1. 40
    1. 5

      My first attempt at recording myself coding is using ffmpeg -video_size 3840x2160 -framerate 2 -f x11grab -i :0.0 output.mp4 to record two frames per second of my native screen resolution. Any improvements or other ideas?

      1. 7

        I’ve found OBS Studio painless for desktop recordings, but it sounds like you’re set already.

      2. 3

        If most of your work happens in the terminal you might use script or asciinema. They result in smaller recording sizes.

      3. 2

        If it works for you, it’s probably not worth changing software.

    2. 4

      I think an almost-rephrasing of the longer post is, “if you have the resources and motivation to make a hobby of getting better at X, and a reasonable approach to doing so, you’re ahead of a lot of people.” Another way of putting that is that a lot of people either lack the resources and motivation to work on something persistently (like an ongoing hobby), or lack a good strategy for getting better–as he mentioned with public speaking, you don’t always have a vocabulary to nail down what being good is. And detached observation and getting the help of others are a couple of specific strategies he cites as helping.

      Other than that I’d like to flag his link about picking the right problem dominating execution speed. Not that execution speed doesn’t matter (you can’t think all day, plus you learn things from execution that can help you do the right next step), but if you’re doing reasonably well at putting one foot in front of the other, thinking more broadly to try to find the right thing to do can have more impact than getting a little faster at carrying your plans out.

    3. 3

      Anybody read the Inner Game of Tennis? #4 is straight out of, I don’t know, chapter 2. You gotta look at film/replays, &/ get a coach.

    4. 4

      This is like when somebody who is 6’8” says 6’2” isn’t tall. I mean, it’s all relative, man.

    5. 3

      The headline made me think of machine learning problems. In ML, 95% can be “fantastic” or “abysmal” depending the problem.

    6. 2

      Don’t see anything about programming there?

      1. 8

        It’s under a heading about three quarters of the way down. Long intro, but topical.

    7. 1

      It took a long, rambling way for me to get to the point where i asked: “wait, people spend the majority of their programming time in a code editor? I spend the majority of my programming time either offline or reading and absorbing requirements”

    8. -6

      I stopped reading this as I rejected the premise early on, the example I had in mind was chess, try get in top 5% on Chess.com or Lichess. “Isn’t that good”? Well, it’s about a 2150 or 2200 chess rating.

      1. 26

        I think bragging about not reading the article is not a good habit to encourage on lobsters, but you’ve also misread–and the clarification was in the second paragraph.

        The relevant comparison isn’t players on chess.com, it’s people who play chess, and that’s a larger group. For instance, I’ve played games against my daughter in the past year, and against a friend or two within the past few years, but I’m not active on chess.com.

        Similarly, I’m about 50th percentile in people who play Go tournaments (maybe a little lower even, I can’t remember), but I’m well above average (at least for players in the US. I don’t know what the distribution is like in China/Japan/Korea–they have tons of strong players, but also millions of players overall). I don’t know if I’m 95% percentile, but definitely not near 50th.

        1. 2

          I skimmed the article, and it’s a rambling mess. There’s some nuggets there but they’re really hard to sift.

          I liked the author’s coinage(?) of the word “ridiculable”.

          It would generally be considered absurd to operate a complex software system without metrics or tracing, but it’s normal to operate yourself without metrics or tracing, even though you’re much more complex and harder to understand than the software you work on.

          This is a good observation, slightly marred by the existence of a plethora of products designed to track employee’s every move on screen. I guess a programmer interested in improving their productivity can get an evaluation license of this kind of software.

          Also this link looks interesting:

          1. 6

            I wonder if it’s the game analogy that’s giving people trouble, because it immediately clicked for me. My game of choice is different – Magic: The Gathering, which first came out right as I was a teenager likely to be able to pick it up and enjoy it – but my experience with it absolutely lines up with the article.

            There’s a very large population of people who play Magic. And reaching 95th percentile within that population is something literally anyone could do by putting in the work. It seems like a high bar, but it really isn’t, because even things like reading a few introductory articles on competitive strategy and practicing what you learn from that will quickly advance you past the average kitchen-table Magic playgroup. Not that much more effort will put you up to the level of being able to win at a typical Friday-night tournament in a local game shop. And at that point you are undeniably going to be 95th percentile, if not higher!

            Even within the specifically competitive-focused subset of the Magic-playing population I think this holds up. Within the last couple years a new digital version of the game (called “Arena”) has come out and been promoted heavily, and it has competitive play with a ladder of ranks and tiers. It’s attracted a fair number of streamers who are new to the game, and again it seems that anyone willing to put in some effort and practice can start consistently reaching the higher ranks, which again put them into the 95th percentile or higher of Magic players, and even of that specific subset who play on Arena.

            Though I think some of the problem here is also perspective: people won’t compare themselves to the general population, or even to the subset who do things like go to tournaments or participate in ranked play on Arena, where it would be clear just how low the skill-level bar of 95th-percentile really is. Instead they compare themselves to the population of established elite professional players, and see a huge skill gulf between themselves and the pros and draw the wrong conclusion. Being only 1% as good as a top-level pro (assuming we could quantify that) does not mean being only in the 1st-percentile of all players, simply because the pros are such a microscopically tiny subset of a very large population, but people often think about it in those terms.

            1. 1

              I guess it’s only applicable to fields where casual and enjouable participation is possible.

              On the one hand we have fields like algebraic geometry that you cannot participate in without extensive preparation. For someone with high school math level, it will take years to even start understanding the papers. Even then you are are only ready to start doing any research of your own at all.

              On the other hand we have fields where, until some point of profiency, it doesn’t matter if you are are better than N% participants. There are many people (mostly kids) who play the violin. You can get better than most of them just by learning not to tune the strings to diminished fifths. It will take years of dedicated practice until anyone will genuinely want to listen to your playing though.

          2. 3

            The existance of personal tracking software doesn’t imply that it is not “normal to operate yourself without metrics or tracing”. A fairly small proportion of the population use such software.

            1. 2

              I’m thinking of software that allows an employer to track how much time their employees spend in different windows and applications, so they can take action against “incorrect” behavior.

        2. 0

          But 2200 is top 5% of people who have ever played chess online, including those 4 or 5 games and so on. As in top 5% of participation metric “has played chess online before”, I thought

      2. 13

        This is explicitly addressed in the beginning of the article:

        Note that when I say 95%-ile, I mean 95%-ile among people who participate, not all people (for many activities, just doing it at all makes you 99%-ile or above across all people). I’m also not referring to 95%-ile among people who practice regularly. The “one weird trick” is that, for a lot of activities, being something like 10%-ile among people who practice can make you something like 90%-ile or 99%-ile among people who participate.

        1. -1

          But this is for people who participate. 2200 is top 5% of people who have ever played more than, say, 5 or 10 games of chess online.

          1. 12

            This is triply incorrect and once misleading.

            First, many chess players never play online. I’d even guess that most don’t, so that is not the correct population to compare to.

            Second, chess.com’s displayed percentiles are not for every player who’s ever played, only for active players. There was a change was a number of years ago, before this chart was made.

            Third, if you look at that chart, top 5% among active players is roughly 1600 on chess.com, not 2150 or 2200.

            Fourth, when you say it’s an X chess rating without qualification, I think this would imply to people in the U.S. that this is a FIDE or USCF rating. 1600 on chess.com from when that table was made converts to 1500 USCF and, again, that’s an overestimate because that’s only active players on chess.com which is going to be overweighted towards players who have put more time in.

            Your stated number, 2200, is in the top 0.2% of active chess.com rapid players. 2200 must come from lichess blitz ratings. At the top of their blitz ratings graph, it notes that it’s for players active this week, so that also has the incorrectness mentioned above. Additionally, it’s well known that lichess generally has inflated ratings and blitz is particularly inflated even for lichess. It is extremely misleading to say that top 5% is “2200 chess rating” when referring to lichess blitz ratings.

            Even if you look at people with USCF ratings, which is a tiny subset of the people who have played or play chess in the U.S. (roughly 85k USCF players, out of probably over 100M people who have played in the U.S.; that chart is old and has 65k but the distribution shouldn’t be wildly different), top 5% is still only 2000 USCF. “2200 chess rating”, as you put it, is someone roughly in the top 1000 USCF. Across all U.S. players, even accounting for strong players who don’t maintain a USCF rating, that’s probably at least the top 0.001%.

      3. 4

        I got to around 1900 in lichess classical in about a year without any specific effort starting from scratch, but a lot of play time.


        My greatest victory is against a 2314 rated player.

        In school I was about 90th percentile, so in general I think for a lot of tasks, with practice you just slot into where your intelligence level is, with deviation around quality and amount of practice.

        I think looking at the graph of all time rating of players is really fun.