Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

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

danluu.com

Post reply on HN
Print view View original

171–180 of 191 posts

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

#171

Post by aaron695 »

JohnFen wrote: post #151
brigadier132 wrote: post #72

I think the way this is phrased is weird. It should probably be rephrased as "becoming 95% percentile at a thing is not that hard". Which I might agree with. Becoming good at a lot of tasks and skills really just comes down to finding the right approach to training and then doing that training. The last 5% is much harder because that's generally filled with truly talented people that also know how to train and are di…

I think it was Adam Savage who put it this way (talking about the importance of talent): Getting 90% of the way to mastery of a thing is just a matter of practice. Talent is what gets you that last 10%.

[dead]

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

#172

Post by throwaway22032 »

I guess it's a perception thing but I feel as if this becomes more intuitive if you look at it as 1 in 20.

Amongst a group of 20 people you know, assuming you haven't just picked 20 people with highly correlated hobbies (e.g. 20 software devs that like to game), each person is probably going to have "their thing" and they'll be the 1 in 20 or 95th percentile in the group at it. Extrapolate out and there you go.

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

#173

Post by jwestbury »

tester756 wrote: post #90
brigadier132 wrote: post #72

I think the way this is phrased is weird. It should probably be rephrased as "becoming 95% percentile at a thing is not that hard". Which I might agree with. Becoming good at a lot of tasks and skills really just comes down to finding the right approach to training and then doing that training. The last 5% is much harder because that's generally filled with truly talented people that also know how to train and are di…

last 5%? I've been top 1% and slightly below 1% at peak in video games that have milions of gamers and I wouldnt say that top 5% is filled with "truly talented people" Top 5% is just decent people, but they are very, very, very far from top 0.1, 0.5%, etc. At 1% I've been faaaaar from 0.5%, 0.1% people.

Likewise -- I played in CAL-M in Counter-Strike back in the day. This statement dates me, I realize, but for those who aren't aware, CAL was the top online Counter-Strike league, and CAL-M was the second tier. IIRC, there were ~10 teams in CAL-I, the top tier, and about 40 teams in CAL-M, separated into multiple regions. (Lower tiers were CAL-IM, i.e., intermediate, and CAL-O, i.e., open - anyone could join CAL-O, but getting past that required promotion.)

Anyway, point is, I was probably in the top ~1000 Counter-Strike players in North America in the early to mid aughts. But the gap between CAL-M and CAL-I was just ridiculous -- we're talking about the gap between a college sports team and a pro team. The average CAL-M team wouldn't win a single match in a CAL-I season.

So, yeah - the gap is increasingly large the closer you get to the top of a given pursuit.

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

#174

Post by antisthenes »

mlyle wrote: post #93
antisthenes wrote: post #89

Earlier quoted context omitted.

As someone who used to be an avid gamer and reasonably competitive, "a little bit of careful self-reflection" is definitely not enough to get you in the top 5%. Top 20%, sure, because the bottom 50% doesn't even try, and the next 30% tries just enough to learn the basic mechanics. You need a certain time commitment and the ability to learn reasonably complex abstract mental models. Learning curves for many games are…

The author describes people who have >1000 hours but make basic mistakes that are easy to fix and hold down their rating. The author also describes the experience of people who are stuck, receive a small amount of quality coaching, and then massively increase in rating.

Yes, both of these exist. I just disagree that these 2 phenomena are conclusive proof of the title/hypothesis of the article.

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

#176

Post by mewpmewp2 »

lamontcg wrote: post #166

> This is a pretty fundamental disagreement that I have with a lot of people. Just for example, I recently got into a discussion with an authority who insisted that it wasn't possible for me to reasonably disagree with them (I suggested we agree to disagree) because they're an authority on the topic and I'm not. It happens that I worked on the formal verification of a system very similar to the system we were discuss…

If you don't have time to argue then just say that, don't say that it isn't possible for someone to reasonably disagree.

I agree with the sentiment that you should consider the arguments, not whether someone is an "authority". I enjoy arguing and I never start by mentioning whether I have relevant experience and if someone assumes I don't, it's clear to me that they are not really considering the arguments themselves, but assume I must have no experience.

And when I'm arguing with someone I only focus on their arguments, not who they are.

It's also not about them being a dick. It's about them indicating that they are not really intellectually honest.

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

#177

Post by mewpmewp2 »

CSMastermind wrote: post #46
madamelic wrote: post #26

I would have to agree about 95th percentile being not that difficult. A streamer that I occasionally tune into (Asmongold) says that if you are seeking information outside of the game such as watching streams of the game, that makes you a top 10% player automatically. Most people are passive consumers and participants in their activities and that isn't exclusive to the individual. I am a passive consumer & participan…

> Most software engineers aren't listening to podcasts about software engineering. If you are religiously listening to and learning from podcasts, that makes you more suspectible to being a higher percentile performer in software engineering. Does it though? I don't feel like listening to podcasts is a particularly good way to build software engineering skill. The same way listening to a podcast about art doesn't mak…

I agree, I think building things is the main thing that makes someone better. So building side projects and experimenting with different tech besides your work.

Podcasts I think are quite superficial. They can only be beneficial if you are critical about what they are saying, but to be critical you need to have already made something yourself. Otherwise you are blindly learning facts like you said.

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

#178

Post by lamontcg »

mewpmewp2 wrote: post #176
lamontcg wrote: post #166

> This is a pretty fundamental disagreement that I have with a lot of people. Just for example, I recently got into a discussion with an authority who insisted that it wasn't possible for me to reasonably disagree with them (I suggested we agree to disagree) because they're an authority on the topic and I'm not. It happens that I worked on the formal verification of a system very similar to the system we were discuss…

If you don't have time to argue then just say that, don't say that it isn't possible for someone to reasonably disagree. I agree with the sentiment that you should consider the arguments, not whether someone is an "authority". I enjoy arguing and I never start by mentioning whether I have relevant experience and if someone assumes I don't, it's clear to me that they are not really considering the arguments themselves…

Actual people rarely argue to some kind of platonic ideal.

And the author of the post clearly indicates that he didn't bother to inform this person that he had expertise that he considers relevant. Elsewhere he also kind of shits on using "terms of art". But those "terms of art" are useful shorthand to experts, establishing that both sides are experts and familiar with the field, and more rapidly communicating information with less words.

That strongly makes me suspect that the problem is one of communication.

The author then asserts that the argument from authority makes him dismiss the expert. Well, that's just the fallacy-fallacy[1] and is no more relevant. The expert can still be 100% right and you can be 100% wrong, and they can be using a logical fallacy and all you can really conclude is that they're a bit of a dick about it. You can't ascertain truth.

If you want to have productive discourse with that expert, you might ask yourself why they're being a bit of a dick about it, and if you've done anything to warrant that breakdown in communication on a /social/ level rather than intellectual-technical level. Or the person may just be a dick. And they might still be a dick and be wrong, we can't really tell. But really from reading that paragraph all I gather is that there was certainly a breakdown in communication, and that it reflects negatively on the author that he assumes he's right because of it. And there's a certain amount of hypocrisy apparent in his conclusion.

[1]https://en.wikipedia.org/wiki/Argument_from_fallacy

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

#179

Post by wing-_-nuts »

hirvi74 wrote: post #168
wing-_-nuts wrote: post #103

Earlier quoted context omitted.

I'm in the top 15 bf4 tankers in the US. I run a very unique loadout that's very specific to my PTFO playstyle and the most common threats I deal with. It has glaring weaknesses that few ever exploit, but if you knew the game and actually tried to counter me, you absolutely could. Sometimes, the .001% have simply found their particular niche in the ecosystem

I miss that game so much. What's the server population like these days? What is your load out? I was a Jet main.

Ah, there's probably still 500 players still playing on PSN. Probably the same on XBL and maybe 2x that on PC I'd guess. The only problem I see is a 'dead sea' effect, where a lot of the folks still playing are very good, and the games can be lopsided and not fun for the average player. I hate to see them stomp on average dudes. I always hold back capturing the last objective, and I never shoot into deployment even if I'm taking fire. The game should be fun and fair for everyone.

As I said, I'm a tanker. I run smoke screen, thermal camo, and LMG. It's an insanely fun play style because everyone relies on their minimap and thermal optics and I'm basically a 70 ton ghost on the battlefield. It's crazy how easy it is to sneak up behind enemy armor or juke incoming shells because everyone is looking for a white blob or a red marker on their map.

Also, given I'm almost unspotable, aircraft tend to leave me alone unless i'm dealing with a pro jet pilot that's looking for me. Sometimes I will piss off the enemy team and by the 3rd strafing run I'm like 'wow bud, you really got nothing better to do with your day, huh?'

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

#180

Post by TehShrike »

theoldlove wrote: post #158
TehShrike wrote: post #142

Earlier quoted context omitted.

I aim to be a 95th percentile parent. It's not as hard as you'd think

How can you tell? Parenting isn’t exactly quantifiable.

Most 5-year-olds are assholes. Once you realize that it is possible to train a child to not act like an asshole, you're pretty far ahead. Once you've trained your 5-year-old to not act like an asshole most of the time, you're easily in the top 5%
Post reply on HN