok are those who are in a real pickle, and there is no reasonable strategy to do better. Those clients' plans have to include downward adjustment of their goals if the initial trajectory is in or too close to the failure window.

[*] obviously most of the true failure chance happens when the client is unable to follow the plan at some point. Financially, some of that can be insured against (health and disability, life for dependent survivors) and some can't.

[-]Vaniver10y60

I'm putting perhaps 7% of my income into long-term retirement savings, and significantly more than that into generic investments. This is mostly to take full advantage of various subsidies offered to me to incentivize retirement savings, but if there weren't a sharp cutoff to use as an anchor I'm not sure what exact value I'd settle on. (I haven't run the numbers, but it looks like the tax ramifications of investing in a retirement account and then pulling the money out early are bothersome enough to not make it a good idea unless you're subsidized in some fashion or have a very high probability you won't want to pull the money out until you're old enough.)

It seems to me like hedging arguments should be most convincing. That is, in the case where there's an egalitarian singularity, it doesn't matter how much wealth you have saved, but in the case where there's an inegalitarian singularity, it matters a lot how much wealth you have saved. So it makes more sense to structure your savings in light of an inegalitarian singularity / you burn out early or have some medical issue / tech growth plateaus or slows.

[-]Adam Zerner10y10

I'm putting perhaps 7% of my income into long-term retirement savings, and significantly more than that into generic investments.

Thank you for the data point.

It seems to me like hedging arguments should be most convincing... So it makes more sense to structure your savings in light of an inegalitarian singularity / you burn out early or have some medical issue / tech growth plateaus or slows.

Would you mind commenting a bit more on what you think the likelihoods of the various possibilities are? In particular, what do you think the likelihood of !1 && !2 && !3 is? Because that's what I'd be saving for if I put money in a 401k or something where I can't take it out until I'm 59. (The opportunity cost to me saving for retirement is spending more time working on startups and increasing my knowledge.)

[-]Vaniver10y00

Thank you for the data point.

You're welcome!

Would you mind commenting a bit more on what you think the likelihoods of the various possibilities are? In particular, what do you think the likelihood of !1 && !2 && !3 is?

I don't feel like I have any special knowledge, and don't want to put where my gut points now down in writing.

My hedging point is that I don't think that number is very relevant, though: it seems like you should condition on no x-risks and not being in the !1 && !2 && !3 state, because both of those states are ones where the actions you take now don't meaningfully impact your position, relative to their impact in the other alternative.

But there are strong counterarguments to that: you might think that you can nudge the x-risk number, or that it matters to you how soon a positive singularity happens, or so on.

The opportunity cost to me saving for retirement is spending more time working on startups and increasing my knowledge.

Most retirement funds allow you to pull it out early with a minor penalty--I think it's 10%? If you have access to matching funds, I think you could put money into a retirement fund, get the match, and then pull it out and be ahead. You would have to look at the particulars of your match and fund, though.

[-]RowanE10y60

I would suggest an option 5), complementary to 4) - you work a normal-ish job, have an otherwise normal retirement plan that you put a much larger-than-normal fraction of your income into, and retire at a totally abnormal age. The go-to example of this approach is Mr Money Mustache. Past a certain level of income, most people are spending most of their money on what are essentially positional goods, and if you can resist the urge to buy what all your peers in the same income bracket are buying, you can save a big chunk of your income, and if you can get as high as 2/3rds you can retire in about 10 years. Should be much more comfortable, both in terms of risks and in terms of actual comfort, than spending 10-20 years on failed startups until you get one that succeeds.

[-]Adam Zerner10y00

Hm, that has the opportunity cost of me not spending the next 10-20 years working towards startups. That seems like a pretty big opportunity cost to me, and it seems that there's a high likelihood that I'll retire comfortably regardless of whether or not I save (because of 1, 2 and 3).

So your proposal has the benefit of a marginal increase in retirement security, but has the cost of me not working on startups for 10-20 years. What do you think of these tradeoffs?

[-]RowanE10y00

Well, I'm not very optimistic about the idea of spending 10-20 years working on startups, I didn't comment on it because I thought it had been addressed enough by others, but I suppose I'll add some thoughts on it: The "75% of startups fail" is from a group already selected for being confident enough that they're startup founder material that they actually go and make a startup. A single failed startup might be the best way of getting the experience you need to make a successful startup, but a track record of multiple failed startups is probably a bad sign, and even if it's not necessarily evidence you're "not startup founder material", that kind of track record will probably affect you negatively.

Although, really my misgivings aren't from any particular argument like that, I just have a feeling that the calculation in 2) must be hopelessly naive, and that's followed by the picture of an impoverished person who once had lots of potential declaring, to the dismay of his family, that with four failed startups under his belt he's going to try founding a fifth one, and this one will be a success for sure.

I mean, maybe all the patterns that option 2) is matching to in my mind are actually totally inapplicable, but it would take an awful lot to convince me the opportunity cost of not following a predictable path to retirement in 10 years was worth it.

[-]gwillen10y60

It might be worth considering a mixed or hybrid path -- if you can get a high-paying traditional (software, finance, etc.) job, while being very frugal and saving money (and you don't have much or any debt), you can get pretty quickly to the point of having a very comfortable savings buffer. If at that point you use your savings for living frugally, but not for funding your startups (do that with Other People's Money), I think you'll have more runway and more options than you might otherwise, and should you find that the startup life is not working out you should have a good backup trajectory available where you go back to a high-paying traditional career and retire.

I'm curious why you think (4) seems unlikely -- is this a fact about your personality, or about the world (i.e. you think there will be no normal-ish jobs, and/or the world will not survive to see your retirement?) Considering again high-paying jobs like software and finance, I think the math shows that a very frugal person can effectively retire after very few years in such a job -- perhaps as little as a decade, certainly as little as two, which is similar to your stated startup timeframe -- depending on what your goals are. And once you're in "effective retirement" (i.e. you don't have a problem supporting yourself), then you're in a great position to try to get a startup going.

[-]Adam Zerner10y00

I'm curious why you think (4) seems unlikely -- is this a fact about your personality, or about the world (i.e. you think there will be no normal-ish jobs, and/or the world will not survive to see your retirement?)

Let me try to clarify. What I'm trying to say is that 1 & 2 seem likely enough that I won't need to worry about saving for retirement. And even if (!1 && !2), it seems very likely that I'll be both willing and able to continue earning money well past retirement age. Given these three things, it seems like pursuing 4 is unlikely to be worthwhile.

[-]gjm10y40

Those dates are not estimates for when a positive singularity will happen. They are estimates for when we will get AGI at roughly human level. The same people who made those estimates seem to expect a median ~20y between AGI and superintelligence.

[-]Adam Zerner10y20

Thank you for clarifying.

[-]Paul10y20

I agree with the advice to get a normal-ish but high-paying job, save and invest a lot for around 10 years, then "retire" and think about doing a startup using the skills you have developed.

The craziest thing I would do is a) put my money in an index fund instead of some sort of retirement account, forgoing the tax benefits of a retirement account. And b) keeping a rather short runway.

If you're in the U.S., you can put your money in an index fund within a retirement account with potential tax benefits without any downside. Just open a Roth IRA at Vanguard, for example, and invest up to $5,500 per year in after-tax dollars. You can withdraw your previous contributions (but not earnings) at any time, tax-free and penalty-free. After maxing that out (and your 401(k) if available), you may be interested in the soon-to-be-released Schwab Intelligent Portfolios for additional investing.

[-]RomeoStevens10y20

Save some cash and keep your ear to the ground for ideas/cofounders (equally important). If this fails hard you have plenty of time to enact a different plan. If you'd like to chat in more detail about starting companies pm me and we can skype or something.

[-]Adam Zerner10y00

If you'd like to chat in more detail about starting companies pm me and we can skype or something.

That'd be fantastic! PM'ed.

[-][anonymous]10y10

Decide what you want to live for. I see three reasons to live: discharge duties, pursue personal goals, or gather hedonistic enjoyment, of which there are many kinds but let's just summarize as sex, drugs and rock and roll. In the first two cases, if you expect to also get paid for them, the goals are probably never reached and thus you never retire and never stop earning. The third option reduces lifespan anyway, running out of money and health/lifespan roughly at the same time can actually be a kind of local optimum.

Quite honestly, I only save for emergencies. By emergency I mean 3 years unemployment. OK I can expect some state pension (EU), but generally speaking I don't see an old age spent in poverty a threat, if my naughtier, hard-drinker etc. subagent wins, I cannot really expect to live long, and most likely the only way to beat that subagent is to develop another one that has some kind of life goal, and if that happens I will probably never retire from that, and it is also likely I will always get paid for that. If for example I retire and start writing and realize nobody buys my books, then I am maybe at year 1 of my three year unemployment fund so obviously then I un-retire.

[-]michaelsullivan10y00

To answer your specific question, there are a bunch of potential alternatives.

You can use a Roth IRA to have access at least to your contributions without penalty, and have tax deferral and tax free earnings.

[-]Epictetus10y00

Best case scenario is that your startup succeeds wildly or the singularity leads to a world where finance is moot.

On the other hand, the singularity may end up being rubbish and your startups could flounder. Saving now and letting compound interest pile up is a way of hedging against this scenario. It doesn't take all that much to build up a reasonable nest egg should things go south. Here is a compound interest calculator if you're interested.

It's a personal decision whether you find certain scenarios likely enough to be worth hedging against.

[-]diegocaleiro10y00

lesswrong.com/lw/hfw/why_is_it_rational_to_invest_in_retirement_i_dont/

[-]TobyBartels10y10

http://lesswrong.com/lw/hfw/why_is_it_rational_to_invest_in_retirement_i_dont/ (although the link created in your comment also works)

Moderation Log