[Previous] Stories and Actions | Home | [Next] How Badly Run Are Cryptocurrency Exchanges?

Hardness, Emotions, Mental Automation

On FI, someone keeps asking how to feel that one is overreaching. Shouldn't there be an emotion to tell you what's going on?

No. Emotions are software, developed by our culture (mostly thousands of years ago). It's kind of like: every minute, or when something notable happens, that software runs through a checklist of 5000 things. If one is going on, it runs through another checklist with more detailed stuff for that issue. The results of the software analysis is presented in a short summary which we call an "emotion".

Our emotion software doesn't know about all situations or issues. It knows about a lot of stuff but not everything. There aren't emotions for everything. There's no reason it'd have to cover everything. It's not complete. Even if people had ten times as many emotions it still wouldn't be complete.

If there was an emotion for everything, emotions would be useless. A summary has to condense information. Emotions are focused on areas our culture considers important. They're selective. They prioritize. They help direct our attention to what matters. If there were emotions for everything, they wouldn't be a summary anymore, and they wouldn't be worth paying attention to. You'd have to have a second layer of software that screens the emotions for the ones you consider important and ignores the rest. If you want information about everything, use your eyes and ears, they are better at doing that type of thing (you won't see infrared but eyes are much less about summarizing than emotions are, they are more about giving you a reasonably complete picture of what's in front of you).

The whole concept of expecting an automatic response or indication of a situation, whether emotional or not, doesn't make sense. You have that for many things but not all things. Such responses have to come from somewhere, they don't just exist automatically in all cases.

With that out of the way, the best indication of overreaching is hardness. When something feels hard, maybe you're overreaching. Feeling confused or overwhelmed are other relevant indicators. These aren't about overreaching but they have some overlap. They're in the right ballpark.

The feeling of hardness indicates inefficiency.

People are confused about hardness partly because there are two types. Hardness-A is an evaluation of the issue in general. Hardness-B is the feeling, it's about a person's experience. Hence people do hard things and say "it was easy for me". Touch typing is fairly hard. It takes practice. It's a skill you have to develop. But it's easy for me. I do it without thinking. It doesn't require conscious attention. I used to experience it as being difficult, but I don't now.

The same is true of walking. There was a time in my life when I couldn't walk and I had to learn how. Most ways of using leg muscles do not result in walking, they result in falling. Only some specific actions succeed as walking and staying balanced. And people have tried to write software to make a robot walk around and they've found that's difficult.

Speaking English is hard. There are thousands of words to learn, each with a spelling and a pronunciation and at least one definition. There are grammar rules and exceptions. There are different forms of words with similar but different meanings, e.g. "hammer", "hammered", "hammering", "hammers", "hammer-like". But once you get used to English enough, it can feel easy, intuitive and second-nature.

It's the same with chess. The better you get at chess, the better you can autopilot it and play without trying very hard. Chess players have a skill level they can play at when trying hard, and a separate, lower skill level they can play at when taking it easy. The skill level for taking it easy isn't usually far behind – a top player can easily beat a good player. In other words, trying hard doesn't make a big difference. A top 1% player when trying his best is still a top 2% player when not trying very hard. Trying hard at chess does make a big difference when playing serious tournament games against players of similar skill, but it wouldn't make any difference against most opponents.

Trying hard is a big effort for small results. That's an inefficient use of effort. 10% of the effort lets a chess player achieve 90% of his maximum skill. Then trying ten times harder only adds a one ninth increase in skill. It makes sense to try hard in a competition where only the best player wins and everyone is trying their best, but it usually doesn't make sense to do that in life in general. And all this applies to many other examples besides chess.

Trying hard means you're not autopiloting. You're using conscious attention, which is a limited resource. You're using focus and mental energy and creativity and that kind of stuff. And, in general, people can do that for around 2 or 3 hours per day. They can do more in the short term but it leads to burnout if they keep it up over time. (For knowledge workers, the 8 hour work day is a myth. For chess players, they focus longer when competing, but they don't compete on most days.)

Trying hard means working at the edge of your abilities with less margin for error. It means more mistakes are made. It means you're trying to do things that you don't know an easy way to do. It means trying to do things you can't do habitually, can't do while multitasking, can't do while not at your best.

Most of our life is automated. Our minds are complicated and do tons of stuff. Our consciousness is like a factory manager who can go around and inspect any one workstation at a time and make changes, but the work in the factory always keeps going everywhere else. Hard stuff is stuff that only the manager can do, there's no workstation to do it. Doing hard stuff means the manager is busy and can't go around inspecting and improving the workstations in the factory, nor creating new ones. Some people don't notice the loss because their manager (conscious mind) is usually mostly idle anyway, rather than going around checking for problems. If you have a lazy manager who wasn't going to do much anyway, then keeping him busy doesn't appear to have much downside. It's still bad though: a busy manager isn't going to reform. Keeping the manager distracted from the ongoing, unsolved problems is not how to change things so that he becomes a better manager.

In general in life, you need to figure out easy, repeatable, low-error ways to do things, then automate them (add them as workstations in your mental factory that can keep producing even when the manager isn't there). Adding more of those is how you get a lot done in life. Having your manager do any work himself is a huge loss of productivity. Your manager can do the work of, like, three workstations. Maybe even ten. He's really good at stuff compared to the automated processes (which you can think of like robots or low skilled workers). But in the long run, having your manager do the work of even a hundred workstations is a terrible deal. It makes way more sense for him to help set up thousands or millions of workstations. Much more will get done if he doesn't do it. (Also, remember, if the manager works more than three hours a day, that's overtime and he starts getting tired.)

Stuff feels hard when your manager has to do it instead of a workstation doing it. You don't know how to do it using only the sorts of cheap, plentiful mental resources that form automatic workstations.

People are confused because having their manager do something is a common step in workstation setup. First you figure out how to do something using conscious attention and maximum focus. If you can do it at all, that's a good step one. Then you figure out how to do it more easily and reliably. Then you get good at it to the point it's easy and automatic/intuitive/second-nature.

But doing something for the purpose of learning and setting up an automatic workstation, and doing it to get it done, are different things. The goal matters to how its done. Like, is the manager taking notes on how he does it so that he can then hand off the job to an unskilled worker later? Is he looking for what could be automated, as he goes along? Is he trying to figure out how to break down the task into small, simple parts that could be handled by dumb workers or robots? Doing those things helps work towards a day when the manager can stop being involved and delegate everything – which means he can move on to new projects.

On the other hand, sometimes people think they will only do something once, so they don't worry about any of that stuff, they just try to get it done and even getting the manager to do it a second time would be hard (they have no notes, they don't even know exactly what they did, they just fiddled with stuff until it worked and they lost track of some of their actions).

And sometimes people think automating something is too hard, so they won't bother. Right now, the easiest thing to do is get it done without worrying about the future. Figuring out how to automate stuff is extra work. Then they do the same thing again the next day, and the next, and they keep wasting manager effort and never get a workstation created. (This is more common with things that come up sporadically, e.g. every few weeks, but sometimes people do it with daily tasks.) Or sometimes people partially automate a task, e.g. typing, but they never fully automate it, it's always a bit of work and a bit distracting.

People talk about inspiration and perspiration. But it should be inspiration and automation. Instead of working hard, figure out how the great new idea can be done easily and repeatedly.

A big obstacle to automation is errors. Every time something goes wrong, the low skill workers or robots at the workstation can't do much troubleshooting. They aren't very creative. They'll go through a checklist of troubleshooting steps if their manager told them to (that's highly recommended!). If that doesn't work, then either the manager has to come along and fix things (like if a machine is broken), or else they can throw out that work product and start over (if only half of the things the workstation produces actually work, it can still produce stuff, although there better be some quality control steps to actually find the broken ones and get rid of them).

Automating requires figuring out how to do stuff in a highly reliable way with a low error rate. You have to figure out not just a method to accomplish a task, but an easy, reliable method that doesn't have many ways to fail. If every step is easy, and that are steps to check for problems and standard ways to fix them, then it can work pretty well and the manager doesn't have to be called in very often to clean up a mess. That's good if your goal is to get millions of workstations running, with just one manager, so that you can get a lot done in life. And yes millions is realistic.

Your brain is a computer. It's a more powerful computer than my iMac. My iMac can do around four billion CPU cycles per second, and each cycle can get several small tasks done. If the average workstation involves a million small tasks to complete one work product, and I have a million workstations, then they might all be able to average a work product completion every 10 seconds, while all running simultaneously. That's the ballpark of how powerful the brain is. And it's better to have a billion workstations and turn some on and off – some are general purpose, but most are only used when doing a specific kind of activity, e.g. a workstation that is only used when playing or thinking about chess. (Figuring out more general purpose workstations helps keep things manageable – it means you need fewer total workstations and you can get stuff done with fewer running at once. Thinking in a more principled way can mean a hundred million workstations, with a million on at a time, instead of a hundred billion workstations with ten million on at a time.)

Elliot Temple on February 3, 2019

Messages (4)

Kate asked about this post on FI:

> I think the meaning of hard/easy used in the statement is the second one, i.e. hard/easy for me (now).

> Whether or not something is also considered inherently hard doesn’t matter. The key is whether it’s

> currently hard *for you* — whether your manager is going to have to do it.


> It’s still unclear to me whether “only do things which are easy" is suggesting that people not try to fix

> irrational thinking methods or figure out how to use FI if they consider those things to be hard.

there is a learning/doing distinction. first you learn how to do something, say dentistry, then you do it (fill cavities, etc). so one of the meanings is you should learn enough that dentistry is easy before it's your job. don't learn enough you can do it, learn enough that it is now easy. dentists should have *mastery* so they can do it with a low error rate (and VERY LOW rate of major errors) even when tired, distracted, unfocused, etc.

and also the learning process shouldn't really be hard. say you're trying to beat a level in a video game. if your goal is "beat the level" then that's hard. but that's about doing, not learning. if your goal is "try strategy X and see if it works or not", that could be a step towards learning to beat the level which is also easy. if strategy X is too hard, then you could have easy immediate goals like "try action X1" and "try action X2" and so on – try out individual parts of it before trying to do the whole thing.

in Vindictus there are lots of boss fights you can do by yourself and get a gold medal for being hit 3 times or fewer. success is hard in some sense. but the learning process doesn't have to involve hard steps. first you can just stand there and let the boss hit you and watch what he does. that's easy! after you watch a bit, you can start to figure out what his attacks are. lots of the bosses only have like 5 different attacks. if remembering is hard you can write them down. you can even record video clips of each attack. that's more work but it isn't *hard*. so this step of seeing what the attacks are can be pretty easy, especially if you aren't rushing yourself. like if you are trying to remember every attack after you've seen it once, that's hard. but if you take your time and are OK with remembering an attack after seeing it 10 times, then it's not very hard.

the next step is blocking/dodging attacks (each character in the game has a few defensive options, mostly dodges and blocks). you can figure this out without doing anything hard, too. for each boss attack, try your first defensive move at various different timings. you can get a good idea of the right timing by letting the boss hit you and seeing what you take damage. your defensive move should generally be used around .5 seconds before the time you took damage, though it varies by move. if this isn't working well, try your character's second and third defensive moves and see if they work better for dealing with this attack.

many boss attacks have multiple parts. like they swing their sword 3 times in a row, and it's a set pattern of those 3 swings. so you can figure out a series of 2-3 defensive moves to defend against all 3 sword swings. (sometimes attacks come close together and you can stop multiple attacks with one defense.)

for each attack, there is some kind of clue that it's coming. the main clues are animations like a boss moves his sword or shoulders back before swinging forward. you see them getting ready to attack in some way. so you also need to learn some kinda thing that you will react to – the signal that it's time to do that defensive pattern for that move.

this can all be done pretty intuitively but it can also be done by conscious design and you write a list of every attack, every signal its coming, and what defensive moves you plan to use for that attack.

which part of that is hard? no part. if you do it in this methodical way, every part is easy. it's not like you need super fast reactions times. the game isn't hard in that way. if you calmly watch for the signal that a specific attack is coming, and you aren't worrying about anything else, then you can block/dodge it with a bit of practice, it's not that hard (and if a different attack happens first you just let it hit you and wait until the boss does the one you're trying to stop).

the individual parts of the game aren't that hard. but the complexity adds up when you're watching for 10 different possible attacks (on the harder, more complicated bosses) while also doing your attacks and also there are other allies on your team who the boss might target (if the boss does a move aimed at you, or aimed at a guy off to your right, then the patterns of blocks and dodges that protect you, and the timing to do them, can be different. where the boss is aiming changes where his sword ends up at different times.) and also you can be remembering to drink health potions every 4 seconds and use your cat statue every 70 seconds and tracking how much SP you have (points for doing special moves) and then managing which special moves to use, when, and so on. and then your ally dies and you want to go resurrect him but that requires standing still for 3 seconds so you have to find a safe time to do that between boss attacks. etc.

but basically all of that can be learned as a sequence of easy steps, too.

once you learn to defend attacks you practice until it becomes more of an automatic habit. you get it to the point its easy to do all the attacks for a boss, it's second nature, its intuitive, your error rate is low. then you try attacking in between the bosses attacks. you'll already have a sense of how much downtime there is after which attacks since you've seen them a bunch. so you can estimate how big of an attack you can fit in after each boss attack, and you try it out and see what works. that's assuming you can already do your attacks easily. if you can't, no problem, you just practice attacking without worrying about defense (initially just do this in an empty area with no enemies). and then practice on easy enemies where getting hit isn't a big deal, so even if your error rate for defense is high, cuz you're focused on attacking, it doesn't matter much.

before you actually use your attacking or defending as a skill – before you try to DO it for real instead of doing it in a learning/practicing context – you need to get it to be easy, you master it so an automatic mental workstation can do it. so by the time you're trying to kill the boss, you have all the skills needed to do it, and it isn't scary or hard like it would be if you just went up to him the first time and tried to win.

and after you practice, you still don't expect to win. if your goal was to go straight from practice to success then that'd be hard. instead, you practice and then you try fighting the boss for real as a test to see how well you do. you're checking how effective your practice was, what your error rate currently is. that's easy cuz the goal isn't "make no errors", the goal is "see how many errors i make". so you do the blocking and attacking in easy, automated ways, which is important cuz now your conscious attention is mostly used for just watching to see how often you screw up. that's not a hard thing to do! you just autopilot attacking and defending while consciously watching how well it works. that's it. ez. then you can see if you need more practice, and if so for which parts. and you can also identify problems like a particular strategy for blocking a particular move is unreliable, so maybe you need more practice or maybe you need to change the strategy – do a different defensive option or do the first block after an earlier visual cue. there are other errors you'll see happen, like a boss can have two different attacks that look similar at first, so you mix them up and sometimes you do the defense for attack 1 when the boss is doing attack 2, so it doesn't work. so while you're autopiloting and seeing how it goes, you can watch for issues like that with your conscious attention, and then you can figure out a solution, like you can look at the attacks more closely until you find a difference which is pretty easy for you to recognize once you know what to look for, and then you can start looking for that and, with a little practice, autopilot doing that. or you can also find a defensive option that works for the first part of both attacks, so it's ok if you don't know which is which until you're doing the second defensive move.

people find the game hard cuz they are trying to e.g. do lots of attacking right now instead of just focusing their attention on defense. or they never practice alone, they just fight in groups where other people are always moving the boss around and creating chaos, and everyone is rushing to keep up with everyone else on doing damage. and if you are just *less ambitious* in the short term, you can make tons of stuff way easier. i was having trouble with some bosses in the last two days and what i started doing is only using my simplest attack which takes the shortest time. that immediately solved the problem of doing an attack that is too long and then i'm not ready to defend against the boss's next attack. and it meant attacking took even less attention and i could focus on defense more. the downside is that the simplest, fastest attack does the lowest damage. but so what? a bit of patience made it way easier and actually saved time overall (cuz it takes longer to kill the boss, but fewer retries, so actually that saves time). that works great on bosses where my goal is to get the gold medal one time – if it's 5 minutes slower but saves some retries that's fine. i don't need efficient offense for a boss where i just want one good kill. there are other bosses that you fight more often so you want to learn to do your offense more efficiently, but it's not needed in all cases. (also part of the issue is some of the old bosses i was fighting, which i only needed one good kill on, actually have different designs than some of the modern bosses that people fight more. some of them actually have overly short windows for you to attack during if you are playing alone. it's fine if you play with an ally cuz then half the time the boss attacks the ally and you can just go stand behind the boss and have time to attack. but for certain heroes, soloing some of the old bosses involves shorter attack windows than you're used to with the modern bosses, so partly you just need to be willing to use your small attacks and be content with that. and if you had to fight that boss every day it'd be annoying, but you don't, and the newer bosses you fight more often have some larger downtime parts built in, on purpose, to let you do your big attacks sometimes.)

the point of this example is if you approach things step by step, every step can be easy. cuz you have a specific goal in the current step which is *not* big picture success, and you just do that.

curi at 11:52 AM on February 4, 2019 | #11779 | reply | quote

There are lots of videos available of me learning Vindictus stuff on my Twitch, e.g.:


In this video I practice and gold medal 3 bosses, in addition to playing a bit of other stuff.

The first one is a yeti. I'd practiced him some before. I came back to him after doing other bosses more and getting better at playing in a consistent reliable way. And also I figured out a solution to the hardest part. The move where he jumps in the air and then stomps a bunch can hit like 5 times and then he can attack again right after. So you have to keep blocking during the move. The timing window there to stop blocking and start blocking the next move is short and I was not reliably getting the timing right. Note that my block (the sphere around me called mana amber) lasts 3 seconds maximum, I can't just keep blocking indefinitely, I have to stop and restart my block sometimes. So I started actually counting out how long the stomping lasts so I'd know when to restart my blocking. That worked well. Additionally I figured out that I can safely let go of block a little bit before the stomping animation ends. You have to keep blocking through like 90% of the move but not the whole thing. i think that's because the move is divided up into like 6 different hits. like you can be hit 6 times max. and once you block the 6th one you're safe. cuz the way the game usually works is if the boss has an attack that lasts 1 second, once you block it you're safe for the entire attack, and you don't have to worry about blocking again until there is a separate hitbox. there are pretty long attack moves where you just block it at the start and then you can attack during it cuz you're immune to the rest cuz it's all the same move and you already blocked that move. and with the yeti stops, they are small moves but i think they work that way so that one i block the last small stop i'm immune to the small amount of it still to come (the alternative is the last bit of the animation just can't hit you. i could test by standing back and walking into it right at the end and seeing if i get hit or not, but i don't really need to know that.)

curi at 12:41 PM on February 4, 2019 | #11780 | reply | quote


> yeti stops

yeti stomps

Anonymous at 12:59 PM on February 4, 2019 | #11781 | reply | quote

Today's twitch vod is mostly just Ein Lacher (solos trying to get hit 3 times or fewer). I spent a long time doing Lionotus on staff then did a bunch pretty quickly on Fiona then did Weeping Queen with staff and needed like 6 retries. Also I topped damage in RAR spiders at the very end. https://www.twitch.tv/videos/375435535

curi at 12:54 AM on February 5, 2019 | #11783 | reply | quote

Want to discuss this? Join my forum.

(Due to multi-year, sustained harassment from David Deutsch and his fans, commenting here requires an account. Accounts are not publicly available. Discussion info.)