TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

95%-ile isn't that good (2020)

166 点作者 DantesKite超过 3 年前

27 条评论

xigency超过 3 年前
Admittedly, I only skimmed this. But within the space of video games, I’m not sure Overwatch is a stunning example since it’s generally a casual game.<p>I’ve been playing various iterations of Counter-Strike and StarCraft for decades. Sometimes with deliberate practice, including reading strategies, playing custom maps to practice certain skills, and watching series aimed at improving level of play. And despite the hundreds of hours invested I don’t believe I’ve ever cracked the top 50% in any leaderboard rankings.<p>Anyone in the top 5% is definitely <i>that good</i>, executing complex strategies with extreme precision.<p>There are lots of areas in life that are highly competitive — where being the best takes raw talent, a huge time investment, and a bit of luck. I wouldn’t try to diminish that.
评论 #30412723 未加载
评论 #30411869 未加载
评论 #30412482 未加载
评论 #30412368 未加载
评论 #30411919 未加载
评论 #30413924 未加载
评论 #30417208 未加载
评论 #30412772 未加载
评论 #30415235 未加载
评论 #30414269 未加载
评论 #30411765 未加载
评论 #30415279 未加载
评论 #30417371 未加载
elil17超过 3 年前
A video game and public speaking seem like really weird&#x2F;unusual examples.<p>A ton of people play video games, very few people do it professionally. A ton of people give talks, but it’s usually a small part of their jobs. Being even 50%-ile among <i>professionals</i> for either would certainly be extremely impressive.<p>When I think about normal jobs, which are less competitive fields than professional gaming or speaking, being 95%-ile is very good (although 50%-ile is not great). Certainly, the top 5% of people at my company (200 people in my office, so 10 people I work with) are on the “top of their game”, so to speak - they always drive towards the most important objectives, rarely make errors, come up with incredible ideas, etc.
评论 #30411634 未加载
评论 #30411541 未加载
评论 #30411701 未加载
评论 #30411781 未加载
lordnacho超过 3 年前
He&#x27;s basically right. Here&#x27;s another example, math contests.<p>When I was in high school I was really into math. I was way ahead of my year group and I figured it was more interesting to do special problems of the &quot;what is the last digit of 2^2022&quot; variety, the kind of thing that didn&#x27;t turn up on normal homework, but also not terribly complicated, the kind of thing you can figure out with a short burst of concentration.<p>So I went to math contests around Europe with other schools. The thing was that most kids I met there were totally ordinary at maths. Sure, some were really quite good, but most of the bottom 75% were indistinguishable from your normal kid who doesn&#x27;t know at 15 what an integral is. The friends on my team weren&#x27;t even in the top set math class at my little school, but my observation was that most kids from the other schools weren&#x27;t either. Most were along for the ride, much like myself at school sports events where you also got a trip across Europe for participating.<p>I did okay at it, and in my final year we had a couple of really good kids on the team and won the contest. I thought that was pretty cool, but I also knew I wasn&#x27;t close to the pinnacle of math contestery. Fact was I was like that kid who just had some natural ability but I didn&#x27;t really practice a huge amount and nobody suggested there was a way to get much better. If you added up all the extra practice compared to just sitting in math class, maybe it was a couple of extra lessons a week, with more interesting homework. My math teacher was a great motivator whom I still keep in touch with, but we only had him and his experience.<p>Fast forward to now and I recently discovered a friend of mine got a top 10 at the IMO. Basically top 10 among all high school kids on the planet. I can ask him the occasional interesting question that I see on FB, and he&#x27;ll know how to do it. He did proper years of training in math, had a very famous coach among others, and went through the eye of the needle several times. There&#x27;s a world of difference between that and your casual have-a-go person.
rr808超过 3 年前
Trivial to be top 5 percent of the general population at programming<p>Possible to be top 5 percent of the programmers<p>Very difficult to be in the top 5 percent of FANG programmers<p>Nearly impossible to be in the top 5 percent of Google Principal engineers
评论 #30413163 未加载
评论 #30416892 未加载
评论 #30411817 未加载
rp3超过 3 年前
More interesting is that reaching the 99.99th percentile is very difficult, and yet completely unremarkable. I love David Foster Wallace’s essay on this subject: “ Tennis Player Michael Joyce&#x27;s Professional Artistry as a Paradigm of Certain Stuff about Choice, Freedom, Discipline, Joy, Grotesquerie, and Human Completeness.” He follows a top 200 and is impressed by how absolutely amazing he is at tennis, but ultimately knows the player will know neither fame nor fortune.
评论 #30411746 未加载
ummonk超过 3 年前
The caveat is that in any activity you’re disproportionately likely to encounter those who practice that activity the most. Therefore, you could be in the 90th percentile of participants, but below average against those you encounter.
评论 #30412924 未加载
Buttons840超过 3 年前
On lichess, the 95th percentile rating of active participants is 2125. I wonder if this author would agree such a rating &quot;isn&#x27;t that good&quot;? I think it&#x27;s pretty darn good.
评论 #30412084 未加载
评论 #30412624 未加载
twofornone超过 3 年前
&gt;But I think it&#x27;s just the opposite: most people can become (relatively) good at most things.<p>I think this point of view is not only blatantly false but also much more dangerous than most people realize. It sets an unrealistic standard across all disciplines, with a wide range of detrimental effects:<p>1. People entering majors that they are not qualified for, taking on more debt and failing later than than they otherwise would<p>2. Incentives for schools and colleges to inflate grades, since people are less likely to question unrealistically high graduation rates<p>3. 1 and 2 combining in the worst case to uncrease the number of incompetent workers across technical industries<p>We&#x27;ve had decades of this sort &quot;you can do anything if you work hard&quot; propaganda in the west now and I fear that in addition to reducing competence across the board, it reduces the average person&#x27;s ability to <i>recognize</i> competence, since the default assumption is that we are all endowed with the same potential, something fundamentally incompatible with human nature.
评论 #30412267 未加载
评论 #30412089 未加载
评论 #30417494 未加载
bombcar超过 3 年前
A not quite obvious side effect of this - you should be aware of the vast majority of things you do that you are not in the top 5% - and don’t sweat it. Most things in life just need to be done, not done exceptionally well, so spend your effort on the things that matter.<p>For example, I shop pretty badly at the store, no list, no plan - not worth it to me to develop those, I can spend time elsewhere on the things I can be good at.
评论 #30411395 未加载
评论 #30411496 未加载
sheikheddy超过 3 年前
Wow, a lot of these comments seem to misunderstand that this is 95th percentile amongst people who participate, not people who practice actively. The article is mostly correct.
评论 #30420823 未加载
HDMI_Cable超过 3 年前
I think this only applies to some things, not others. If you&#x27;re in the 95th percentile of high-school students, then you&#x27;re probably getting into a good university. For university, the 95th percentile is enough to get into medicine, law, or a good business school. The 95th percentile for overall intelligence is 125. The 95th percentile for household income is $250 000.<p>For things like video games, which have a broad net of casual players and a few who are very competitive, it isn&#x27;t that good. All you need to do to get in the 95th percentile is to be competitive; but for applications where everyone is acting somewhat competitively, or for non-zero sum competitions like school or income, the 95th percentile can be very good.
D13Fd超过 3 年前
I think this is really &quot;it&#x27;s not hard to get into the 95th percentile of things that you are already fairly good at.&quot;<p>Yes, if you have a history of playing video games, it&#x27;s not hard to pick up another game and do well--and, with some focus, do very well.<p>But what if he tried to pick up playing the flute? How about drawing comics? Competitive swimming? Investing?<p>What about being in the 95th percentile of a complex field like medicine or law?<p>Assuming he doesn&#x27;t already have a background in those things, he&#x27;s probably going to have a hard time reaching the &quot;95th percentile&quot; of skill.
评论 #30418542 未加载
dang超过 3 年前
Discussed at the time:<p><i>95th percentile isn&#x27;t that hard to reach</i> - <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=22265197" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=22265197</a> - Feb 2020 (286 comments)
rapjr9超过 3 年前
In machine learning 95% accuracy is not very useful. It means the algorithm will be wrong 1 out of 20 times. So every 20th word in speech recognition will be incorrect, your medical diagnosis will be wrong for every 20th person, your face will not unlock your phone every 20th time, every 20th person will not be allowed to log into a web site. There are some machine learning applications where 95% is good enough or even great such as web search engines, but for many important applications it is nowhere near good enough and you need something more like 99.9% or better to be actually useful. Think about how often the algorithm can be allowed to make a mistake. If it&#x27;s never, then don&#x27;t use machine learning. If it&#x27;s 1 in a million you need 99.9999% accuracy. 95% looks good in machine learning papers in computer science, but in the real world 95% often means dismal failure. Think of an autonomous killer robot, is mistakenly killing 1 out of 20 people ok? For a self driving car is harming the driver or car (or pedestrian&#x2F;dog&#x2F;child&#x2F;other car) mistakenly even 1 out of 1000 (99.9%) times ok? I think not. One measure of what accuracy you need is how well people do the same task. If you can do better than people most of the time, maybe the algorithm is useful, though you also have to compare exactly how the algorithm fails to the way people fail. If people fail by choosing a fender bender and the algorithm fails by killing someone, the algorithm has to work better than people do.
评论 #30417174 未加载
评论 #30417305 未加载
评论 #30417224 未加载
mi100hael超过 3 年前
Golf is a real-world activity that can be used for comparison here. It&#x27;s a good candidate because there is a standard handicap rating system issued by a central governing body, and there are a large number of amateur players who play regularly, compared to other sports.<p>- Roughly 10% of golfers in the US maintain a handicap rating.<p>- &quot;Today, the 2,417,905 who have a handicap included 2,051,675 “active posters,” meaning they posted at least one score in 2020. However, the average number of posted scores was 38, almost double the average number of rounds played by golfers overall last year—so it’s pretty clear that those who get a handicap are among the game’s most engaged participants.&quot; [0]<p>- &quot;it has been found that on a given day the average golfer would be expected to post a score of approximately 100 strokes when following all the rules of golf&quot; [1] which is 28 strokes over par. A 28 handicap is around the 5%-ile of golfers with a handicap [2]<p>- As such, those 10% of golfers with a handicap should overlap pretty heavily with the top 10-20% of all golfers. The 50%-ile of golfers with a handicap is 13 [2], so by extrapolation that should be 90-95%-ile of all golfers.<p>- 13 is still a solid handicap. That&#x27;s someone who is breaking 90 (better than bogey golf) pretty much every round and likely has years of practice. Saying it &quot;isn&#x27;t very impressive because it&#x27;s not that hard to do&quot; would be an incorrect statement in my opinion.<p>- (as a sanity check, the 90%-ile of golfers with a handicap is a handicap of 5. Most D1 golfers will have a handicap right around 0, and D3 will be maybe 2. Given the number of casual golfers, it seems likely that only about 1 or 2 out of 100 are nearing that level.)<p>[0] <a href="https:&#x2F;&#x2F;www.linksmagazine.com&#x2F;how-do-you-match-up-against-the-average-golf-handicap-in-2020&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.linksmagazine.com&#x2F;how-do-you-match-up-against-th...</a><p>[1] <a href="https:&#x2F;&#x2F;golftips.golfweek.usatoday.com&#x2F;average-golf-handicap-america-2455.html" rel="nofollow">https:&#x2F;&#x2F;golftips.golfweek.usatoday.com&#x2F;average-golf-handicap...</a><p>[2] <a href="https:&#x2F;&#x2F;www.usga.org&#x2F;content&#x2F;usga&#x2F;home-page&#x2F;handicapping&#x2F;handicapping-stats.html" rel="nofollow">https:&#x2F;&#x2F;www.usga.org&#x2F;content&#x2F;usga&#x2F;home-page&#x2F;handicapping&#x2F;han...</a>
评论 #30418546 未加载
BasDirks超过 3 年前
After reaching that percentile the grind begins. And that&#x27;s when learning begins. And that&#x27;s when you develop character. The dopamine doesn&#x27;t flow as freely any more as obstacles to improvement become the rule rather than the exception. Braggadocious anecdote: on December 25th, 2021, I started solving Kattis problems. I set myself a goal to reach a top 1000 rank in 2022. This took me two weeks instead of an entire year. I adjusted the goal to top 100. This is significantly harder, and after two more weeks I was tempted to dismiss the goal through excuses like: &quot;these are just puzzles with limited real-world application&quot;. This is obvious bullshit, and I just wanted to avoid doing the work. I caught myself in the act and doubled down on the goal. I bought some books and learned a lot. Currently sitting on #229, and I _will_ reach top 100 this year.
screye超过 3 年前
This article belies a key assumption that is also the biggest challenge when making such a statement.<p>Distributional assumption - The assumption is that the mean and std dev of the participating class is in line with the general population.<p>Overwatch and public speaking sneakily satisfy this assumption. Everyone is forced to speak in public at some point, and overwatch is the archetypal &#x27;normie&#x27; game of the of the past few years.<p>Dan then moves from these 2 scenarios to programming, where this assumption doesn&#x27;t work. There are many formal and informal filters involved in making it an employable programmer. (Say complete a BSCS or equivalent experience). There are also supply-demand factors, where the top %ilers are presorting themselves into CS like no other time in history.<p>Landing a FANGesque job straight out of college is already putting you in the 95th percentile, because of sheer number of filters that one person passes through. Other situations with strong filtering include inaccessible games like Dota2 or high barrier of entry sports like climbing.<p>Now, I agree with Dan&#x27;s title and the contents, but for a different reason. There is perpetual headroom in the pyramid of excellence. Eventually, those better than you at a particular thing go from being a number to a tangible list of names. But, whether that should mean anything is a purely personal decision. The argument that anyone can go beyond x-percetile through deliberate practice is always true until it isn&#x27;t. That&#x27;s because we tend to take our personal IQ, resilience, discipline and acumen for granted.<p>So, I would rephrase that as:<p>&#x27;When the distributional assumption holds, 95th percentile would not be considered noteworthy by any institutions or people that the average FANG&#x2F;HN reader would find worthy of respect.&#x27;<p>Doesn&#x27;t sound as useful with the qualifications.<p>P.S: something something nuance for large scale communication is impossible. Criticizing danluu for one post lends credence to another post of his. Such is life.
dTal超过 3 年前
Sturgeon&#x27;s law: 90% of everything is crap.<p>Corollary: 95th percentile is only (approximately) 50th percentile of non-crap.
westcort超过 3 年前
The biggest takeaway I got from this was that recording yourself doing something, then observing yourself completing it as an outsider can dramatically increase your performance. Relevant content:<p>After watching a recording of myself writing code, I realized I was spending about a quarter of the total time implementing the feature tracking down which functions the bugs were in! This was completely non-obvious to me and I wouldn’t have found it out without recording myself<p>Now that I’m aware that I spent so much time isolating which function a bugs are in, I now test each function as I write it to make sure they work<p>This allows me to write code a lot faster as it dramatically reduces the amount of time it takes to debug my code
XorNot超过 3 年前
The classic &quot;oh actually that&#x27;s a 1 in 20 occurrence rate&quot; issue.
评论 #30412095 未加载
ajuc超过 3 年前
&gt; Most people probably don&#x27;t have the talent to play in a professional league regardless of their practice regimen, but when you can get to 95%-ile by fixing mistakes like &quot;not realizing that you should stand on the objective&quot;, you don&#x27;t really need a lot of talent to get to 95%-ile.<p>In the only 2 competitive games I played enough to talk about this - sc2 and dota2 - the mistakes people make at my level (which at my peak was ~2000 mmr in dota and low plat in starcraft - probably top 60-40% of the active players in both) - were nothing like &quot;not understanding game objectives&quot;.<p>In starcraft it&#x27;s easy to say what went wrong because there&#x27;s much less variables. Usually it&#x27;s going for a wrong build order blindly, not reacting to enemy unit composition in time, or just being too slow mechanically (usually in the economy part of the game, fights you can just attack-move into the enemy army, but you have to build buildings and units FAST or you don&#x27;t have enough units to fight). I&#x27;ve been watching sc2 games on top level, I&#x27;ve even played vs people in top 90% in 8v8 free for all games, and there&#x27;s just no way for me to win against them.<p>I&#x27;ve played one 8v8 game where I was the only one in plat and the rest was in master&#x2F;grand master. They left me alone for 20 minutes and fought it between themselves. I&#x27;ve build maxed 3&#x2F;3&#x2F;3 skytoss fleet with mothership and carriers, full wall of photon cannons and a lot of gates for instant remax. 20 minutes of reasonably fast base-building. After I started to fight I lost in like 5 minutes vs mass marines + medivacs :&#x2F; Guy just harassed me to death from 10 different angles while escaping my deathball of carriers. I had 10 times his resources but not the skill to defend from so many points at once. And he wasn&#x27;t even grand master. Sc2 will show you exactly how much you suck.<p>In dota there&#x27;s much more chaos, sometimes it&#x27;s lost in the picking phase, other times someone just feeds for 10 minutes, sometimes people don&#x27;t understand who is more late-game and farm for 40 minutes vs later-game carry or constantly fight vs earlier-game lineup instead of farming. Other than these I usually don&#x27;t know why we lose. It&#x27;s never &quot;I don&#x27;t understand the game objective&quot;.<p>I could probably get much higher if I wasn&#x27;t spamming winter wyvern as pos4&#x2F;5 every game. But it&#x27;s the only hero I enjoy, so whatever.
flerchin超过 3 年前
The observations about 95% percentile being good or not are distracting from the point that focused, reviewed practice is necessary to level up.
评论 #30416978 未加载
aronhegedus超过 3 年前
article definitely made me think about what else I could be 95th percentile in. I agree that dedicated practice is very important compared with people that just play casually
whoomp12342超过 3 年前
okay yeah but context matters here. Who cares about being in the 95th percentile of overwatch compared to say being in the 95th percentile of income earned?
评论 #30417186 未加载
birdyrooster超过 3 年前
Seems like their real life was Overwatch
whateveracct超过 3 年前
don&#x27;t compare yourself to others
themacguffinman超过 3 年前
&gt; Despite all of the caveats above, my belief is that it&#x27;s easier to become relatively good at real life activities relative to games or sports because there&#x27;s so little delibrate practice put into most real life activities.<p>&gt; You&#x27;re probably 99%-ile, but someone with no talent who&#x27;s put in the time to practice the basics is going to have a serve that you can&#x27;t return<p>I have a soft spot for this kind of insight because it is supposed to be inspiring and I know I should take it in that spirit, but one thing that&#x27;s always bothered me is the contradictory assertions that on the one hand, almost no one deliberately practices, but on the other hand, this makes it &quot;easier&quot;. How can both assertions be true? Any sensible definition of difficulty has to consider how many people succeed at it. If very few people succeed at a task in practice, <i>it is a difficult task</i> no matter how &quot;easy&quot; it looks in theory.<p>I&#x27;ve come to see passion &amp; motivation as indistinguishable from &quot;talent&quot;. Having the passion &amp; motivation to put in the effort is part of one&#x27;s &quot;talent&quot; and realistically is not that easy to change. Everyone knows that to write well, you need to practice writing a lot. Scott Alexander notes in his Parable of the Talents [1] that he never needed to really muster up the energy to practice writing a <i>lot</i>:<p>&gt; I know people who want to get good at writing, and make a mighty resolution to write two hundred words a day every day, and then after the first week they find it’s too annoying and give up. These people think I’m amazing, and why shouldn’t they? I’ve written a few hundred to a few thousand words pretty much every day for the past ten years. But as I’ve said before, this has taken exactly zero willpower.<p>I&#x27;ve found this to be true in a lot of walks in life. While there are certainly people with genius intellect&#x2F;ability, most of the &quot;talented&quot; people I meet aren&#x27;t of that kind, they are just effortlessly more motivated or passionate about something than I am. For example, they often started programming when they were 10 or something and spent most of their childhood choosing to do little side projects instead of the normal variety of childhood time wasters.<p>I think this essay&#x27;s (95%-ile is not that good) thesis is roughly this: to be good at something, you just need to care a little, because most people don&#x27;t care at all. Just need to care a little, that doesn&#x27;t sound so hard, right? All I want to say is yes, in my experience caring even a little is <i>so, so hard</i>. There&#x27;s a reason why most people don&#x27;t care at all and that reason is probably not &quot;you&#x27;re better and more special than most people as a reader of this essay&quot;.<p>[1] <a href="https:&#x2F;&#x2F;slatestarcodex.com&#x2F;2015&#x2F;01&#x2F;31&#x2F;the-parable-of-the-talents&#x2F;" rel="nofollow">https:&#x2F;&#x2F;slatestarcodex.com&#x2F;2015&#x2F;01&#x2F;31&#x2F;the-parable-of-the-tal...</a>