[Previous] Mixed Society | Home | [Next] Social "Intellectuals"

Time-Based Metric For Overreaching

How can you tell if you're overreaching? Here are simple guidelines:

90% of the time, thinking should take 2 minutes or less. (1 in 10 things goes past 2 minutes.)

90% of cases that take longer should be under 15 minutes. (1 in 100 things goes past 15 minutes.)

90% of the cases that take longer than that should be under 2 days. (1 in 1000 things goes past 2 days.)

Next steps should be fast. You shouldn't be stuck for long periods of time. ("Long" means longer than the amounts of time above. A main point of this post is that people have the times wrong and are routinely stuck for a few hours and don't realize how long and bad that is.)

Most stuff you do should be small and easy. If it's not, break it into smaller parts (so that you can be making progress frequently by finishing one little part) or find easier stuff to do.

If someone says something, you should usually have an idea of your reply within 2 minutes. A clarifying question is fine as a reply. It doesn't have to be a big thing. Or if you are going to give a big reply where you make 5 points, then you could think of each point as a mini project and figure each one out in 2 minutes.

If you're writing an article or novel, most steps should take less than 2 minutes of thinking before you do them. A paragraph is a reasonable step. You decide what the next idea will be, then you write the paragraph for it. If you stop midway through the paragraph, starting again is another step. If you need to do planning for the paragraph, e.g. checking your notes about the plot and your chapter outline, those activities are also steps. If you spend 10 minutes reading your notes before starting a paragraph, that's fine, that's time spent making progress on the activity. The time limits are to deal with the time you aren't doing anything, where you're just thinking and not actively, directly getting anything done. When the breaks between actively doing stuff are larger than these time limits, that indicates it's hard for you and a lot of problems are coming up and you're probably making a bunch of mistakes.

Don't try to cheat. This will only help people who approach it honestly. Like if you think of a clarifying question in 10 seconds, just ask it. Don't save it for 1 minute 50 seconds to try to get extra thinking time.

If you're usually going near the time limits, something is wrong. Sometimes it should be 5 seconds, sometimes 30 seconds, sometimes 90 seconds. If you're frequently just under 2 minutes (or a little over and rounding down), you're probably overreaching. For the 2 day timeframe, most of those should only take a couple hours of time you actually spend on it. Actually spending a large portion of one day, let alone two days, should be much rarer. Two days gives you time to sleep on it, or leave it on the back burner for a while, wihch is good to do occassionally.

These guidelines are not exact but the simplicity and ease-of-measurement are major upsides. They can give you a ballpark of what to look for. Compare what you do to this and see if it's even close. I think people don't have much understanding of how long "too long" is, in concrete numbers, so this will help.


Elliot Temple on February 12, 2019

Messages (12)

Fortnite overreaches by adding new features (with new bugs) while they have a bunch of bugs. Ninja says this could kill the game.

https://www.youtube.com/watch?v=1v67uvxJ4LY


Anonymous at 4:25 PM on April 10, 2019 | #12121 | reply | quote

Some comments on overreaching from the Fallible Ideas Discord chatroom today:

[7:48 PM] curi: If you have experience both with projects where you get overwhelmed (that's a typical way that overreaching feels) and projects where you have things under control and succeed, then you can make judgments about future projects: is it planned out enough to know what to expect? Do you know the relevant skills/fields enough? Do you have the resources needed? etc. Most people are really used to being rushed all the time, and glossing over lots of mistakes all the time – they are in a chronic state of overreaching which they interpret as normal. So they know a new project will be like that, but they think that's normal, so they do it anyway.

[7:50 PM] curi: Another thing to try is asking yourself "Am I confident I'm doing this right? Would it be surprising to me if I made a mistake?" or "Am I confident that this paragraph I wrote is correct? Would a mistake surprise me?" People often know they wouldn't be very surprised by a mistake being pointed out, and/or know they aren't very confident. Those things indicate overreaching (if your goal is success. it's different if your goal is just to practice or try something out in an exploratory way).

[7:52 PM] curi: High errors rates are often caused by lack of resources, e.g. trying to do something super cheaply or super fast. (Time and money are typical important resources.) A good rule of thumb for a project is to have 50% extra for resources. That means if you think it should take $100 and 10 hours, you budget $150 and 15 hours. That allows for random variations (sometimes things go better or worse than expected) without it causing big problems.

[7:54 PM] TheRatWay: Wow that's quite a helpful way to think about it

[7:54 PM] curi: A lot of people try to add margin for error to individual steps within a project (e.g. they should be able to do one task in an hour, so they give it 90min). But it's better to add margins to the whole project.

[7:56 PM] curi: Because statistical fluctuations are bigger on a small scale. A task could take 3x longer than expected. But for a whole project, your luck evens out more (some things take close to the expected value, and the things that take more or less help cancel each other out)


curi at 7:58 PM on July 18, 2019 | #13126 | reply | quote

In the work I’m doing on *Structure and Interpretation of Computer Programs* (SICP), I’m not meeting these time guidelines.

> 90% of the time, thinking should take 2 minutes or less.

A lot of my SICP thinking takes under two minutes, but a rough guess is it’s more like 50% than 90%.

But

> A main point of this post is that people have the times wrong and are routinely stuck for a few hours and don't realize how long and bad that is.

I am not routinely stuck for a few hours on SICP, although there have been some exercises I’ve been stuck on for a few hours (maybe 2-4 out of the 73 I’ve done so far). I estimate that around 80% of the exercises take me under an hour, including editing and formatting and posting time, but very few take under 20 minutes.

I think I am learning SICP stuff. I think I am correcting some of my errors and not building on them or getting overwhelmed by them. But there are some exercises I haven’t been able to complete to my satisfaction. And I don’t think I’ve mastered the course material, just learned a lot from it.

Am I overreaching? Should I do some simpler coding courses first before continuing with SICP? I’m comfortable continuing with SICP but I also wonder if I’d be better off doing easier stuff such as more of Simply Scheme.


Anne B at 12:40 PM on October 29, 2020 | #18528 | reply | quote

> In the work I’m doing on *Structure and Interpretation of Computer Programs* (SICP), I’m not meeting these time guidelines.

>> 90% of the time, thinking should take 2 minutes or less.

> A lot of my SICP thinking takes under two minutes, but a rough guess is it’s more like 50% than 90%.

> But

>> A main point of this post is that people have the times wrong and are routinely stuck for a few hours and don't realize how long and bad that is.

> I am not routinely stuck for a few hours on SICP, although there have been some exercises I’ve been stuck on for a few hours (maybe 2-4 out of the 73 I’ve done so far). I estimate that around 80% of the exercises take me under an hour, including editing and formatting and posting time, but very few take under 20 minutes.

> I think I am learning SICP stuff. I think I am correcting some of my errors and not building on them or getting overwhelmed by them. But there are some exercises I haven’t been able to complete to my satisfaction. And I don’t think I’ve mastered the course material, just learned a lot from it.

> Am I overreaching? Should I do some simpler coding courses first before continuing with SICP? I’m comfortable continuing with SICP but I also wonder if I’d be better off doing easier stuff such as more of Simply Scheme.

Do you have a criticism of trying more e.g. Simply Scheme stuff and comparing your speed and error rate on that to your speed and error rate on SICP stuff?

Also have you identified any patterns in the stuff you do get stuck on? I realize you said it's only 2-4 things but were you able to see some common theme? I ask cuz if you did maybe you could work directly on something relevant to that.


Justin Mallone at 8:47 AM on October 30, 2020 | #18537 | reply | quote

Big picture: I’m uncertain about all this. Below are some rambling thoughts that that aren’t well organized and don’t come to conclusions.

> Do you have a criticism of trying more e.g. Simply Scheme stuff and comparing your speed and error rate on that to your speed and error rate on SICP stuff?

Trying to calculate speed and error rate would be complicated. Time taken to complete an exercise might not be a good measure of thinking speed on it. Some exercises require more thinking steps than others. And what is error rate? Error per exercise doesn’t seem useful. And what counts as an error? A try at a program that doesn’t work but gives me feedback that I can use to fix it? Posting something that is wrong and not knowing at the time that it’s wrong?

> Also have you identified any patterns in the stuff you do get stuck on? I realize you said it's only 2-4 things but were you able to see some common theme? I ask cuz if you did maybe you could work directly on something relevant to that.

Hmm. Two instances of where I got stuck were Exercise 2.3 and the stuff at the beginning of Section 2.2 about closure. Those aren’t about the same topic. What they have in common is that I couldn’t figure out what the authors meant by some stuff they wrote. But that’s a pretty general property that would apply to lots of things someone could get stuck on.

——

General thoughts

If I want to learn coding and I want to follow the Time-Based Metric for Overreaching, I should look for something easier than Simply Scheme.

What would it look like for me to only do FI things that fit the Time-Based Metric for Overreaching? It would be very different from what I’m doing now. I wouldn’t be doing SICP. I wouldn’t be doing Simply Scheme either. I wouldn’t be doing the TCS stuff I’ve started doing. I’d skip a lot of the responses I give to other people in various FI places. I wouldn’t have a learning plan or weekly evaluations. I might be able to do things like watch an FI video or read an FI article and write down a few ideas I saw in it. I might be able to read a book and maybe come up with a question or two. But I couldn’t follow most of my questions to resolution.

Often I have no coherent thought within two minutes of seeing something, but I do have coherent thoughts within ten minutes or an hour or something, especially if I’m writing stuff during that time.

The grammar work I did last year, which I’m pretty sure resulted in me learning a bunch of stuff, also did not come close to meeting the Time-Based Metric for Overreaching. Maybe my problem is slow thinking or lack of focus or something rather than that I’m doing things that are too difficult. Or maybe I really did learn stuff effectively but it was still overreaching and I’d have learned more in less time if I had picked something easier than grammar.

It took me way more than two minutes to think of what to say for each part of this comment. Does that mean I shouldn’t have answered? Even trying to find out if I’m trying to do things that are too difficult is too difficult for me? That doesn’t make sense.

Maybe I could learn to think in smaller steps. I don’t have ideas on how to do that, though. Actually, I do have an idea, but it seems silly: if I haven’t thought of something after two minutes, I could think “I haven’t thought of anything. But I think I will if I try for another two minutes, so I’ll do that.”

I’m not convinced that the Time-Based Metric for Overreaching is a good idea. But I could investigate it some more. I could try drastically reducing my FI difficulty level so that I meet the metric and see how it goes. It seems like it would be really hard to time how long it takes me to think of each idea I have while I’m doing something, though.


Anne B at 2:31 PM on October 30, 2020 | #18541 | reply | quote

> Two instances of where I got stuck were Exercise 2.3 and

You didn't know you had a problem on 2.3 until I said something. You presumably have problems on other exercises without realizing it.

> It took me way more than two minutes to think of what to say for each part of this comment. Does that mean I shouldn’t have answered?

When you have a conversation in person, you are able to reply to the vast majority of things people say in under 2 minutes. Usually you can start replying in under 2 seconds. You do this routinely for many topics.

You could have a conversation with an IRL friend about this stuff (who is familiar with this stuff enough to discuss; don't imagine trying to explain it to someone who's never heard of it) and think of things to say in under 2 minutes.

When you allegedly can't think of stuff in under 2 minutes that usually means you're preventing yourself by adding extra barriers, requirements, rules, goals, etc.


curi at 6:32 PM on October 30, 2020 | #18545 | reply | quote

My first thought in response to this, in under two seconds, was: “Well, I can think of random things to say in two seconds that correspond to the topic, but they’re likely to be wrong.” Then I gave it some more thought as I tried to write that down. Then I looked back at the original post on this page. My next thought was “ ‘90% of the time, thinking should take 2 minutes or less’ is about how to tell if you’re overreaching on a topic. If I can always think of something random to say right away but it’s likely to be wrong, I don’t think that counts as thinking taking 2 minutes or less.” So yes, I add the extra goal of saying something that I have some confidence is right. I check my first thoughts against other things I know and against the thing I’m responding to. When I’m writing, I add extra goals of wording/grammar/punctuation/spelling.

Example: I read through SICP Exercise 2.29 for the first time yesterday. The exercise has four parts and I read through all four to get an idea of the whole thing. For each part, I thought of something to try first within a few seconds. But it’s likely that none of these will be my final answers when I do the exercise. I started to write the code for part a and realized I needed more information so I thought through what I needed and got that information. Once I write some code I’ll test it and see if it does what I think it’ll do. It’s likely that it won’t and I’ll do some more thinking about why it didn’t do what I thought it would and how to change it. So yes, I think of something right away and that’s under two minutes. But the whole process of doing an exercise often includes some thinking steps that take more than two minutes.

So how do I apply the metric? “thinking should take 2 minutes or less” applies to what kind of thinking? I could be studying something that’s definitely too difficult and still think of *something* about it right away. Should I just say that first something if I can’t think any further? Then I’d be studying that too-difficult topic in a non-overreaching way? That doesn’t seem like it would be very productive.

When I wrote this post, I alternated thinking of ideas, writing them down, and re-reading to see if what I wrote made sense with the other stuff I wrote and with the other posts on the page. (And I wondered if the word “alternating” can apply to more than two things and I looked that up. I think it’d be better to use “switched between” or something rather than “alternated”.)

Now I’ve written four paragraphs and been back over them a few times. I’m not confident that what I’m saying makes sense. To stay in non-overreaching territory, should I add a sentence that I’m not confident in what I’ve said and just post it? Or should I take longer than 2 minutes or longer than 15 minutes to really think through what I want my main point to be and whether it makes sense and organize my writing so it’s more clear what my main point is and maybe change my mind about what I want to say? This time I’ll just post it, but usually I’d do more thinking when I have the feeling that I’m not clear in my own mind about what my main point is and whether it’s correct. I think that’s the part where I learn stuff and that’s also the part that takes longer.


Anne B at 11:29 PM on October 30, 2020 | #18549 | reply | quote

Oh, and I’d also go back through the things I’m responding to and probably catch places where I didn’t take into account something that was already said by me or someone else.

Where I might get stuck is, after all the going back over things, *then* realizing that I don’t know what I’m talking about. And then I’d feel like I had wasted my time. That seems like an overreaching process. I could have made it quicker by just writing my first thoughts and saying they were just first thoughts and not spending all the other time. But I can’t always predict when I’ll figure out something new and worth thinking by doing a bunch of extra thinking and when I won’t.

Again, I’m posting this without a lot of extra thought. But it doesn’t seem fair to readers to write a bunch of thoughts without making an effort to make them good thoughts.

My gut feeling is that I'm overreaching with these posts.


Anne B at 11:38 PM on October 30, 2020 | #18550 | reply | quote

> > Two instances of where I got stuck were Exercise 2.3 and

> You didn't know you had a problem on 2.3 until I said something. You presumably have problems on other exercises without realizing it.

My first post on my blog about SICP Exercise 2.3 included this:

https://aelanwave.wordpress.com/2020/08/28/sicp-exercise-2-3/

> Hmm. Other people’s answers took points and sometimes other things as the arguments for “make-rectangle”, rather than segments like mine did. I think my answer is okay, although it’s harder to give my “make-rectangle” its arguments. I did not put anything in my “make-rectangle” procedure to check that its given arguments meet the requirements, but I could have.

However, my first post to FI about SICP Exercise 2.3, on the same day, before anyone else said anything, included this:

https://groups.google.com/d/msg/fallible-ideas/ErcZk3tL7BQ/5R3fhVqhAAAJ

> 1) I’m not sure I understood correctly what they wanted in this exercise. Did I do two different representations of a rectangle? And I’m not sure I understand well enough what they mean by “abstraction barriers”.

So I claimed both that I thought my answer was okay and that I wasn’t sure I understood what they wanted in the exercise. I posted about the exercise because I had some suspicion that I had a problem with it.

I hope that I learned from that exercise that I should pay attention when I have a suspicion that I might not be getting something, to follow through on the suspicion, and not to claim that I think my answer is okay. My guess is I haven’t learned that well enough yet.

I agree with your point that I probably have problems on other exercises without realizing it. I don’t think I’m good enough at knowing when I do or don’t understand something.

- - -

The below questions are not addressed to anyone in particular.

So how does not understanding something and not realizing you don’t understand something relate to the time-based metric?

I can imagine someone writing or saying a bunch of wrong things pretty quickly and not realizing they are wrong. Is this overreaching? My guess is yes.

I can also imagine someone taking a while to come to some conclusion, including some time along the way where they were stuck for a while, and yet their final conclusion and their thinking along the way might be valid. Is this overreaching? Is the fact that they were stuck for a while in the middle a sign that they’d have been better off tackling easier topics?


Anne B at 7:41 AM on October 31, 2020 | #18551 | reply | quote

> The below questions are not addressed to anyone in particular.

> So how does not understanding something and not realizing you don’t understand something relate to the time-based metric?

> I can imagine someone writing or saying a bunch of wrong things pretty quickly and not realizing they are wrong. Is this overreaching? My guess is yes.

Elliot's comments here seem relevant: https://www.youtube.com/watch?v=urGtEzrVJuM#t=1h8m01s


Justin Mallone at 8:22 AM on October 31, 2020 | #18552 | reply | quote

> I hope that I learned from that exercise that I should pay attention when I have a suspicion that I might not be getting something, to follow through on the suspicion, and not to claim that I think my answer is okay. My guess is I haven’t learned that well enough yet.

FWIW I've often found, upon reflection, that I make various mistakes *when I had a suspicion something was going wrong* but disregarded that suspicion for whatever reason. I think it's an important issue and I don't have a great solution. But it's important cuz the case where you have a suspicion something is going wrong is the case where you can actually do something about it. If you're legitimately totally blind to a problem, that's a tough situation, but if you actually have even a tiny bit of awareness of a problem, that's something really critical to pay attention to. When you have no awareness of a problem and then it sneaks up on you, it's kind of like blindly walking into a wall, so anything that lets you avoid doing *that* is precious.


Justin Mallone at 6:37 PM on October 31, 2020 | #18553 | reply | quote

SmallAnt demonstrating not-overreaching

https://youtu.be/GBETTDE2Nyc?t=178 - SmallAnt - I Tried One Of The Hardest Speedruns Without ANY Preparation... (SM64 1 Star)

Re: learning, speedrunning

Here's an example of a speedrunner (SmallAnt) breaking down a technique (BLJ) into smaller parts as a learning technique. This happens 4min 36s after console power-on (he didn't practice beforehand). I think figuring out something like that in less than 5 minutes and taking corrective action is really good. He's not overreaching.


Max at 6:08 AM on November 15, 2020 | #18669 | reply | quote

(This is an unmoderated discussion forum. Discussion info.)