[Previous] curi's Microblogging | Home | [Next] Less Wrong Banned Me

Elliot Temple on September 13, 2020

Messages (30 of 276) (Show All Comments)

#19926

> I specifically ask for help on this issue. I don't know if you're intending to be helpful or not

yes, I was trying to be helpful. I wouldn't have bothered commenting if I wasn't trying to be helpful.

Your response was long and detailed but failed to actually engage with most of my points, so it was difficult to respond back to. You also seemed defensive, and doubted that I was even trying to help you, despite you opening with "Thank you!"

#19941

> I think the project was successful

what were your initial goals? what do you think you succeeded at?

In your first blog, you said stuff about wanting to have a better understanding of women. From what I did watch and read, I didn't think you had a good understanding of my position or experience, or that of other women I know. (Not saying you should continue the project, btw - I don't think you were doing the project in a way that would help you gain an understanding of women either.)

I also don't really think you are interested in discussing it, or interested in the crits that I have. (Not saying you should be - not everything has to be a high priority for you.)

And, again, I am writing this as an attempt to be helpful.


a girl anon at 2:14 PM on February 11, 2021 | #19944 | reply | quote

finished *it's not luck*

I finished reading *It's Not Luck*. This is the 4th Goldratt book I've read (the others being the goal, critical chain, and the choice).

I like it. I found the audiobook was lacking in some parts b/c there are some important diagrams early on and about half way through. I'm going to review those sections particularly b/c I want to make sure I understand the methods and can replicate them. I looked at the diagrams (I have the print version too) but didn't follow them closely.

I like the end; there's like a quick-fire recap of the methods used. There's some info that was new (or at least I didn't spot it earlier) like: goals have obstacles; big goals have intermediate objectives (which are in some ordered dependency graph with both parallel and serial parts); each intermediate objective should address some obstacle. That sounds pretty common sense, but the related method is the simplest and clearest method of planning I've heard. (the main part of the method that's missing from that list is the "what are you going to do about it?" question/answer.)

I also like the company strategy stuff in the last few chapters, especially the idea of deliberately leaving excess capacity in market segments to keep a company flexible. The idea is that company resources are general/flexible across multiple segments, then you can prioritise resources to service high-profit segments, tho make sure to always leave some excess capacity in each market segment. When market segments lose demand or become unprofitable, you can always move to another market segment b/c you left excess capacity (i.e. didn't full exploit that segment). It also means that you can avoid over-saturation of a market becoming a bottleneck (which could also mean that a company becomes like over-specialised for a segment; i.e. loses flexibility).


Max at 11:30 PM on February 11, 2021 | #19951 | reply | quote

#19944 I want to think about what I should say before I reply fully. I agree that I was defensive. I think some of my greatfulness was genuine, even tho I was a bit defensive. There's a conflict/tension there tho I'm not fully sure of the detiails yet.


Max at 11:34 PM on February 11, 2021 | #19952 | reply | quote

right/consistent/fast <-> eureka/chewing/integration

the steps of learning (do it right/once, do it consistently, do it fast) work well for explaining physical tasks like speedrunning, and some think-work like coding (where those 3 steps are sometimes said: make it work, make it right, make it fast). but how do those steps work for learning more abstract ideas (e.g. calculus, theory of constraints, philosophy)

I think the first way of thinking about the 3 steps can work for calculus (but it's harder to see for ToC or phil). With maths you generally get shown a method / proof / formula / etc and then get problem exercises of increasing complexity, and sometimes challenge questions. the exercises are usually: some basic ones first (completing them shows you can do step 1), then some complex ones (completing them shows you can use the idea consistently), then some more complex ones (completing them shows you can use the idea consistently enough to mix with prior/other techniques, mb with some creative thinking). challenge questions push your ability to use the idea (consistently and quickly) s.t. you can use the idea in creative new ways with previously learned ideas (this helps to create/improve structural knowledge). the progression of exercises maps well to correctly, consistently, quickly.

ToC and Philosophy ideas are harder to relate to the 3 steps. I think, roughly, the process follows the same pattern but doesn't make sense with those steps as they're currently named. WRT abstact ideas, what does "do it once" mean? excepting analogies to things like pyramids and structural integrity (like, engineering), how do the ideas of "consistency" and "speed" relate? They *sort* of have natural ~connotations, like consistency <-> 'know how/when to use the idea', and speed <-> 'idea has low overhead / it's intuitively able to be mixed with other ideas'. Even those ~connotations are a bit of a stretch, tho.

I think these steps are a good ~translation for the original steps: realise the idea (eureka), chew on the idea (so that you get a fuller understanding and develop structural knowledge), and integrate the idea (use the idea in new ways and in combination with your everyday thoughts). These steps don't necessarily happen all at once for a ~complete idea. Like "theory of constraints" is a big idea made up of lots of smaller chunks, some of which aren't documented in Goldratt's *The Goal*.

How can we still learn complex ideas, then? When we try to learn complex ideas, we learn them incrementally with smaller ideas. Some ideas are more easily learned than others, which can be due either to better (or more fortunate) structure, or because they're smaller ideas themselves (fewer components), or because they overlap a lot with stuff you already know. When ideas are big *and* unintuitive it means that there are a lot of new concepts that you need to learn all at once. That's hard. Sometimes it get's easier if you realise an instrumental goal that let's you use a subset of the idea (e.g. ToC applied to factories, rather than the full ToC). You can't rely on that, tho; a shortcut like that isn't always available.

The biggest idea you know is basically your full mind, and it's made up of lots of smaller ideas (sub-ideas), which are themselves still massive and complex compared to their sub-ideas, etc.

Learning complex ideas can have a long first step, and might involve some chewing on smaller ideas. When you get the big eureka moment it's because you've integrated the smaller ideas. I think the chewing process for the big idea will help work on integration for the smaller ideas, too (you can still have work to do with the smaller ideas *after* you start using them).

Note: I *think* I am using 'chewing' and 'integration' in the objectivist sense, but IDK. If I am not I would appreciate someone letting me know.


Max at 1:04 AM on February 12, 2021 | #19953 | reply | quote

a small contribution to (the discussion/work that is) CF's integration

DD says in BoI something like: some subset of objectively true ideas implies all other objectively true ideas. This make sense; if there's some ~objectively true and complete body of knowledge to describe reality (infinite or not), and progress is moving us towards that body of knowledge, then things that are ~true (in the everyday CF sense: they work) will overlap more and more, like they will imply parts of each other and they can be ~easily integrated.

one way that i thought about this after reading BoI was the idea of a "stand alone complex" (the subtitle of s1 of the anime *ghost in the shell*). The way i imagined things in my mind was like a graph of ideas/explanations that connected to each other. those idea-nodes only connect if they don't contradict. there are like 'bubbles' of ideas that are compatible with one another; all the bounded ones are parochial, but the single unbounded one (finite atm but forever growing and unbounded) is like the 'biggest self-supporting collection of ideas'; the "stand alone complex" (SAC).

integration is like creating stronger interconnections between the idea-nodes in that graph, and reducing the number of nodes via ~unification (not that this lessens the ~magnitude/measure of the SAC of ideas).

Goldratt has a lot of good ideas in his books, and concepts like 'excess capacity and how it relates to constraints' helps us understand the core of falliblism: that we can still 'know' things even if there are problems and ultimately all ideas will be refuted by ever-better ideas. it helps us answer why it's okay to use newtonian gravity in a toy simulation.

This sort of thing has already been integrated and expanded upon by IGCs, breakpoints, yes/no, and curi's other work.

I wrote this in part because I am continually surprised by the convergence between ideas as I learn more about CF and it's antecendents. I love it; it's great.


Max at 1:24 AM on February 12, 2021 | #19954 | reply | quote

#19954

> Goldratt has a lot of good ideas in his books, and concepts like 'excess capacity and how it relates to constraints' helps us understand the core of falliblism: that we can still 'know' things even if there are problems and ultimately all ideas will be refuted by ever-better ideas. it helps us answer why it's okay to use newtonian gravity in a toy simulation.

Can you explain more about how “excess capacity and how it relates to constraints” helps us understand “that we can still 'know' things even if there are problems and ultimately all ideas will be refuted by ever-better ideas”? That connection sounds interesting but I don’t get it.


Anne B at 6:06 AM on February 12, 2021 | #19960 | reply | quote

Makeup Mastery?

Max, do you think you achieved *mastery* of some significant, new things in your makeup project? If so, you could list those. If not, I think you should have higher standards and stop overreaching.

Can you self-evaluate the correctness of any of your new makeup knowledge with similar confidence to your self-evaluations of counting to three or judging whether the word "with" is spelled correctly? Those are examples of what mastery looks like.

The same goes for all your other philosophical work. Keep it simpler. Practice things. Aim for mastery. Aim for a low error rate where correct criticisms are uncommon, surprising and treasured.

Consider what you do have mastery of and build on it. Plan out projects intentionally with goals and trees, keeping issues like mastery and overreaching in mind.


curi at 12:35 PM on February 12, 2021 | #19963 | reply | quote

#19963 Where are the 5+ successful past projects at 90% of the size, complexity and difficulty of the makeup project? And at 80%, and 70%, and 60%, etc., all the way back incrementally to simple projects like crawling to a location as a baby.


curi at 12:51 PM on February 12, 2021 | #19964 | reply | quote

#19964 You don't have good examples of what success looks for to compare your project to. There's a huge gap from the makeup project to your most similar projects that are clear, confident, decisive, unambiguous successes.

And these are not new things that I'm saying.

Start way smaller, get quick, clear wins, and iterate. Start with multiple successful (micro) projects per day. Finish 100+ in a month with a not-decisive-clear-success rate under 10%. Establish a baseline of what you can do that way and get the iteration started.


curi at 12:57 PM on February 12, 2021 | #19965 | reply | quote

re: Makeup Mastery?

#19963

> Max, do you think you achieved *mastery* of some significant, new things in your makeup project?

No.

I think I got better at some things (some makeup related, some video production related, some people/social related). Even if I gained mastery over those things, I don't think they're *significant* compared to the size of the project or the project's goals. I think that relative scale matters, e.g. mastering teeth-brushing is significant if the project is like 1-2 hours long (mb split up), but it's not significant if the project is days or weeks long or way bigger in scope.

It's not worth me spending 30+ hours on something just to get a relatively small skill like doing video editing quickly (relative to filming + copying + rendering), or knowing whether lipstick is applied well/badly. It's nice that I have those skills now, but I could have used movies or stock footage and just practiced editing them to learn that skill in a few hours. (I don't think I mastered those anyway, but I am consistent in some ways.)

> If not, I think you should have higher standards and stop overreaching.

It just occurred to me that 'high standards' and 'high ambitions' are, in some ways, contradictory. Or mb not contradictory, just that it's easy to mistake the two and, when they are mistaken, it can be really bad for the project (i.e. it'd fail).

> Can you self-evaluate the correctness of any of your new makeup knowledge with similar confidence to your self-evaluations of counting to three or judging whether the word "with" is spelled correctly?

No.

Not for anything significant, at least. Super basic stuff maybe, but a lot of that is just common sense.

A case in point is my claim that Mecca did color matching wrong (against my face, not my neck). I have an explanation for that, but I only learned it after like 8 days of watching videos about makeup, and I don't know other ways of checking if I'm right or not besides like asking ppl. I have lots of ways of checking 1,2,3 and the spelling of "with".

> The same goes for all your other philosophical work. Keep it simpler. Practice things. Aim for mastery.

I think I lack some of the skills needed to self-eval. If I think about trying to self-eval on e.g. Goldratt stuff, IDK how I can do that besides writing about it to find contradictions or holes in my understanding. Discussion helps, too, but that's not self-eval anymore. Convergence and non-contradiction with other knowledge is good, too, but I think that's more like a hint than self-eval; it doesn't help to decisively find that I'm doing something wrong. Finding contradictions does tho.

I think there might be some like good general methods that could help with self-eval of philosophical work, but I'm not sure of them. Like I suspect they're out there but don't know what they are yet.

Mb writing bridging explanations (roughly: doing integration) is a good way to self-eval some. Feels like that's only a partial strat tho.

> Aim for a low error rate where correct criticisms are uncommon, surprising and treasured.

This struck me as really important.

I think my mindset around criticism still has some serious issues, like I value crits a bit but I can't tell quickly and reliably if crits are correct, surprising, and valuable. Sometimes I can, but those are exceptions, not the norm. I think mb I am still chewing on the idea that crits are valuable/desirable. I know they are in a bunch of explicit ways, but the idea isn't fully integrated w/ my mindset yet.

I think I wasted some opportunities when replying to 'a girl anon' that I could have made better use of, but IDK exactly how.

An issue for me atm is that I can tell when crits are uncommon and surprising, but I don't treasure them. yet (i hope).

-----

Notes: I feel like mb some of my paragraphs that expand on my 'No' answers are like excising things or evasive but I'm not really sure. I wanted to write more than just 'no' and to think a bit about the ways that minor successes are used to justify major failures. If there's dishonesty in those paragraphs then I don't think I know how to expand honestly, yet. Mb they're okay as reflections. i also wanted to reconsider whether I should think of the makeup project as successful via writing things out, and I am changing my mind on that, but there's more to do.

Ctx: I thought off and on about this reply for approx 6 hrs. So I did slow down some. I didn't write down a goal for this reply, tho. It's much easier to think of goals for posts that aren't replies compared to posts that are replies. I guess I put some of my goals for the reply in 'notes' above, but IDK if I had all those in mind before starting, or if I added them later on / retroactively.


Max at 9:13 PM on February 12, 2021 | #19969 | reply | quote

kb layout + accidental post -- postmortem

I posted #19969 before I meant to. I don't think it's obvious since I was in the final stages of editing. On my KB layout at the time (I have just changed it now) I had tab next to enter on the right thumb pad. this meant that I could accidentally hit <tab> then <enter> when I meant to just hit <enter> or just <tab>. i've avoided this sort of proximity issue in the past (predicting the error and avoiding some key locations b/c of it), but it's something that I should have known to avoid.


Max at 9:22 PM on February 12, 2021 | #19970 | reply | quote

#19960

> Can you explain more about how “excess capacity and how it relates to constraints” helps us understand “that we can still 'know' things even if there are problems and ultimately all ideas will be refuted by ever-better ideas”? That connection sounds interesting but I don’t get it.

yes. i think this would be good practice for me so I want to write a bit about it.

first, tho, do you agree with the second bit you quoted? i.e.:

>> the core of falliblism: that we can still 'know' things even if there are problems and ultimately all ideas will be refuted by ever-better ideas.

(I included a bit extra of the quote for context)

If you don't agree with that then I guess that something I write that directly responds to your request would not be very useful to you.


Max at 9:27 PM on February 12, 2021 | #19971 | reply | quote

afternoon project - SMM2 level

this is a project plan that should be achievable in an afternoon/evening (today). my method is based on things i've learnt from curi and method at the end of *it's not luck*

**goal:** produce a fun SMM2 level that teaches and demonstrates the 3 steps of learning: do it right/once, do it consistently, and do it fast.

**ctx:** I own SMM2 and have a little experience with the level editor. I have played 10-20 hrs of SMM2 and am reasonably familiar with the game modes, their features, enemies, interactions, etc. I've done a bit of planning and pre-thought about this, including some exploratory level creation last weekend. I drew up a level concept last night that I think can work -- that was something I wasn't sure about for a while.

**obstacles:**

1- lack of knowledge about vertical levels and camera lock

2- haven't tried to create the level, mb there are inconsistencies with the features I need (e.g. vertical + rising lava in SMW or New SMB modes)

3- haven't uploaded a level before and don't know the process

4- might not have left enough time today

5- it might take me too long to clear the level if it's challenging

6- the level might be too easy so it doesn't effectively teach ppl anything

7- i might not be able to fit enough height in the level to make it interesting and challenging for the player

8- the level is boring or frustrating

9- ppl don't like it

10- i need a level concept+design

11- rising lava might not be fast enough to be challenging (used to demonstrate step 3 in learning process)

12- level isn't challenging

**intermediate goals/objectives:**

- learn about SMM2 level editing via the in-game lessons and supplementary stuff like YT vids; solves obstacles 1,2,3

- work quickly on things that matter (aesthetics last), pay attention to avoid inefficient creation patterns (e.g. use copy-paste); solves obstacle 4 and partially 5

- integrate quality of life features: good checkpoints, avoid lengthy resets, losing progress, softlocks, etc; helps solve obst 5,8

- 3 1-ups + THX, helps solve obst 9

- ensure there's a decent level of challenge, even for experienced players, helps solve 6,8,9

- use doors sparingly and only where necessary. helps solve 7; note: doors could be added at the end to help with quality of life stuff if I'm under the limit of 4 doors.

- **done** come up with a level design/concept to iterate on, solves obst 10.

- use tracks and saws, podoboos, or spinning fire things to create the necessary time pressure, if required, solves obst 11. i can also research more ways to create time pressure and look at what ppl have previously done (e.g. by searching for specific speedrun levels)

- adjust position of blocks to make jumping up harder if it's not hard enough, solves 12 (could also make the gaps in the ledges shorter to make jumping up a bit harder, or add spikes in strategic places to make the jumps a bit more challenging, put higher standards on the players method, and create tension)

I might not need to solve obst 5,7 and I only need partial success with 8,9 (I don't need to please everyone)

most of the intermediate goals are about level design stuff and already include some knowledge i have about level design. I could do exploratory learning if obst 8,9 is an issue at the end.

at this point I chose to re-read *it's not luck ch 26*. I noticed that there's a q Julie asks that I didn't write in my notes during this chapter: "Do you have enough intuition about the subject?" similar to Alex, my answer is "I think so." after reading the chapter I think my next steps in the project are to plan the order I'm going to meet intermediate goals (and the corresponding actions, but that's fairly trivial and can be done ad-hoc), and then to execute on the plan.

**sequence of actions:**

- do the relevant SMM2 tutorials, identify gaps in my knowledge or things I'm unsure of, and supplement with more SMM2 tutorials and/or watch relevant YT vids

- refine the level design with my new knowledge and decide things like possible game modes and the required maps (I think particular backgrounds/level-themes are required to do vertical levels the way I want to). ideally I should have several options for game mode and level theme combinations that can let me achieve the goal.

- create a prototype level without quality of life features to verify the design can work

- iterate on design if necessary, do additional research if necessary (in general I can do this after any step)

- refine the prototype level and include the quality of life features

- test the level start-to-finish and evaluate it against the goal. iterate as described if need be.

- refine and polish the level, test again, add aesthetic stuff if I have time.

- publish the level.

there are some trivial steps I haven't included like 'name the level', 'write the level description', 'turn on my switch', etc. don't think it's necessary to document those, tho. I can do all that with excess capacity and they can be done at any point (except like turning on the switch, etc).

**level design prelim**

the main tech the level teaches is to reliably climb up a cliff via 1-block wide ledges

note: the 2nd image is of the sub-world and the note on the bottom right that's a bit cut off says "enuf time for player to use pipe to reset".

**meta-goal of the project**

my meta-goal for this project is, having now planned it, to not need to alter the plan but still complete the project. if I fail this goal, then my goal will fall back to altering the plan, documenting what went wrong/what I missed, and continuing with the plan. (I might have to alter it more than once.) another way to put this goal is that this project and plan has a low error rate.

doing this plan took approx 1hr


Max at 10:27 PM on February 12, 2021 | #19972 | reply | quote

SMM2 afternoon project success

I went through my plan and successfully uploaded the level 3hr 15min after I finished planning, so 4 hrs and 15 min all up.

level code **GL0-1HH-MVF**

I iterated on the level mostly while uploading, which took 45 min. There were some difficult jumps that I made easier; particularly at the end, and I reduced the difficulty of the 'consistent' phase to be more forgiving to the player since the hazards aren't meant to actually kill the player at that point, tho they can still die if not careful (that's deliberate).

the rest of the level dev was done fairly incrementally; I didn't change much after I put down an initial design and did some short play-testing (to make sure jumps worked and there wasn't any cheese).

I included 2 challenge areas, the first one of which is secret.

The second challenge area is right at the end and can kill the player if they're not careful. if they're really good then they will have been able to keep the mushroom this far, so it's easier.

the entire course, including secrets/challenges, can be done without damage.

I didn't alter the plan, but I think I had some oversights: I didn't plan for experiment/prototyping to overlap with the game-mode/level-theme choice. as it turns out there's a lot of flexibility here so the initial stuff I chose worked fine, but I should have planned some for some experimenting there. realistically it'd probs be okay to include it in the 'prototype' step I planned, but I feel like I should have included it. i count this as a minor error that I should be conscious of when doing future plans.

another minor error I made was spelling 'consistent' with an 'a': 'consistant' in the level. that error was also made in the diagrams I uploaded before.

I am surprised by how smoothly and quickly everything went compared with my past level-creation explorations.

I'm reminded of a line from *It's Not Luck* (ch29):

> We didn’t have time for mistakes, so we had to spend extra time planning!


Max at 2:06 AM on February 13, 2021 | #19974 | reply | quote

I recorded a video of the SMM2 level for anyone who is curious but doesn't have SMM2: https://s3.wasabisys.com/xert/2021-02-13_22-44-18.mp4

I recorded it with a capture card but didn't use HDMI throughput so there was a bit of input latency; it's not as difficult/inconsistent as it looks in the vid.


Max at 4:07 AM on February 13, 2021 | #19975 | reply | quote

#19971

>> Can you explain more about how “excess capacity and how it relates to constraints” helps us understand “that we can still 'know' things even if there are problems and ultimately all ideas will be refuted by ever-better ideas”? That connection sounds interesting but I don’t get it.

>

> yes. i think this would be good practice for me so I want to write a bit about it.

>

> first, tho, do you agree with the second bit you quoted? i.e.:

>

>>> the core of falliblism: that we can still 'know' things even if there are problems and ultimately all ideas will be refuted by ever-better ideas.

>

> (I included a bit extra of the quote for context)

The statement seems reasonable to me. I can't think of anything better. But saying I agree with the statement seems like it would require more knowledge of fallibilism and epistemology than I have.


Anne B at 6:46 AM on February 13, 2021 | #19978 | reply | quote

> #19927

> ** --- (the rest of this post is some free-writing / reflection / notes) --- **

> This is mb also related to an attitude I have towards learning, where I like think it's better or more valuable to explore things on my own than bootstrap via other ppl's knowledge. Something like that. This is a bit funny, tho, b/c I am still bootstrapping via other ppl's knowledge (tutorial vids), just not as ~directly as Q&A that I could do via like an hour of consulting time.

> There's some merit to that learning attitude generally. like DD made some significant improvements WRT structural knowledge & Popper's ideas. he thought about them in a diff way that had more reach (or something like that); they were more useful b/c of the way that he understood them, and that understanding was possible b/c he didn't do too much like direct bootstrapping from contemporary Popperians.

> Mb that learning attitude makes some sense to do for a bit, or if you're a really good thinker and working on hard/complex philosophy, but I think I've just convinced myself that it's not worth doing most of the time (e.g. learning makeup).

I think you have some really big mistakes in your thinking here. They are the kinds of mistakes that could be affecting all the other learning you are doing too, so it's worth actually thinking about and figuring it out. It's not specific to the context of makeup (which is where it came up).

I know you identify that you may have a mistake in your thinking. But I think the mistakes are worse/bigger than you realize.

I just wanted to note this because I have seen other mistakes in your writing & learning processes that I think may be caused by this same line of thinking.


trying to be helpful at 12:57 PM on February 13, 2021 | #19985 | reply | quote

re: 19985 by 'trying to be helpful'

#19985 I think what you're saying is that the part of my learning attitude "where I like think it's better or more valuable to explore things on my own than bootstrap via other ppl's knowledge" has problems. (There could be other problems related to the other stuff I said too, but the above is like the essence of it?)

I originally had trouble understanding your msg because I *thought* I rememberd/knew what I wrote, but after re-reading it, I didn't. I thought I had said something more like 'this is a part of my thinking that has a mistake' but really I just said 'this is a part of my thinking'.

Part of the reason I thought I'd said something more self-criticising is that I've talked a bit with curi about it in the tutorials and I've made some posts about it here, e.g. #18021 (I ctrl+f'd for 'exploratory' to find that, but I didn't find other examples, so I may not have posted about it as much as I thought.)

I think your msg is saying that, even if I have posted about this problem and mb recognise it a bit, that it's a bigger problem than I realise and has bigger impacts than I realise, even still. Also that if I want to get better at learning and thinking then I have to address this mistake as somewhat of a priority because it's a major error.

Does this line up with what you meant?

(my goal with this post was to ask the right q's that would give answers that let me figure out if I understood what you said or provide some way fwd if I still didn't understand)


Max at 4:41 PM on February 13, 2021 | #19986 | reply | quote

> #19985

> I think your msg is saying that, even if I have posted about this problem and mb recognise it a bit, that it's a bigger problem than I realise and has bigger impacts than I realise, even still. Also that if I want to get better at learning and thinking then I have to address this mistake as somewhat of a priority because it's a major error.

> Does this line up with what you meant?

Yes, that is what I meant. I know you recognize it as a problem. But I don't think that your view goes far enough. You still think there is merit in doing things from scratch, not building off of other people's knowledge.

I think you noticed a mistake in your thinking, but your analysis seems kind of like you think you are taking a thing that is good in a general way, and applying it in an area where it isn't good.

You aren't taking the mistake seriously enough - you are downplaying it, when you should be emphasizing it.

I think this mistake has come up in other places. You said something in discord the other day:

> I ~never look at solns for coding or maths stuff.

I think that view is mistaken, and is related to the mistake I am talking about in this post. You value starting from scratch, figuring things out on your own, not building off of existing knowledge. I don't think that is the best way to learn.


trying to be helpful at 8:11 PM on February 13, 2021 | #19987 | reply | quote

I have been intensely focused on work the past 2+ weeks (and have had some major successes). That will continue for at least another 2 weeks, and probably much longer. That means I won't post here nearly as much. FYI.


Max at 11:33 PM on March 2, 2021 | #20086 | reply | quote

Mb some progress on my learning conflict stuff

#19987

>> I ~never look at solns for coding or maths stuff.

> I think that view is mistaken, and is related to the mistake I am talking about in this post. You value starting from scratch, figuring things out on your own, not building off of existing knowledge. I don't think that is the best way to learn.

I think I just realised something.

I've changed my attitude a bit since mid-feb (in the right direction but not really going past any major breakpoints).

I was thinking about project planning, and I noticed that I don't do stuff like explicitly make trees for everything (I do for some things and particularly for hard things).

Mostly I do that in my head, and incompletely -- if I wrote the tree down and did proper brainstorming I'd have a better tree. For lots of stuff it doesn't matter *at the time*. Like I have enough *excess capacity* in my automatic-tree-making that it usually doesn't cause a project to fail. Partly that's b/c I can update it easily and usually there's nothing high stakes enough that missing it is really bad. However, have I *mastered* tree-making? Well, I just said that I haven't: "if I wrote the tree down and did proper brainstorming I'd have a better tree".

Why does this attitude seem okay at the time, but actually isn't? B/c, in effect, I am and have been compromising my self-judgement (when I move on too quickly and also don't keep improving those skills). I end up thinking I have more excess capacity (WRT e.g. planning skills) than I do, but how will I know that before I start a project? Certainly I do have *some* idea of whether I can complete a project, but I still make mistakes, and I'd have a much better estimate if I mastered relevant skills.

It's not that I'm immediately overreaching or doing it consistently, but I *am* consistently *at risk* of overreaching (to diff degrees in diff contexts).

I guess this isn't actually a direct reply to *trying to be helpful*, but I think it's related. I've thought a lot about sources of organization of knowledge recently, and there have been some ~minor changes in my choices b/c of it. One of the reasons this idea is related is that I think it implies that, b/c master is so important, the benefit of using well organized info is increased.


Max at 4:10 AM on March 24, 2021 | #20250 | reply | quote

#20250

Rereading, the way I put the first two sentences is misleading. I said:

> I think I just realised something.

> I've changed my attitude a bit since mid-feb (in the right direction but not really going past any major breakpoints).

That makes it sound like the thing I just realized was something I did in February.

That's not the case; the second sentences was meant to be like bg context but I didn't fix the placement in editing. I meant to include it, but to find a good way to integrate it as bg context.


Max at 5:59 AM on March 24, 2021 | #20251 | reply | quote

It's common that people can do stuff with their autopilots/intuitions/defaults without writing much down. Some people are even really good at some stuff (their expertise) that way. But it's limiting. It puts a cap on how advanced your knowledge can get. (It's not an exact cap. It's more like it creates diminishing returns on your learning, so going further takes more and more work.) Explicit analysis and error correction of the pretty-good stuff removes the cap on progress. It makes it easier to realize better can exist (people generally don't see what they're missing out on) and access it.


Anonymous at 10:46 AM on March 24, 2021 | #20252 | reply | quote

learning strats

this is part of my ongoing work on my major idea-conflict about learning.

What's the return on learning?

Let's say that you devote 1/2 of your working life on self-improvement. Or that your employer lets you devote 1/2 your working hours to self-improvement in a business-focused direction.

How long does it take for you to hit Return On Investment (ROI)? If you're spending 1/2 your hours on learning, then the point you reach ROI is when you get more than twice as good at your job (or your hobby, or whatever).

Say that: if you are good at learning, you get 1% better at doing your job per day. I think that's actually achievable for context-specific things (i.e. right person + right ideas + right subject matter). Well, your productivity "principle" is 1, and the "compounding interest" that you get from learning is 0.01 and the period is `d`. So `(1 + 0.01)^d > 2` is our ROI point. That means `d > ln(2)/ln(1.01)` which means, approx `d > 69.7`. *70 days* is the ROI point.

Is that practical in general? probs not. What about an improvement of 0.1% per day? That leads to `d > 693.5`, so like ~2 years. I think the 'real', general, maximum rate of improvement is between 1% and 0.1% for most ppl WRT most things, i.e., most ppl can get twice as better at most things in less than 2 years. There's definitely a period of high getting-better growth for completely new subject matters, tho.

Is there going to be a major factor in someone's growth rate? Yes. It will be, provided they're willing and motivated to learn, *the organizational quality of the learning material*. Organization of materials covers *which things in what order with what other context?* That *must* be the major factor because, provided someone is willing and motivated, the order and quality of ideas that they're introduced to is directly related to their learning progress. How could it be otherwise? If order and quality were not decisive factors, then the order of learning materials wouldn't be that significant (contradicted by structural epistemology) and/or the quality of materials wouldn't be that significant (contradicted by error-rate and overreaching). Note that *organizational quality* necessarily includes both order and content.

Mb a good time to mention: there is some excess capacity in the order of learning material; that's good because it allows for chewing and higher-quality self-judgements about one's learning. There's also some excess capacity in the quality of learning material; it only needs to be good enough to meet major breakpoints in the quality of the student's understanding. The student can't learn ideas perfectly, but provided they avoid major structural issues w/ what they do learn, then the student will have enough excess capacity in what they have learned to do useful work. They can also improve their ideas later without the overhead of bad structure. Humility helps here -- if they think their knowledge has too much reach then that can inhibit future learning.

So, one's ROI on learning is heavily dependent on the organizational quality of learning material. Material that meets major breakpoints (i.e., doesn't introduce major structural issues) is worth seeking. Material that is well written and easy to understand is worth seeking.

**Seek organized learning material. Follow through.**

One of the benefits of doing philosophy is that the skills you develop help you to mix and refine multiple sources. If there is a simplistic but accessible source of info, and a high quality but badly-written source of info, then being able to consume both quickly and efficiently with minimal errors is *profitable*. Not every subject has easy-to-find material that's high quality in both regards, so this skill matters.


Max at 2:40 AM on March 31, 2021 | #20312 | reply | quote

learning strats - addendum/errata

An addendum:

> most ppl can get twice as better at most things in less than 2 years.

That's provided they have the philosophical knowledge about learning to do so. That's the major constraint for most ppl. But I don't think those ideas are necessarily that hard to learn. I should have made this bit more clear, tho.

One reason I didn't say it as explicitly in the prev post is that I think this stuff is included in "the organizational quality of the learning material", like good learning material will also have good knowledge about learning that it teaches you beforehand or along the way.

That's a bit contradictory to what I say at the end tho: "One of the benefits of doing philosophy ...". The contradiction is b/c I say this helpful skill is helpful as an afterthought, but don't mention it when it's more important -- earlier in the text. It's a contradiction because that sort of bad oraganization is probs not good to have in a mini-essay about the importance of quality organization.


Max at 2:50 AM on March 31, 2021 | #20313 | reply | quote

#20252 Btw anon, I appreciate you commenting and agree with you. I think you put this idea well and succinctly:

> Explicit analysis and error correction of the pretty-good stuff removes the cap on progress.

I guess that ppl think that, b/c they're good at something, whatever specific limits they have are like ~universal among all other comparable ppl. Like they're at a "hard cap" rather than a soft one. A problem that comes out of this is that a ubiquitous soft cap is difficult to tell apart from a hard cap -- if ppl don't have the philosophy skill to tell the diff then they're self-limiting.


Max at 2:55 AM on March 31, 2021 | #20314 | reply | quote

#20334 Thx.

For anyone reading this after the fact: I replied to curi's post, and if there's more discussion specific to that topic it'll probably be under that thread.


Max at 8:06 PM on April 4, 2021 | #20336 | reply | quote

#18750 Max, you should not be trying to write a philosophy book or a long series of articles. That's large overreaching. You should practice the things from our tutoring.


curi at 2:42 PM on June 14, 2021 | #20585 | reply | quote

#18803

> This is a bit of a guess at a general method for doing FI.

You need to organize your life and form some good habits (or otherwise set up things you happily, easily do regularly). They should include practice, reading and writing (e.g. freewriting, notes, forum posts, outlines, summaries, attempts to clearly explain something).


curi at 2:47 PM on June 14, 2021 | #20586 | 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.)