It's cool, but dealing with bad-faith participants is what is really difficult about these types of exercises. It works great until the hurt people find it, and then the hurt people hurt people -- perpetuating their sadness unto others by upvoting the bad things.
I am glad that Hitler isn't #1 on this list anymore. Because Hitler, despite his best efforts, could only be a distant 2nd when pitted again the true #1 best human being : Rupert Murdoch.
I don't think higher rating is an endorsement. The question is only: which do you rank higher? I think Adolf Hitler ranks quite highly on notoriety, especially if the comparison is inherently notorious (ie. Hitler vs. Jeffrey Epstein.) So I'm not sure Hitler ranking highly is in bad faith.
> So I'm not sure Hitler ranking highly is in bad faith.
I think they may be ranking Hitler the artist or Hitler the leader surely. It can't possibly be the more simple explanation of the internet being full of trolls who think anonymous graffiti is funny.
Why is it only on HN that this sort of contortion happens?
ELO in most contexts is a sort of "strength" rating so I decided when doing a few dozen that I would choose whichever had more impact on history. Or to put it another way, I was choosing the thing that would cause the most differences if it didn't exist. In my case I got Hitler vs Electricity so the latter is the obvious choice but anyone else using the same criteria as me has perfectly good reason to choose Hitler over some random sport team or most other individuals.
Though yes it's the internet, so it's probably mostly people doing it for the funny.
Maybe the site needs a quick explainer of what Elo ranking is. Perhaps people should have to pass a multiple choice question about it before they can start ranking.
Conflating Elo ranking with goodness is shockingly dimwitted - I'd think it much below Hacker News commentators, but apparently that's not the case. Of course Hitler would have a high Elo ranking; he's objectively one of the most powerful and influential humans of all time.
Its at 2031 elo now, while the closest to it is 1500 elo. Lot of catching up to do but I think there is still good chance to catch it before it becomes much harder.
Unironically, I want this, but for restaurants/stores/etc, essentially a replacement for Google Maps' helpful but deeply flawed one-to-five-star (but actually 4.0-4.9-star) rating system.
Sorry no lol, not the randomness aspect. What I mean is that Google has a very good guess of which restaurants I've been to, so rather than asking me to rate it 1-5 stars, I want it to ask "Is it better or worse than <Other Restaurant You've Been To>"? and then building ratings based on those results, with the option of me answering " I have not been there".
It just seems like better system than the current one, in which the Papa John's near my undergrad college and the best Italian place in my neighborhood are both rated 4.5 stars.
When you search for something on Yelp and you get 10 results, to places you've never been to before, and despite not visiting any of them beforehand, you choose one. That's what's happening here, and that's what's happening in Google search results. No difference.
... but that doesn't help informing another person's decision. Someone should track all of everyone's restaurant visits and quiz them in this style and publish the data for everyone's benefit
I wonder if ELO makes sense for something like this that is so inherently unstable/not-normally distributed? I.e., in chess, if you're playing an opponent whose rating is +800, you have less than 1% chance of beating them. But in this case, we have Peppa Pig currently at 1875, vs. the BMW 3 series at 1072. Can we really say with any confidence that Peppa would beat the BMW 99 times out of 100? It seems unlikely that increasing the number of matches will change the instability of this rating.
In it he mentions that he kept the site up for only a short amount of time, in part to prevent anyone from writing a bot which could game it. Unfortunately, that has not been the case with Elo Everything, which has a clearly gamed leader board.
What’s the practical application of this? It makes sense in chess where you have two players using the same rules, so a comparison makes sense. But is Postgres a better database than Redis? Is The Godfather a better film than Lord of the Rings? These questions make no sense. Political candidates can’t get an increased value simply for being less popular.
So is the idea that you use this in a very technical space for, say, ranking jobs in a queue? Or is it intended for an e-commerce site to order by product ratings?
I suspect I'm overthinking it even more! I couldn't even come up with an answer to the first matchup because I have no idea what I'm supposed to base my ranking on.
Idea would be great for product recommendations, as long as the spam issue can be solved. Most people can make a reliable comparison between two products but maybe not among many.
Something must have broke/exploited. Hitler is still #1 with over 2000 matches but Baseball is in #2 with over 100M matches. The next 20 are 1000-2000 matches.
Considering this thing ranks Adolph Hitler as the best "thing" currently, above other things such as sexual intercourse, the entire Earth, and Oxygen, I don't think I would trust any product recommendation such a system would spit out.
Well that's because of spam. If you ask people to genuinely make a bunch of 1v1 comparisons from first hand experience, vs say, blogspam of top 10 ____ products lists, or recommendations of people who have only tried 1 option...
It's cool, but dealing with bad-faith participants is what is really difficult about these types of exercises. It works great until the hurt people find it, and then the hurt people hurt people -- perpetuating their sadness unto others by upvoting the bad things.
[1] https://imgur.com/a/mQFsg7x
I think they may be ranking Hitler the artist or Hitler the leader surely. It can't possibly be the more simple explanation of the internet being full of trolls who think anonymous graffiti is funny.
Why is it only on HN that this sort of contortion happens?
Though yes it's the internet, so it's probably mostly people doing it for the funny.
Conflating Elo ranking with goodness is shockingly dimwitted - I'd think it much below Hacker News commentators, but apparently that's not the case. Of course Hitler would have a high Elo ranking; he's objectively one of the most powerful and influential humans of all time.
[0] https://en.wikipedia.org/wiki/Prince_Family_Paper
It just seems like better system than the current one, in which the Papa John's near my undergrad college and the best Italian place in my neighborhood are both rated 4.5 stars.
Each outcome adjusts the winner's and loser's Elo ratings according based on the current difference.
For example:
- A 1500 elo beats a 1300 elo. The winner gains 5 and the loser loses 5.
- A 1300 elo beats a 1500 elo. The winner gains 15 and the lower loses 15.
The math is such that a 100 point difference is 2:1 advantage. A 400 difference is 9:1 advantage.
[1] https://en.wikipedia.org/wiki/Elo_rating_system
I.e. a scissors-paper-rock scenario.
https://en.wikipedia.org/wiki/Electric_Light_Orchestra
In it he mentions that he kept the site up for only a short amount of time, in part to prevent anyone from writing a bot which could game it. Unfortunately, that has not been the case with Elo Everything, which has a clearly gamed leader board.
I guess the optimal strategy would be to target the leader you want to knock out with losses to the current last-place entry, right?
I guess that last place entry would change pretty quick though.
Yup, this could work
[0] https://support.torproject.org/about/why-is-it-called-tor/
[1] https://www.britannica.com/technology/Tor-encryption-network
So is the idea that you use this in a very technical space for, say, ranking jobs in a queue? Or is it intended for an e-commerce site to order by product ratings?
This feels like retro Internet, I love it.
How could you ever choose?
At the time of writing, Adolf Hitler is #1 on the leaderboard with Elo 1754.
And democracy - "I love democracy!" - at #15
1. https://youtu.be/ALy6e7GbDRQ
1) Milk 2) Toronto 3) React
Thanks for the laugh! Fun little app
Actually I had the idea it would match people up who had similar views.
For several reasons the choice isn't that obvious.
Maybe the API is just very simple and somebody is just sending upvotes to Hitler infinitely.
Yep, it has no authentication. You can just keep posting to the API with Hitler beating everything.
Yeah I think someone gamed this
Predictably, someone is gaming it and spamming Adolf Hitler.
1. People that think its funny and choose Hitler
2. People who compare Hitler to something like "Ferrari" or "Rome, Italy" and go "yeah he's had more impact (from my perspective)"
In that same sense, Hitler is arguably one of the greatest leaders the world has ever seen. He was just a horrible person who did horrible things.
I am (obviously) not a fan at all of Hitler, Nazism, etc. but it's had a profound impact on our world, regardless of whether it was horrible impact :(
But yeah, it's obvious there is some gaming of the system. Hard to rank Hitler higher than "earth" or "electricity"
(Also the history of Rome has definitely been more impactful than Hitler.)