I just got this random idea that people who want to become better at life could benefit from a common scale of "leveling". No, I don't mean vague Lesswrongey things like "changing your mind". I mean a set of concrete criteria like "you qualify for level 2 if you can do 5 pull-ups, have solved 30 Project Euler problems, and did 10 cold approaches". Obviously there would be separate ladders for different character classes, but not too many. Also obviously, my example was a bit too high for level 2. So I guess I really want to ask some meta questions here:

1) Do you think agreeing on a common leveling scale would be a good thing for a substantial subset of LW users? Would you feel good about leveling up and telling other people about it on LW?

2) Is there some good way to determine leveling criteria that are neither too high nor too low? Maybe make an intermediate scale of "experience points"?

New Comment
126 comments, sorted by Click to highlight new comments since:
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

It would be more useful to describe what abilities one would have at a particular level, than the amount of grinding it takes to achieve it. Being able to solve Euler problems at all might be a useful indicator of being a certain level, but having done a lot of them doesn't seem directly important, and if it teaches certain skills, then we should look at those skills directly.

I entirely agree with this. I'd rather take shortcuts to levels of ability rather than grind through them, if shortcuts are available. XP is only useful when there's only one way to get to a higher level. I'm pretty sure that's not true in real life.
Agree with the general point about grinding, but Project Euler is the best way that I know of to look at the relevant skills directly. I'll be happy to use something else if it works well enough.
I am not very familiar with Project Euler, so to get an idea of what you are talking about: what skills are you looking at with 50 problems that you can't look at with 5?
The problems have rising difficulty level. You need much more understanding to solve a problem like this than to solve this one.
(Ooh, I like that first problem. It reframes in all sorts of interesting directions.)
Ah, the increasing difficulty level makes a difference. Though for purpose of proving skills, it seems you could just skip to the harder ones, even though that may not work well for training.
They are not monotonically harder, and they are not all the same exact skill being tested. So someone who has completed Project Euler problems 1-30 has done more to demonstrate her ability than someone who has only done Project Euler problem 30. If your level is such that 30 represents your current challenge, then doing 1-29 won't take toooooo much time anyway. And you can still have fun trying to improve on the best solution offered so far for the problem -- there are multiple ways to solve a given Project Euler problem.

Chore Wars and EpicWin exist. (Haven't actually used ChoreWars. I used EpicWin for a while, I still technically do, but it's not as effective as I'd like.

Up until recently I'd have been highly supportive of this. A few days ago I read a counter argument, specifically to the gamification of education, which made the point that a) when you have extrinsic motivators, people are less likely to care about the things for instrinsic reasons, b) people tend to game the system rather than actually accomplish the thing you intended.

I don't know how true those statem... (read more)

One thing I've experienced is that you do NOT want to assign points for number of things done unless you have some way to assess that they're done well -- I spent a while maximizing the number of things I did per day, and quality went wayyyyy down.
Thanks for the links! But EpicWin and Chore Wars are trying to solve the problem of getting things done, while I'm trying to solve a completely different problem.
I use(d) EpicWin for something closer to what you're talking about.
Dan Pink discusses some of the research regarding this problem in an RSA Animate video. (He also wrote a book about it.)

This is an incredibly cornball idea, completely worthless for any naturally rational mind, which sounds like it was designed to help out irrational game-obsessed apes instead.

As an irrational game-obsessed ape, I wholeheartedly approve. Thumbs up.

I think a single common leveling scale isn't a useful idea. For instance, I'm not a salesman, and in the other popular context of "cold approaches", doing them would probably reduce the level of my marriage! A scale for each of a few dozen skills would be interesting, though.

Unless this sort of game b... (read more)

There's no guarantee that your life situation will be compatible with leveling up in all categories. That's not ungamelike, there are tradeoffs. OTOH there is no guarantee that everyone will progress through the same levels in the same way.

I definitely want a leveling scale. I want to earn merit badges and compete annually for friendly trophies. Maybe they can be gamed - that's fine, we'd have to start somewhere, doesn't bother me. This is a project I am and have been excited enough about to actually work on.

Nice! Do you have any thoughts on the second question in the post?
I haven't thought about how to optimally set levels, or about e.g. xp as sublevels. I have thought some about the "merit badge" approach. Subcategories of Anna's 9 branches? And from my notes "Should badges have levels, like level 1: you can talk a good game, level 2: you've beaten others in contests, level 3: you are good enough to be a target for those trying to beat the experts, level 4: you are the current champion amongst level 3s." This makes sense for some things but not others - the ability to do 5 pushups for instance is good, but being able to do more pushups than everyone else may not be. I think I would be less interested in obtaining some super-broad "Level 4!" with checkpoints than in embarking on a 3-month journey to achieve my "You have at some point demonstrated at least apprentice-level skill in novel estimation problems!" badge.

So what's the next actionable thing people want to do to move this forward?

I propose that we talk here and agree on a definition of Level 1 that would be achievable by most of us within a week or two (just to get the foot in the door), then start working toward it and discussing what Level 2 should be.

The ExRx table of strength standards has a level called "untrained," defined as "the minimum level of strength required to maintain a reasonable quality of life." Of course, when I started exercising, I was below this "minimum," but it took only a little bit of dedicated effort to get there. That's what I think "Level 1" should be like.

Social: be able to initiate a conversation with someone you've met and make plans to meet again.

Programming: be able to write a program with a for loop in some language (i.e. multiply the numbers from one to N.)

Discipline: be able to work for two hours without stopping.

Math: I'm not sure, because I'm so wrapped up in it that I don't have a good gauge of "minimum necessary," but perhaps, be able to prove Bayes' Theorem, or understand differentiation and integration on a more than mechanical level.

Endurance: be able to run for a mile without stopping or walking.

Memory: learn a short poem or passage by heart. (Maybe the Gettysburg Address, to be U.S.-centric but standard.)

Empiricism: find a question you REALLY don't know the answer to (and can't instantly google) and either design an experiment or read experimental studies until you have convincing evidence for one side or the other.

I think that for the reference class of humanity, the minimum amount of math that you need is the ability to consistently not get cheated out of money, and to be able to count without your fingers.

How consistently do we have to be able to do these?

Discipline*: Wake up on time at least 3 days a week.
Good start, thanks a lot for the work you put into your comment! Harsh critique follows. 1, 6 and 7 are precise enough to be part of a formal leveling scale. (But we'd need to i18n-ize them a little because all three use the American cultural assumption. I don't know offhand the weight of a pound, the length of a mile, or the number of words in the Gettysburg address :-)) 3 could be made more precise with a little work. I'd prefer it to say "write and run a program", to make the winning condition clearer. Also I'd rather use one of the classic examples, e.g. Hello World, 99 bottles of beer on the wall, or any of the first ten Project Euler problems. IMO we shouldn't be in the business of inventing new exercises that teach programming! 2 is not very informative: if you were able to initiate a conversation with a stranger once, that doesn't mean you can do it again. Same for 4: can Bob qualify for level 1 if he's pretty sure he worked for two hours straight at least once in the last month? I'd like to have criteria that are about as informative and insensitive to flukes as #1. 5 and 8 are too vague and need to be completely rethought, IMO. A level 0 person doesn't know what a proper mathematical proof or a proper piece of evidence feels like.
Agreed on substituting Project Euler problems. (God, I should really be doing those at some point...) Maybe 2 and 4 need to add a frequency requirement? Let's say, in a month, 4 invites and 2 days a week that involve a 2-hour sprint of work. (BTW, I think meeting strangers is less important than initiating conversations with people you've already met; my "social" criterion was about the latter. Can you ask that guy you know from class if he wants to meet up and study? People who can't arrange to meet are at a disadvantage, and I've struggled with that for a long time.) 5 and 8 are too vague. The trouble with LessWrong is that there's a very wide range of math abilities. I think calculus is conceptually important, though; maybe it should be something like "complete a Khan academy course," "complete a Schaums Outlines calculus book," or "Pass the AP calculus test." I honestly don't know what to do for empiricism.
Great idea about frequency for 2 and 4! It adds an element of grind (leveling is now guaranteed to take time), but that sounds like an okay tradeoff to me if we can't think of anything better. Also I agree that talking to people you know is a better idea for level 1 than talking to strangers. I have set up a wiki page with our current draft. I think you should be able to edit it :-) And let's commit to finalizing the requirements for level 1 within the next couple days - do you agree?
Could this link be edited into the article? I missed it before.
Awesome job starting us off! Sorry I forgot to mention that earlier. Maybe a minimum level is realizing that you don't know something, and being willing to update on information related to the fact that you're curious about?
First, let me explain what I'd want out of the "leveling" process -- a way to "become more awesome" by holding ourselves to rather high competitive standards in a bunch of different, but not too obscure, areas of human achievement. I don't think a person needs to get to ExRx's "untrained" level of strength to have a "reasonable quality of life" -- I know lots of people who seem to be doing fine without being that strong, or without being able to run a mile. However, an elite athlete would think "Oh my god, how do you live if you can't even run a little?" That's what I was getting at. What the elite levels of a skill consider "minimal" is, of course, way above the actual minimum that ordinary people practice. That kind of "minimal" is a good, challenging goal for an ordinary person. (For example: I am really not a finance expert. I am also a terrible cook. I would want to know what a professional in finance, or a professional chef, thinks is the "bare minimum" of financial understanding or culinary skill, and see if I can get there.)
While I see the intuitive appeal of this from a status standpoint, I don't understand why one would want to be other-optimized by someone who, by definition, doesn't understand enough about one's situation to know whether the given advice matches the stated criteria for applicability or not. I will admit that my reaction could easily be the result of an ugh field, though. Ugh, people who think they know how I should be living my life.
It really may not be for everybody. Some people -- I'm one of them -- do want to be "other-optimized" in the sense that we want to know "how we're doing," generally, by the standards of other people, and we're willing to try meeting an external set of challenge goals. Maybe it's an age thing; I am not old enough that I think I have my life figured out and that I know what "works for me." A lot of things are still in flux, so I'd be willing to learn entirely new skills and add new identities. That seems to be common for people in their teens and twenties. I wish somebody had shaken me and told me "No, seriously, learn to program NOW" when I was a lot younger. I expect there will be other things that I don't yet know I will want to be good at. This "leveling" business seems like a way to facilitate the process.
I'm pretty much behind the idea of leveling, actually. Coming up with a coherent, stepwise map of how to get from a layperson's ability level to mastery of a certain topic seems like it will have all sorts of benefits. I just think we can do a lot better than asking an authority on a given topic what they think everyone should know about it, to figure out which steps should go first or what should be considered a very basic level of competence - I think we'll end up optimizing for the wrong things, if we do it that way.
This is exactly what I want from the leveling process as well. Thanks for expressing it so clearly, and for the suggestions of cooking and money management. These sound really nice to have!
Some of these don't seem very well calibrated for 'minimum necessary to maintain a reasonable quality of life'. Programming: Most people don't program at all, and do fine, so this might not be a useful category in the first place. 'Computer interaction' would be better, and level 1 might be something like the ability to use reasonably-user-friendly web-based email or the ability to write, save, and retrieve an essay in a word processing program. If we want to talk about actual programming, though, level 1 might be something like understanding what variables are, or that computers 'think' in a purely procedural way (that there is no ghost in the machine). Math: I agree with atucker that basic arithmetic is appropriate for level 1, though I'd add 'ability to parse simple word problems' to it. Endurance: I don't run. At all. This has approximately never been an issue, and doesn't affect my quality of life in any way that I can detect. Level 1 for endurance should probably be something more along the lines of being able to walk a few miles (2, maybe 3) on flat ground without becoming winded - this seems about in keeping with what someone would need to do in the course of shopping at a large store. Empiricism: Again your suggestion seems too advanced for a 'minimum necessary' level. I'd suggest something more along the lines of being able to notice intrinsically flawed arguments or arguments where the arguer is obviously biased.
Perhaps we should refine "minimum necessary to maintain reasonable quality of life" to something like "minimum necessary to maintain reasonable quality of life, given that the area is relevant". By way of analogy, I don't own a car. The minimum necessary skill in driving to maintain reasonable quality of life is absolutely nothing for me. But for many, driving a car is necessary; for them, the minimum necessary skill is being able to remain within the lines, use indicators, park and three point turn, etc. Pass the driving test, in other words. There's no point standardising levels of driving skill to include my lifestyle! Given that driving is relevant to your quality of life, you need pass-driving-test levels as an absolute minimum. So that has to be Level 1. Being able to get by without a skill isn't the same as having the minimum amount of skill necessary. Being Level 0 in driving and still having a reasonable quality of life is a fact about me, and maybe about the public transport system where I live, and maybe also about the willingness of my friends to give me lifts - not a fact about how much skill is required in driving. (When I crossclass into Professional, though, and Driving becomes a class skill, I'll have to get to at least Level 1)
I see, and basically agree with, your point, but that benchmark seems to still have some problems: Specifically, I'm having trouble coming up with a scenario where programming would be a relevant skill but level-1 ability (equivalent in difficulty to other level-1 benchmarks) at it would be sufficient.

I had a data entry job in the summer of 2002 when staying with family between years of college. After a day or two meeting people and finding out where the bathrooms were and getting started with the nominal data entry task I installed a macro recorder so I could factor out some of the human tedium by writing scripts to speed things up.

By the time I left the job 8 weeks later to go back to school I was teaching the "real employees" how to automate the boring parts of their own jobs and had them hire a friend who lived in the area to continue their macro lessons and to write the really "tricky" macros on the side (he'd upgraded the job to writing perl scripts within a few weeks).

Basically, if someone thinks they can be a "white collar worker" without any "algoracy" (cognate to literacy and numeracy), I suspect they are in the process of becoming economic road kill. The space of AI-hard jobs is steadily shrinking. Maybe some people can switch to "blue collar work" and learn to drive a tractor or pick strawberries instead? At least for a while? See, there's this thing called the singularity... but if you're here reading and commen... (read more)

I agree on programming. (I'm progressing towards level 1 in programming currently, and programming so far has allowed me to write a script that eats a .txt combat log from an MMO and spits out information I care about, it allows me to use a Python console as my daily planner (from collections import deque -> create a stack+queue of tasks), and it allows me to solve Project Euler-type problems. So not a whole lot.) However, I am treating levels of programming much like I treat levels of wizard - low level spells suck, high level spells are game-breaking-ly awesome.
Writing simple, convenient shell scripts. Solving low-level Project Euler problems.
Those sound like what's needed. Thanks for making awesome concrete suggestions! I wonder whether the Math category should be broken down a bit. There's calculation - which is sometimes easier to start with e.g. things like basic arithmetic operations, representing real quantities with mathematical numbers and shapes), then there's algebra/geometry and the things which follow, and probability starts with a new set of axioms and applies to a different sort of problem, so it may deserve a separate skill hierarchy. I might also separate out "proving" as a standalone skill, though with not very many levels. (You learn to prove advanced math by learning the math, but initially you still have to learn how to follow and then how to generate a proof). Or maybe as part of a" logic" skill hierarchy?
I worry that the system will get way too complicated if we're splitting things up at that level. But maybe with many worker bees, it will thrive, like Wikipedia or MathWorld. In my leveling references (various games I've played), you could get to level 4 in Math, but put a note that said "strong on probability theory, average on the rest". I think we need to determine if we want a Level system as found in various games, or a Badge system, that quantifies/commemorates particular tasks (your breakdown reminds me more of Badges, not Levels). Maybe both? Generic human levels for the things most of us agree are important for everyone to work on, plus badges for the specializations, like algebraic geometry or the computational astrophysics of stellar dust? Maybe "numeracy" for the generic human version of math, cf. John Allen Paulos' "Innumeracy" book.
Right now I'm most interested in "generic human levels", but several people here in the comments have independently expressed interest in the idea of "badges" for specialists. Perhaps you could make something out of it? :-)
It seems to me that badges are what you get when your levels become too specific. Like, if we can't agree on math badges and split it up by subfield (say, calculus, statistics, graph theory, topography, cryptology, etc.) and then can't agree on levels (what comes first, optimizing multivariable functions, or Taylor series?) then you essentially just collect badges in a bunch of fields.
Endurance: run a mile is much greater than level 1 endurance for anyone over the age of 30. I'd suggest a brisk-walk for 30min without stopping to catch your breath is level 1 (trust me, there are many people that can't do this).
Middle-aged people are not worse than young people at long distance running. The current male world record for the 24-hour run was set by Yannis Kouros at age 41, and the current female world record was set by Mami Kudo at age 45 (Wikipedia). I really want to revert the endurance requirement to what it said before you changed it, because I'd like the tasks to come with a sense of challenge and accomplishment if at all possible. Making some pancakes for the first time will certainly feel to me like "wow, I made some pancakes!" Running a mile for the first time feels like "wow, I ran a mile!" (That should've been in all caps, the feeling is so intense!) Walking 30 minutes for the first time... feels nothing like that...
YMMV, definitely, but I think walking for a distance rather than a time preserves some of that 'wow'. I know I was fairly impressed with myself the first time I walked a mile and a half to a convenience store, walked around to shop, and walked a mile and a half home without stopping. It might be less of a 'wow' than running some distance, but for level one I think attainability outweighs that.
Here is an calculator enabling equalizing for fitness among distances and times. For example. the fitness level required to do two miles in 30 minutes is approximately the same as that required to do one mile in 14 minutes and a few seconds.
Ok so we need a better understanding of what level 1 means then. I was comparing it to the "strength" level-1 which is given as "untrained"... ie what you could possibly reasonably do without needed any time training - even though it might feel like an achievement to do it for the first time... and the particularly unskilled in an area may have to do some training to get there. Not to you it might not... but I'm guessing you're probably fitter than average, though you might not realise it - average is lower than you think... As to age and running... it's not to do with the actual age.. but the fact that most people over 30 have slacked off for a while and are unfit - especially those of us that have been working a sedentary job now for ten years (ie most of us) . It's not related to age... but to what constitutes the "standard" level 0. I know a hella lot of people that could not walk for thirty minutes without getting winded... and universally these would be over thirty. Those under twenty can still coast on the endurance equivalent of "natural talent". Those over 30... have to actually work. Still, I like the other commenter's example of using distance. How about "walking 5k in under 40 minutes"? I certainly felt "wow" the first time I did that. For me.. running a mile feels more than level 1... I would get a "WOW" instead of just a "wow"... We can have lots of levels, you know. Running 5k can be level 3, running 10 can be 4 etc etc... all wow's of varying amounts of awesome... :)
Actually the ExRx tables that SarahC linked to are pretty hardcore, I've seen people on the internet say that it took them months (up to a year) of lifting weights 2-3 times a week to reach the level described as "untrained" =) Unfortunately I was unable to find any other table of strength standards, so this one will have to do, I guess! My current best idea for endurance is to keep the 1 mile run requirement, but add an option of achieving "Level 1 No Physical" for people who won't qualify for the strength and endurance parts at this time. (In all other respects the levels should be indivisible, i.e. you don't get moral whuffie points for achieving half a level.) Sorry and please don't consider me an ageist or something. Even if you do choose to take your time and do the physical part, I'll be having my trouble with the social part and you can still beat me to the finish line! =)
Ok, that all makes sense then. So to reiterate your point (and make sure I've got it). Level 1 for any particular skill will likely be unchallenging for anybody that already has skill in that level... but will in fact be challenging (possibly quite challenging) for people that have never attempted it before. I didn't realise we were working on making the whole set to be indivisible! That actually makes it more interesting! :) However... optionally taking out the physical I think would be very important. For instance people that have a herniated disc would never be able to do the weight-lifting training... ever. Similarly with running and people that have destroyed their knees. It'd be a shame if somebody could never get level 1 because of an injury even if they'd be level 5 in all the other skills... Anyways - otherwise I think it's a pretty cool idea.
Yeah that's about it. Guess it's time for me to freeze the thing in its current state and write a post. Thanks a lot!
Can I also suggest that those of us working on a level record the time/effort it takes to reach each level (and whether we think we were starting from scratch). That can possibly help us to calibrate the different "areas" a bit better.
Good idea! I'll write about that.
It's also humiliating, but in a motivating way, that I can't do part 2 of the social skil level 1, or discipline or empiricism (maybe not endurance either, I haven't checked). These are concrete things I need to work on, and it's encouraging to "know" (even though this "knowledge" doesn't involve real evidence) that they're no more than 1 level away from where I am now!
As "Finance" (and in particular personal finance) has been suggested as a category, I've added a level 1 proposal to the wiki page: You should understand and be able to calculate compound interest. You should be able to explain what present value, future value, and discount rate mean, how they relate to each other, and given any two you should be able to calculate the third.
"Being able to calculate" is fine (though I'd like a more clear test that doesn't require the user to generate random numbers), but "being able to explain" is not nearly crisp enough. It's almost as bad as using "being able to explain why politics is the mind-killer" as part of a test for rationality. Could you rethink that part?
OK, here's my revised version for finance: You should be able to figure out how much money you will pay in compound interest on a loan, or how much you will earn on an interest bearing account, in a fixed amount of time. You should be able to compare the monetary values of different financial decisions (e.g. borrowing money, paying off a loan, purchasing an investment) by comparing their present value.
It must be telling of my financial ignorance that I don't understand the second part :-( Could you give an example?
Suppose you can have $100 today or $100 in a year. Since you could do things with the $100 in between, but you could also choose to hold onto it for a year if that turns out to be a better idea, you're likely to prefer $100 today. But most people would take $1,000,000 in a year over $100 today, so the value of getting money a year earlier is finite. If you are indifferent between $105 in a year and $100 today - i.e. if a 5% return on the money would exactly compensate for a year's delay - then we say your annual "discount rate" is 5%, and the "present value" of $105 next year is $105 / (100%+5%) = $100. Finance generally assumes that discount rates are constant, and compound just like interest. To give a simplified example of where this is useful, suppose you are deciding whether to buy a home for $100,000 (in cash) or rent a home for $10,000 per year, payable at the end of the year. Your discount rate is 5%. As simplifying assumptions, suppose no transaction costs, you know you will live there for exactly 10 years, and neither the home's value nor the rent would change after that time period. The present value of $-100,000 now is obviously $-100,000. But you could sell the house in 10 years for a present value of $100,000 / (1.05^10) = $61,391.33, so the net present value of buying is the revenue minus the cost, $61,391.33 - $100,000 = $-38,608.67 The present value of the rent over 10 years is the sum from i = 1 to 10 of $-10,000 / (1.05^i) = $-77,217.35. So in this example buying is much cheaper than renting. In a real life buy-vs-rent calculation, you have to deal with complicating factors like the amortization of mortgages, but you can deal with most complications by calculating the present value of each component separately. That's what I did in my own buy-vs-rent calculation.
Thanks! For what it's worth, I was missing the part where everyone has a personal discount rate. If I'm allowed to assume that, then everything becomes obvious, of course. If you don't mind, I have edited the finance part to say "make a buy vs rent calculation, using prices appropriate for your area and your current standard of living". That sounds more crisp to me than "be able to do something".
Thanks, that probably made it much clearer.
Who's "most of us"? Are young adults supposed to be level one? Teenagers? Precocious children? What does level 2 look like? 3?
By "most of us" I meant most of the people who will be participating, whatever age they turn out to be.

I propose the following: anyone can create a merit badge (example to follow in self-reply), and it becomes official if it's seconded and thirded (and not downvoted below -2).

Public Commitment: 1 point for announcing in public (such as to a comment thread) that you intend to accomplish a particular goal, together with a deadline by which you will know whether you did it.
I don't think you've thought that through.
Whoops. I found this link more persuasive, incidentally.

I don't see it working as an external ruleset, but perhaps people could more customarily announce their own leveling up, personal changes that added up to making a difference in some respect, like mastering a major skill, or fixing a mistaken policy (wrong belief) that had influence over you, or gaining a very useful insight.

I've added "make pancakes" as level-1 cooking. Pancakes are a "mix all ingredients and cook" recipe - so pretty straight-forward. I've added a link for a fairly explanatory recipe too.


It took me FOREVER to find http://wiki.lesswrong.com/wiki/User:Cousin_it/Leveling from this page. Is there a way you can edit a link into the OP?

It might be a good idea to suggest alternative courses in each. For example, the Strength test is mostly a test of ability to access weightlifting equipment.

Suggestions (programming and finance already look fine to me):

  • Strength: do series of 10 pushups, on 8 separate days within 1 months
  • Endurance: swim 500 m without stopping
  • Social: initiate a conversation with someone you know and obtain additional information (e-mail address, IM addresses, telephone number, postal address, birthday, family composition). Do that at least 10 times in 1 month, with at le
... (read more)

Self-control: Buy a bag of candy or similar superstimulus-ey multi-portion food item. Leave it in a place where you will see it at least three times a day. Go a week without eating any of it.

Creativity: Make 5 art cards that explore a theme.

I have no stove or oven. I can run a 6 minute mile. I cannot really ride a bicycle. Swimming...well, better than biking, at least. Alternatives will almost always be needed.
Yes, I love the idea of alternatives. discipline: set yourself (and meet) a regular commitment that must be completed at least twice per week (examples include household chores or a new exercise routine). endurance: cycle 10km without stopping social: join a new social group with > 15 people regularly attending (see meetup.com for many examples). Within a month, learn+memorise the names of 15 members.
Great, thanks for the suggestions! Though it seems really hard to make the difficulty of the alternative courses match the original ones closely enough, e.g. your strength requirement is much easier than the current one, endurance is much harder, social is easier... Creativity: "5 drawings" is too vague for someone who doesn't know how to draw (how much is one drawing?), but writing 500 words in one sitting is a great idea and I'm taking it! Games: I like the suggestion, and would prefer to start with just chess. No idea how much work is reasonable to ask from a beginner, though. Altruism: icks me out :-)

There's plenty of physical fitness programs, such as Hundred Pushups that already effectively give you a leveling system AND an actual training program. I've found them quite useful, although I dropped them as soon as I stopped gaining levels easily.

It doesn't seem like it would be significantly harder to come up with alternatives: A NaNoWriMo training program where you produce 50 words/day for a week, then 100, until you're doing a novel in a month, for example.

I don't see any reason to have "generic" levels which combine physical fitness, probl... (read more)

I like the idea of "generic" levels because I'd like other people's appraisals of what they think are standard skills of human competence. (For example, it took me a while to realize that fitness is a standard domain of human competence, and that I was being silly by defining myself as "just not into exercise.")
One could easily have a website listing broad categories (Exercise, Puzzles, etc.) and specific challenges within them, without trying to make a handful of metrics to be a "generic level"
I agree with wanting to know what others think of as standard skills of human competence, and non-standard, interesting/useful skills of human competence. But I don't see how needing generic levels follows from that. You can know just as much by having a (strength, stamina, math, chemistry, literature, instrumental music, etc.) level system, and with folks giving suggestions for what would be most useful to accomplish life goals generally, without trying to organize those skills into a "human level 1, human level 2" system.
If you're interested, I propose that we start defining the first level :-)

Here's one comic's take on the gamification of human incentive structures.

Fictional evidence, doesn't contribute to the question asked, fun and distracting.
It raises the problem of defining the system on which to acquire levels and the potential for perverse incentives when developing such a system, which can accumulate over time, following Goodhart's law.
I don't think we'll have a problem with system-gaming as long as nobody "wins" anything real. It's all about self-improvement. If somebody wants to lie and say they self-improved, that's their own (screwed up) business... it makes no difference to me, or anybody else. People will still be supportive of each other's efforts, even that of the liar.
It seems pretty unavoidable to me that people will 'win' status this way. On the other hand, if 'level 4 programmer' means something specific, and someone claims to be one, that claim should be falsifiable, and they should lose a significant chunk of status if it's falsified - more than they'd gain from the lie in the first place, in fact. So it's still unlikely to be a problem.
Experience suggests that people will in fact game systems to win fake rewards.
Yes of course they will. My point being that it won't make any difference to the rest of us even if they do.
That's probably mostly right. It does mean that public leaderboards are unlikely to help, though.
Not doing this project doesn't degamify our society, so we don't on the margins have a play-it-safe option.

I assume this isn't supposed to match up with anything else, right? Level 6 isn't reserved for superhumans?

An idea for level 2 programming (note, you must complete all of the below):

  1. project euler has its own levels. Become "level 1" within Project Euler (which requires completing at least 25 project euler problems).

  2. create an account on stackoverflow.com - gain at least 100 reputation solely from your answers to other people's questions.

  3. start a blog based around your programming language/niche - and post at least four blog entries each month

I've added to the "memory" one: for short-term memory, using dual 1-back. No idea if I've made that one too hard....

I've also reduced the difficulty-level for the endurance one. I think "brisk walk 30min" is level 1. "run a mile" is at least level 2 as it requires some training to get there.

Whoops, didn't see your comment at first. Sorry. Replied about n-back in another comment. Also, do you think none of the tasks for level 1 should require any training? SarahC, who proposed using the ExRx scale for strength, said she had to train a little before she reached the "untrained" level.
Yes but many people will need to "train a little" to be able to walk for 30 min without stopping for a breath. Same with cooking pancakes, or programming. I think my idiomatic phrasing of "some" is probably the problem here, so let me be more explicit. For me (and many other people), I would need to train a lot to run for a mile. As I am in fact currently training to run, I can tell you that it has taken me two months so far to get to a stage where I can only walk/jog for twenty minutes, not yet actually run a full mile in one go. I expect to be able to do that at the end of this month. It is not level one.
Uh... I'm really sorry to say this, but training for three months to run a mile is not average. You're probably training incorrectly. I just looked on the internet and most programs for beginners say that after 8 weeks you should be running about 2 miles in one go, and they use a very relaxed training schedule (20-30 minutes, 3 times a week).

I like this and had a similar idea. It did not involve "leveles" but a learning tree, extremely similar to the one found at http://www.khanacademy.org/exercisedashboard . So you would not need "character classes" or "levels" but had an immediate view on the persons abilities. Based on the learning tree you can of course introduce "levels" as a branch of the learning tree and calculate "character classes" based on the proportion of the mastered skills.

Ultimately, this amounts to order all human knowledge int... (read more)

I've had this idea before too... but it seems to me like you need to have "character classes", instead of trying to make general "life levels". E.g., here'd be three levels for a character class of "statistician":

Level 1: given data (small enough sample size to calculate by calculator so you don't need a script), calculate basic summary statistics like the median, variance, etc., and be able to interpret them.

Level 2: pass quizzes to demonstrate understanding of basic fundamentals, like the difference between within-group and ... (read more)

Character classes are not necessary, but I agree with you that they're better than generic life levels. Even better is a granular attribute & skill level system, such as found within the World of Darkness system (Mage: The Ascension, Vampire: The Masquerade, etc.). So strength, stamina, programming, statistics, and so forth each have their own level. When you get advanced enough in level in a narrower skill area like statistics, you can sub-specialize that skill. But I think we don't need to worry about special cases like that for now.

I like this idea. Describing sufficiently granularr levels would also contribute to positive-feedback cycles, and to some extent even taskification (i.e. I know what I need to learn to level up).

Then let's start, I guess.

Level 0 rationalist: You think that "Either possibility A must be or possibility B", means "A 50/50 chance either way".

Level 1 rationalist: You no longer think that.

1) Nah, probably not. Some common standards could be nice, but they should be specific, and once we start trying to do something LW-specific rather than some flavor of "general," karma ain't broke. Self-assigned standards might work if you can resist or somehow (like with another person) police gaming the system, but there's already systems for that.

2) The problem is inherent - putting things in this structure requires that you assign them relative worths. There may be an average answer, but average is wrong for a lot of people.

Sounds like fun, but it will take a lot of thought to come up with good leveling criteria. A naive set of criteria is just going to be boring or unproductive.


I think it's a great idea, and I would like to participate (I've thought about bringing up this idea, too). However, I think it would work best as an external (perhaps affiliated?) site.

Please see my reply to atucker.

Yes, I would greatly appreciate such a scale in my own quest to beat Akrasia :)

I never thought cousin_it of all people would stoop so low as to post a self-help thread.

Thanks for the compliment! No, really. Unfortunately I don't deserve it, because sometimes I get these ideas...

We already have a common leveling scale set out by the universe itself: how many surviving children you have.

If you don't like that one, and prefer an easier proxy, we already have several: the most obvious being your annual income. (Corrected for education level if you prefer.)

If you don't like any of those, and you're looking for easier proxies still, I can only ask: Why? What exactly are you trying to accomplish with the scales you are trying to derive? Once you figure that out, it may be easier to decide what the scales should measure.

Going to the gym does't require figuring out your cosmic goals first. Typically, leveling up is not a game objective in itself, but it's supposed to help you achieve game objectives, whatever they are.
Cosmic goals no - those being far mode beliefs, if anything they are likely to be counterproductive. But it seems to me that leveling up is something to be done in the service of game objectives, not vice versa. For example, consider the following strategies: 1. I'm trying to optimize for lifespan. The available evidence says exercise is beneficial for that, so I will set out a program of going to the gym on a permanent sustainable basis as well as cutting down on calorie intake and driving, and aiming within the next decade to move to a region where cryonics is available. To help me achieve these goals, I will design a metric to measure them. 2. Everybody seems to think you should go to the gym, so I guess I'll go to the gym. I think the former strategy is, apart from anything else, more likely to have you still going to the gym in six months time. -shrug- It seems most people disagree, based on the vote total; so be it.
I don't understand this. It seems like you should want to maximize the amount of things that you really care about that you get done. You should focus your effort on your actual goals. Not some set of parallel goals that you don't care about called "levels." If you want to maximize how much you get done that you do care about, you should aim to minimize the amount of self-improvement that you do. You should make only those improvements that are necessary to get done what you are trying to get done right now. You shouldn't upgrade yourself speculatively.
Income and kids aren't what most people who are interested in self-improvement are striving towards. Do you really disagree with that?
I'm neither agreeing nor disagreeing - frankly, I have no statistical data on what most people who are interested in self-improvement are striving towards. What I'm saying is that if you are interested in something else, it might be better to figure out why, and just what that something else is, and why you are trying to come up with a metric; the answers to those questions might help you figure out what the metric should be.
How do you think people are planning on constructing this "level system", if not through assessment of personal goals ("what you're interested in") and appropriate metrics for those goals? I see no difference between what you're suggesting as "better" and what the OP and discussants suggested.
Well, take a look at one of the concrete proposals that have been made, say memorizing the Gettysburg address. Is this a good or a bad idea? I don't know. Personally I have an explicit policy of delegating that kind of long-term memory to machines; if I were going to try any kind of memory exercise, I'd go for something like dual N-back that tries to train working memory. Does that mean memorizing the Gettysburg address is a bad suggestion? Well, I don't know. It wasn't presented with any accompanying rationale. For all I know, maybe there's actually a very good reason why at least some people might want to thus train long-term memory, that the person who suggested it is seeing and I'm not. It would be easier to assess proposals if they were accompanied by discussion of how they are intended to meet some objective.
Using N-Back is a good measure, actually. What do you think that would come under? Memory? What do you think would be level 1? Dual 1-back? and level 2 would be dual 2-back?
I see you've edited the wiki page to add dual 1-back to "memory". I have some doubts about this, because a) no one really knows whether n-back is good for anything except improving n-back performance, b) I'd like to only include things that clearly and obviously sound like good ideas to the average person, and avoid things that may sound like snake oil (even if they aren't in fact snake oil). Would you mind reverting that edit? I don't want to do it myself because that would mark the beginning of a passive-aggressive edit war. Also, nice idea about pancakes! It's pretty much the perfect level 1 task.
Oh ok. I didn't know that about N-back. I guess, from hearing about it so much here, that I thought it was more solid than that. No problem I'll revert it. and I wouldn't mind if you reverted it as long as I had the explanation :)
Sounds reasonable. Might be worth further subdividing into working memory versus long-term memory since they would seem likely to be useful for different things?
Ok, and have removed it again at a request from cousin_it, who says that it's not confirmed that n-back has any effect on actually improving short-term memory (except as applied to playing n-back). I am not qualified to continue that discussion - if you'd like to continue it... it's in one of cousin_it's comments on the page somewhere... feel free to take up the thread ;)
Have added them as "sub" goals of Memory as "long-term" and "short-term" memory. They don't need to be top-level skills, I think.
I think you are confusing the terminal and instrumental goals in this situation. The "leveling up system" is a means to the desired end. It's an instrumental goal intended to be a fun way in which some of us see that we could motivate ourselves towards our "real" (terminal) goals. If it's not for you - then no problem. It's not going to be for everybody.
I basically agree. I do think that how many surviving children you have is not quite the right metric though. In the developed world we are now emphasizing quality over quantity in offspring. I would rather have 2 children that I have put a lot of time into and who will carry some of my beliefs and values into the future than 8 kids from 4 different baby mamas that will barely recognize me if they see me.