Macro, not Micro
Overview
The basic observation is that, if we think of life as an optimization problem, then redefining the search space is much more important than making local optimizations; as a fact of human psychology it’s hard to consciously focus on both; but we can implicitly get away with doing both by creating mental triggers for when local optimizations are likely to be particularly effective to think about, and by structuring things so that many local optimizations get made automatically.
Introduction
If you have been to one of the Rationality Minicamps or certain other CFAR events, you may have had the privilege to attend one of Anna Salamon’s excellent classes on microeconomics (despite the title of the post, I am being sincere here; you really should attend them if you haven’t already). There is too much content to briefly summarize, but essentially “microeconomics” in this context means applying basic microeconomic concepts like marginal value, value of information, etc. to everyday life. For example, if you spend 30 seconds brushing your teeth each day, then spending five minutes to think of something else to do at the same time (like stretching) will save you 3 hours a year, which is a great investment! (There are some caveats to this calculation, but I’m glossing over them as they aren’t relevant to the post.)
And indeed, spending 5 minutes (once) to save 3 hours (every year) is almost tautologically a good investment. Now that I’ve brought up this example, and assuming you value your time, you should probably actually go through this exercise (or just use the stretching suggestion).
The Problem
I intend to argue against something similar to this but subtly different. Basically, while any given trade such as the one above is good, I think it is a mistake to systematically search for such trades. Note that I also don’t want to argue that you should never search for such trades. If you’re about to buy a car you should almost certainly put a lot of microeconomic optimization into it, and if you can find things that improve your overall work efficiency substantially, then you are winning big-time. But I worry that, sometimes, the wrong lesson is drawn from these microeconomics examples (or cognitive bias examples, or any other rationality skill), namely that X is suboptimal by default and we should go out looking for places to optimize X.
The reason I think this is wrong is because it aims much too low—if you really want to save the world, then your average thought needs to be good enough to save two human lives [source: the average human lives only 3 billion seconds]. Even if each individual optimization you make ends up adding to the amount of time you have, the overall process of concentrating your attention on such optimizations makes you less likely to think other thoughts that would be far more valuable. Perhaps another way of putting this is that, even if each small optimization helps you a little bit, the time it takes to think up such optimizations actually makes you lose out—however, I don’t think this is actually it, I think it has more to do with forming mental habits, where you want to form the mental habit of making huge optimizations rather than small optimizations.
The Solution
What I think people should be more concerned with than micro is what I’ll refer to as macro—the overall structure of the search space (in this case the structure of your life and how you think) --- as opposed to making local optimizations within a fixed structure. For instance, becoming an atheist; or realizing that social skills are both trainable and highly instrumentally useful; or learning to visualize the steps towards a goal; or learning to code; or finding a group of allies that you didn’t previously realize existed; these are all examples of what I’d call “macro” optimizations that are the sorts of things we should be looking for. (I should note that a lot of “macro” skills were also covered in Anna’s microeconomics units.)
I also continue to think that there is a clear place for micro-level skills, as well. The key is to incorporate them into your thought process, both at the level of creating triggers to explicitly call micro-level optimization routines when they are likely to be helpful, and at the level of restructuring your thought process to automatically be more likely to make good decisions by default. For instance, the lesson I drew from Eliezer’s posts on cognitive biases were not that we should go learn about all the different cognitive biases, but that we should develop habits of thought that will automatically notice and decrease the effects of such biases. Then, for a couple of the more pernicious ones like trivial inconveniences, I further added specific alarm bells in my head to watch out for those, but only because I noticed that avoiding trivial inconveniences was routinely harming me.
Conclusion
I’m not sure how good of a job I’ve done of explaining what I wanted to (it’s still not entirely clear in my own head), so I invite your thoughts and feedback. I’d be particularly grateful if someone wiser than me (I’m looking at you, Critch / Wei / Yvain) could figure out what this post was trying to say and then write that instead!
Knuth famously said
Wikipedia explains:
Does that sound like the point you’re trying to explain?
I think that’s related, but different. I really like the quote though!
The difference is that I don’t necessarily think the “design” (in this case I guess overall life structure) will be less clean if you micro-optimize, but it will take away cognitive and habit-forming resources away from the things that are actually important.
Although I suppose in the limit of absurd amounts of micro-optimization, it might indeed lead to a pretty cluttered life...
This is a really good point.
I call this habit of optimizing details “suboptimizing”, because it is both suboptimal, and you are optimizing subcomponents.
The micro/macro thing is a good distinction. I’m suprised you didn’t reference the competitive Starcraft literature; it is full of discussions of micro and macro game.
Starcraft is definitely what inspired the choice of the word “macro”; unfortunately, I’m not a good enough Starcraft player to know that much about that literature. If you have particular references that you think I should read, I’d be interested in doing so.
I don’t even play starcraft, so I don’t know. I just know that they talk about it a lot.
They do, but it’s pretty much the game-theory min-maxing stuff you’d expect in a highly competitive environment dominated by geeks and nerds.
General usage of the terms in the Starcraft context.
Due to the context, personal evolution of the “player” (e.g. realizing that you can just get better at overwhelming your Enemies, and practicing for that) isn’t generally included in this dichotomy, and is considered part of the metagame, so I don’t know how far one could push the comparison.
The problem is that thinking of trivial optimizations is more rewarding in the short term. It gives the impression of working toward improvement and conveniently distracts you from the larger optimizations (which for me are surrounded by ugh fields).
I rarely even implement such “subopimizations” because I skip the part about making the optimized behavior habitual, they are simply compulsive.
So, macro-optimization problem: how do I stop compulsive micro-optimization?
It’s hard to know what the answer is at that level of abstraction, but probably general techniques on habit-forming would be helpful here. Critch (user:Academician) is really knowledgable about this sort of stuff but I don’t know if he’s systematically written out his ideas.
One approach is to, instead of trying to “get rid of X”, you can “displace X with Y”, where Y is something better than X. So if you spend 2 hours each month sitting down and thinking about how to better structure your life, and spend 15 minutes at the beginning of each day thinking about what you need to do that day and how to do it, that’s a good start. I also try to notice when I am doing something wrong (and then ideally stop doing it).
-- Umesh Varizani
-- Anon., quoted by Eric Sosman
My summary of this idea has been that life is a non-convex optimization problem. Hill-climbing will only get you to the top of the hill that you’re on; getting to other hills requires periodic re-initializing. Existing non-convex optimization techniques are often heuristic rather than provably optimal, and when they are provable, they’re slow.
An especially important example of macro choice that deserves some thought is the choice of a professional activity. See 80000 Hours:
http://80000hours.org/
I cannot parse this. Do you mean “five minutes”? That’s the only thing I can think of that would make that sentence parse.
This only added to my confusion. I understand the larger point, but this example is confusing. Stretching while brushing your teeth is impractical and dangerous, 3 hours a year is infinitesimal, and you should really brush your teeth for more than 30 seconds a day. The following part was much more coherent and the examples well thought-out.
Standing on one foot might be a better choice than stretching.
Okay, that’s actually a good point. I’ll try to think of a better example (ir you have any in mind I’m open to suggestions).
I’ve found that doing SRS reps while walking on the treadmill is a much more efficient than walking or reviewing flash cards separately. Your mileage may vary.
brushing teeth and shaving in the shower are decent examples.
This looks like micro-optimization to me. If you have time to brush your teeth while showering, you’re clearly not optimizing your shower time. (Not applicable if the outside is unpleasantly cold, in which case it’s better to brush your teeth inside than outside the shower. Also, shaving is probably faster in the shower.)
It was given as an example of micro-optimization, so that makes sense, no?
Oops.
I really enjoy being in the shower.
Thanks, it was indeed supposed to be five minutes. Fixed (although I will also try to follow gokfar’s advice and pick a different example).
(I found this pretty readable, and I think I got what you meant)
You bring up the awesome point that focus and attention are limited resources, and that we have to trade off between different kinds of thought processes, but I don’t think that you specified the right macro/micro interplay.
How useful a macro thought is depends on how useful locating a neighborhood in the search space is to you, and the level of abstraction that the assertion is at. If you get the right neighborhood, but lack the information necessary to use it, a small optimization might still be better.
For instance, I know that a good strategy for running a business is to create a service that’s valuable to other people, make an attempt to convince some focused group of people that it’s valuable for them, and then distribute it to them. This is much better than literally everything else that I’ve ever known as a business strategy before this, but it is not at a level of granularity that is actionable in any particular way.
Too much macro, and you might know that something should be possible but not how to do it, too little and everything you know how to do sucks. It seems like a more robust strategy is to be good at using macro level thinking to locate things that are likely to be very important, and then micro thinking to actually carry out a plan in that domain.
In this strategy, it seems like you would mostly be coming up with micro thoughts, just because any given thing you might do has more micro steps than macro thoughts. There are fewer abstract thoughts than detailed thoughts, and the search space is just much much smaller as you get more and more abstract, or macro.
I suppose we might also have different thresholds at which we call something macro or micro. Knowing that your tone of voice is a thing seems macro to me, but knowing that a particular tone of voice generally has this impact is micro. Something which is clearly micro, but probably not in general helpful would be something like reflecting on what tone of voice you should be use if this person brings this particular unlikely thing up in conversation.
This seems like a rewording/independent revelation of Humans are not automatically strategic which was a reply to A “Failure to Evaluate Return-on-Time” Fallacy, though it’s always nice to have important ideas articulated in a variety of ways.
I’m not sure that “structure of the search space” is quite what you want to be describing here, so much as “scope of the decision problem.” Yes, when you’re deciding what your morning routine will be, investing five minutes can result in savings of hours, spread over the year. But your morning routine is a small problem that is insulated from the rest of your life. The problem of your life’s goal is a huge problem that informs the rest of your life- a small change there could result in savings of years.
Are you familiar with the Toyota Production System? One of the major components (Kaizen) could be described as the idea that no improvement is too small to implement. Many small changes can add up to a big change.
More generally, it’s not at all clear to me why this post is “Macro, not Micro” instead of “Macro and Micro.” Several sections seem confused: is it a mistake to systematically search for such trades, or should such trades be incorporated into your thought process? I think you would do better with a post discussing integrating the perspectives than a post contrasting the two.
I completely agree that the two perspectives can be integrated (I even spend a paragraph discussing how, the one starting with “I also continue to think there is a place for micro-level skills...”). However, it is possible to micro-optimize successfully and still lose, but this is much less true of macro-optimization, so I actually do think it makes sense to present one as better than the other.
(There is a separate danger in macro-optimization, which is that it is easier to deceive yourself into thinking that a bad or neutral optimization is actually a good one. For classical examples, take going on diet X, or training yourself to sleep for only 5 hours each night.)
I think it’s about more than the scope of the decision problem. The techniques you use to optimize at the macro-level are fundamentally different and I really do mean in many cases “change the structure of the search space”. For many people, reading the Sequences had such an effect; or going to college; for me one of the most salient recent examples was working at Dropbox for the summer, which completely changed the way that I approached writing code. I didn’t decide to work there due to the output of any explicit optimization, though. It was “this has many characteristics in common with the sort of thing that will end up changing my life; so, even though I can’t give any specific mechanism as to how it will actually do that, I’m going to work there anyways”.
Re: Toyota; the difference between Toyota and a human is that Toyota makes millions of cars each year, so even saving 1 penny on each car manufactured is worth tens of thousands of dollars to the company.
I think it’s pretty possible to macro-optimize successfully and still lose. All you have to do is know what to do and not how to do it.
It’s not clear to me that this is the case. The parts of the problem that are hardest change, but the problems are fundamentally similar in a deep way. I will agree that the training to improve macro-level optimizations and the training to improve micro-level optimizations have different focuses, and am working on a post that will be relevant to the former.
Toyota makes cars, their macro optimization is already set. Imagine if they were using something other than the assembly line and just focused on micro optimizations within an outdated macro. They would run out of business within a year whereas if they used outdated micro optimzations they’d just experience a relatively mild loss
Toyota is, among other things, the descendant of Toyoda Automatic Loom Works, and the Toyota group is a Keiretsu.
More generally: it is not my experience that micro optimizations and macro optimizations compete significantly. Yes, people have limited attention, but macro optimizations make micro optimizations easier, and the reverse is true (but more weakly).
I get it man, and I absolutely agree.
As a quick sidenote, the fact that there are patterns in micro functions that contain answers to the macro is something that things like Freemasonry are built off of (from what I’ve read of it at least). The first degree itself is about using different mason tools symbolically to help with everyday life.
Anyway, yeah, there is a heirarchy of logic that we should take into account so that we can learn what to apply ourselves to in the first place, and from there how to apply. You’re definitely right in saying learning to be social is an important semi-macro application. Money may be the most tangible form of power, but people (the ones who give money its power in the first place) are the biggest source of power.