Let Your Workers Gather Food

Crocker’s rules apply to this post, and to everything I post.

Also, after writing this post and googling LW for links I came up with this post, which presents the same ideas. #!@$%


When I was younger, I often played real-time strategy (RTS) computer games. These usually involve running an empire successfully enough to conquer all the other empires. To win, you would have to gather resources like food, stone, wood and gold for use in research, construction and recruitment. Gathering resources is done by worker units. To construct worker units you need food.
Do you see the hack?

I know, the title gives it away. You can construct a bunch of workers and tell them all to gather food. Use the food they gather to make more workers, and put those on food as well. You set up a positive feedback loop and quickly have vast numbers of workers. When you have crazy amounts of food, then you can get to the business of putting lots of workers on each other resource, and using all your resources to take over the world.

Of course, among RTS players this isn’t a new idea, and various forces have arisen to counterbalance it. For instance, players build up small soldier squads and attack right at the start of the game, destroying any player who has only defenceless worker units. Marginal costs of worker recruitment increase with the number of workers you have. There are population limits. But the initial development boom still plays an important role, and the key to winning is often to balance those actions which help directly (building an army) with those actions which help with actions which help directly (making more workers).

Now consider this. If you want to, say, ensure we’re not all dead in 100 years, what do you do? You could become a fireman and save a few lives. Or you could donate to some worthy organization. Or you could get other people to donate to said organization. Or convince people to convince people to donate to the organization. And so on. That’s one orbit under the meta function, but it’s not the one I want to talk about.

Say you decide to throw money at a worthy organization. To do that you need to get money, and to get money you need time. How much buck-for-the-time you get depends on how efficient you are at converting time into money. But time isn’t just useful for conversion into money. It can also be used to increase your efficiency of converting time into money. Or it can be used to increase your efficiency at converting time into [efficiency of converting time into money]. And so on. Do you see the hack?

Use the time you have to get better at using the time you have. You set up a positive feedback loop and end up crazy awesome. Human go FOOM. You might spend some time learning to efficiently manage your time, giving you more free time to work at your goals. You might spend some time thinking about how to manage akrasia and thereby create more quality work-time. You might research how best to learn, and chance upon SRSs. You might learn about nootropics. You might even stumble upon this very site, where you might pick up pointers to things you can do to get better at using your time. Now, of course, you can’t just spend your time becoming more and more awesome. At some point you need to actually use that awesomeness to do what you originally wanted to. As in the RTSs, you need to balance actions which accomplish stuff directly against those which accomplish stuff indirectly.

In other words, you need to balance the various levels of action. To summarize, level 0 actions are those which directly accomplish your goals. Level k+1 actions are those which help make level k actions easier or more effective. As the linked post observes, lower levels tend to be additive while higher levels are often multiplicative or better. Level 0 is useful. Level 1,2 and 3 are much more useful. Level 123 is pretty useless. Sure, travelling by horse gets you places, but having one person invent a plane which a billion people use will cut travel times significantly. Observe that this very site is about going meta, about thinking about thinking, and often about thinking about thinking about thinking, or even more. Also, work can be on many levels at once, and it’s not always easy to figure out the level of an action. For instance, what’s the level of you reading this paragraph?

So we want awesomeness explosions, and to help bring about awesomeness explosions we need to know a bit about them. How can we act to make ourselves explode? What determines the speed of an awesomeness explosion? Is awesomeness capped, and, if so, what’s capping it?

To start, what limits awesomeness? We might ask whether we can extend the analogy with worker recruitment in RTS games, and indeed we can. In RTS games, worker explosions can’t go on forever because:

Marginal worker costs increase with the number of workers built.

There are population limits.

Sooner or later, other players will rudely just up and attack you, and your defenceless workers will die.

In a serendipitous confluence of circumstance, these three limitations on RTS fooms map nicely to human fooms. Respectively:

Different fruit hangs at different heights, so picking low-hanging fruit makes the average fruit higher. In addition, as you become saner you become less neurotypical (Not every change is an improvement, but every improvement is a change.), so it’s harder to use established human knowledge about self-improvement to improve yourself.

There are limits to human knowledge. There are built-in limits on brain hardware. Note that both these can be overcome by the sufficiently awesome. They’re also both special cases of the above bullet-point. Just as population limits are a special case of increasing marginal worker costs, in yet another serendipitous confluence of circumstance.

Your levels of work must be grounded. If you work on level k+1, make sure you do enough work on level k to justify it. (There’s an analogy with Truly Part Of You.)

With all these factors limiting human fooms, there’s no guarantee human fooms will be, well, FOOMs. They might fizzle out too quickly, due to increasing marginal costs of awesomeness. But my mental intuitive estimation machinery says that, while taking over the world might be pushing it a bit, a lot more is possible than we’ve achieved so far.

Now for some ideas on making yourself go FOOM.

Do useful stuff! People aren’t automatically strategic. I think this is the second-most important reason we haven’t all foomed yet (after akrasia). Remember that compound interest isn’t magic – just improving yourself isn’t enough. You have to actually assign higher priority to things which are more important. This is a really important point, and all the rest of these bullet-points are special cases of it. I think those in the LW mindspacecluster are particularly prone to seek knowledge without first applying some ruthless pragmatism. I know I am.

Do stuff which helps you do stuff. In other words: work on higher levels, as long as you’re still grounded. Go meta, like this sentence (whose metaness is too great for even the ordinals). Trying to learn from that physiology book isn’t very useful when you haven’t learnt how to learn. Again, this is a really important point and all the rest of these bullet-points are special cases of it.

Write down your thoughts (Darwin), and preferably ensure you remember them with an SRS. Record your time use every now and then. This is an important.

Explore new mindstates. Trying to come up with ideas seems like mining for diamonds, and often you can get more, bigger diamonds by mining in different places. One reason this is a good idea (and a reason why you should write down your thoughts) is that people (or me, at least) often seem to retread the same thoughtpatterns over and over, day after day. It’s like you have a ‘reset’ button that gets pushed every evening when you go to sleep. I have directly observed this, when I wrote down my thoughts for a few weeks without memorizing them with an SRS. When I write down my thoughts, I’ve taken to calling this Every Day The Same Dream (EDTSD) syndrome.

Look at what other really smart people do, then consider doing that. Look at really smart people and ask yourself why they haven’t taken over the world yet. (David Bennett: If you want to beat the market, you have to do something different from what everyone else is doing, and you have to be right.)

Think a lot, and think in efficient ways. Most people seem to just hope good ideas will tap them on t he shoulder. I’ve been like that for most of my life. Ideas often do tap you on the shoulder, but I get better results by sitting down with a piece of paper and a pen and thinking really hard, vomiting anything that comes to me out onto the page. I call this the Thinking Really Hard (TRH) technique, and I Think it was inspired by Eliezer’s exhortation to sit down and Think for 5 minutes before concluding a problem is unsolvable.

Related to the previous bullet-point: Ensure you focus mental energy wisely. If you really spent all your mental energy where it’s optimal, how much more would get done? Perhaps actively stop yourself thinking about things you don’t care about. Go meta: think about how to improve your thought-focussing abilities.

As a special case of the previous bullet-point: Think long and hard about how to get more time. I really mean that. Perhaps you should add a reminder to regularly do that to your SRS, if you’re awesome enough to use one. Remember the Pareto principle.

As a summary of all these bullet-points: Figure out ways to work faster and smarter and harder. I have lots of ideas about overcoming akrasia which are pretty weird and which I’ve never seen described anywhere else, but which work spectacularly for me. Of course, that doesn’t mean they’ll work for everyone else, but they’re bound to work for some of you. I might write a post.

A few links, which most LWers are probably already familiar with:

The Science of Winning at Life

SRSs: Gwern Branwen, Piotr Wozniak

Nootropics: ImmInst 1, 2

Humans Aren’t Automatically Strategic

All the posts about akrasia

more?

What determines the speed of human recursive self-improvement? The main factor, I think, is how much new awesome you get from a given amount of awesome – the rate of compound interest on awesome. If being awesome causes you to become much more awesome, you will foom quickly, whereas if you get only a bit more awesome for each unit of awesome you have, you will foom slowly. If there’s a set-point of awesomeness towards which you are attached like a spring, it will be very hard to foom. Each of these three situations often occurs in real life, for different types of awesome.

I’m pretty sure that several people here on LW have had these human fooms, since, well, they’ve found LW. I’ve had a miniature human foom, and it’s still ongoing. But it seems to me that this is nothing compared to what’s out there.

As with those of fooming AIs, the actions of fooming humans are hard to predict, and for the same reason: if you could predict what they’d do, you could probably do it yourself. Nevertheless, here are some ideas of what people far on in the fooming process might do:

They would practise extremely fine-tuned control of their own thought-processes; they would waste no thought-time. They could just sit and go into a thought trance, coming up with a brilliant new insight in seconds. When most people think, they’re just executing adaptations, not optimizing utility.

They would be free of cognitive bias.

They would have the ability to flat-out ignore pain. They would do everything the way cold, hard logic says is most efficient. They wouldn’t ever sit, they would stand or run. They would run on a treadmill on one leg while listening to a French audiobook (despite not knowing French) while juggling 5 tennis balls with one hand while doing SRS reviews.

To restate the previous bullet-point, they would have no akrasia. They would find those little voices at the backs of their heads that keep whispering for them to fail. They would drag out those little voices and kill them.

Now, these things look unrealistic. But I think they’ d all be achievable by any average LWer who committed themselves to this, and only this, for a year. I really mean that. And I have a feeling that more, much more, is possible.

The whole universe sat there, open to the man who could make the right decisions.
Frank Herbert, Dune, as quoted by Nick_Roy