AllTimeFootballer – a head-to-head experiment to rank footba

[goat49]

Hi everyone,

I've been working on a small football project that started from a simple question: is it possible to create a meaningful all-time ranking without simply asking people to name their GOAT?

I put together a database of 660 players from different generations, from the great players of the 1950s and 60s through to today's stars.

Instead of choosing a Top 10, the site gives you two random players and asks one simple question: Who was the better footballer?

Every choice contributes to a community Elo rating, so over time the idea is to see whether thousands of individual head-to-head decisions can produce an interesting all-time ranking.

The community voting has only just started, so the current table is still very volatile – Ronaldinho is leading at the moment – but that's part of the experiment.

I'm particularly interested in getting votes from people who know football history, rather than ending up with a ranking dominated simply by the most recent generations.

There's no registration required and it's completely free:

alltimefootballer.com

I'd also be interested in any feedback – especially if you notice an important player I've missed from the 660.

Thanks!

[grantham]

What's to stop someone visiting the site and just spamming player A or player B continuously without even taking either player into consideration, just totally screws your numbers.

Also what is the point of asking about two random players someone may not have heard of either of them. Someone born in the 80's may have no idea about a German player from the 50/60 and as far as I can see you can't opt out of a choice

Nice idea, but the application is deeply flawed
_________________
www.programmehut.com - The Internet Dealers
archives.football Results, line-ups, profiles, programmes, player images and much more

[goat49]

Thanks for taking the time to give detailed feedback – I genuinely appreciate it.

On the first point, there is already a rate limit to make rapid spam voting harder, although you're right that no open community voting system can completely prevent someone from deliberately making meaningless choices. That's something I'll have to keep an eye on as the number of votes grows.

Your second point actually made me think more. My original idea was that the random cross-era matchups would encourage people to come across players they might not know and perhaps even learn something about them. As a football nerd myself, I probably assumed too much that other people would approach it that way.

But you're right that forcing someone to choose when they don't know enough about one or both players can actually make the ranking worse rather than better.

I'm therefore considering adding a "Don't know enough / Next matchup" option which wouldn't count as a vote or affect the Elo ranking. I think that could preserve the cross-era discovery aspect while avoiding forced guesses.

Longer term, I'd also like to provide more information about the players directly on the site, so encountering an unfamiliar player can become part of the experience rather than a reason to leave.

It's still a very early version, so this is exactly the kind of feedback I was hoping to get. Thanks again for testing it.

[Pete’s Picture Palace]

Great idea but.......... Pele, Puskas, Cruyff, George Best - and others - all midfielders?

That's one problem with this, categorising players in certain positions.

[goat49]

That's another fair point. The current position categories are deliberately quite broad, but I agree that they become problematic with players like Pelé, Puskás, Cruyff or Best – especially since their positions sometimes changed over the course of their careers. Beckenbauer would be a similar case. With the players mentioned, my own perception has at least always been that a “playmaker” (Pelé, Puskás, Cruyff) falls into the midfield category, and I would have put Best there as a winger as well.

At the moment, the positions don't have any influence on the voting itself. But in the long term, it looks like I'll have to look into primary and secondary positions as well.

These are exactly the kinds of issues I'm hoping to come across at this stage, so thanks for pointing it out.

[goat49]

Thanks again for the feedback yesterday. I thought about your point regarding matchups where a user doesn't know one or both players well enough to make a meaningful choice.

I've now added a “Don't know enough -> Next matchup” option. It simply skips the matchup without recording a vote or affecting either the personal or community Elo ranking.

I still like the original idea of encouraging people to come across players from different eras and perhaps learn about players they didn't know before, so longer term I'd also like to add some short information about each player rather than relying on the skip button alone.

The discussion about positions is on my list as well. Before I introduce position-based rankings or an All-Time XI, I'll need a more nuanced system for players who played in different roles during their careers.

Really appreciate the constructive criticism – this is exactly the kind of feedback I was hoping to get while the project is still at this early stage.

[goat49]

Another small update following the discussion here.

I've now added separate Community Rankings for each era: 50s/60s, 70s, 80s, 90s, 2000s, 2010s and 2020s.

The Elo rating itself remains global, but the era filters make it much easier to see how players compare within their own generation.

It's obviously still very early – apparently Arjen Robben is currently both the overall GOAT and king of the 2010s, while Antonio Rüdiger leads the 2020s!

I'm interested to see whether those rankings start looking rather different once more votes accumulate.

Thanks again for the feedback here. A few of the suggestions from this thread are already shaping what I'm changing next.

[goat49]

Another small update following the discussion here.

While looking at the new era rankings, I noticed that players from the 2020s were a little too prominent in the overall Top 50.

The reason was quite simple: the completed eras have 100 players each, while the 2020s currently only have 60. That meant those players were getting more matchups per player and therefore more opportunities to move their Elo rating.

I've now adjusted this. Matchmaking also takes overall player exposure into account, and matches between two players from an incomplete era are weighted according to the size of that era. Cross-era matches still count fully.

I recalculated the existing votes using the new system and Pelé is now narrowly leading the overall ranking ahead of Mbappé.

The rankings are still very young, of course, so I'm curious to see how they develop with more votes.

Thanks again for the feedback here – it was actually the discussion about eras that made me look more closely at this in the first place.