455 replies to this topic

[Heyli]

There we have it. Finally scientific proof!

 

https://www.formula1...lunjciH1Cz.html

 

I like how they are even able to calculate it down to the thousands. 

[TheFish]

What a truly awful list.

 

Kovalainen the 8th fastest driver since the 80s.

[Ben1445]

[TomNokoe]

All they have done is the usual "driver x beat driver y who beat driver z". Hardly machine learning.

 

Kovalainen is only on the list because he was teammates with Hamilton and Trulli.

 

Lando is only on the list because he is teammates with Sainz who was teammates with Verstappen.

 

And so on ...

[Kalmake]

if a driver was coming back to the sport after three or more seasons out of F1, that was also considered.

I'm sure the AI decided that one. 

[Heyli]

All they have done is the usual "driver x beat driver y who beat driver z". Hardly machine learning.

 

Kovalainen is only on the list because he was teammates with Hamilton and Trulli.

 

Lando is only on the list because he is teammates with Sainz who was teammates with Verstappen.

 

And so on ...

According to their own statement it was a bit more complex than that. I find the age comment very interesting. They found a way to see the impact of Age on laptime?

 

(just to point out, no I dont believe this is the 1 list to rule them all and we finally have a proper objective meassure...   )

 

"The team behind ‘Fastest Driver’ – headed by Formula 1’s Director of Data Systems Rob Smedley and Director of Broadcast and Media Dean Locke, and Amazon ML Solutions Lab’s Principal Scientist and Senior Manager Dr Priya Ponnapalli – also found ways and means to detect and protect against outliers like crashes, car failures, changing weather conditions in a qualifying session and even age in the data to create a fair ranking."

[PlayboyRacer]

That is literally THE worst list to do with Formula 1... in history.

And that's quite something.

[Jellyfishcake]

I can't stand these types of lists, so pointless, and by doing it on the Official F1 site makes it so much worse.

 

AWS really are awful, 

[Goron3]

Poor Caterham 2010. 

 

Incredible driver line up, awful car.

[M66R]

Erm... Not sure that's quite right.

[Marklar]

F1 challenging f1metrics in awfullness

[PlayboyRacer]

Aside from the obvious ridiculousness of Kovalainen, Barrichello, Fisichella etc... I had a chuckle at Button being there.

Don't get me wrong, Jenson was a great driver. But a banzai, one lap qualifying run wasn't a weapon of his. Whether it was a good car or not.

A ridiculous list.

[milestone 11]

AWS, look at our spectacular tyre usage graphics. Alexa can kiss my arse.

[FortiFord]

Aside from the obvious ridiculousness of Kovalainen, Barrichello, Fisichella etc... I had a chuckle at Button being there.

Don't get me wrong, Jenson was a great driver. But a banzai, one lap qualifying run wasn't a weapon of his. Whether it was a good car or not.

A ridiculous list.

 

To be fair, he did look like a Senna reincarnated in the first few races of 2009. 

[PlayboyRacer]

To be fair, he did look like a Senna reincarnated in the first few races of 2009.

See you'd think that... but out of the first 6 races he still missed two pole positions. And the Brawn was beyond dominant. Kinda proves my point!

Think Jenson only took 8 poles in his entire career iirc and had a race winning car for a few seasons.

Edited by PlayboyRacer, 18 August 2020 - 09:46.

[P123]

That is literally THE worst list to do with Formula 1... in history.

And that's quite something.

F1 Metrics runs it close.

 

Otherwise, not sure I even want to try and work out how they got the result they did.

[FortiFord]

I wonder what happened to Hakkinen, Coulthard, Frentzen, Hill, Villeneuve?

 

Particularly the latter 3, since we had Frentzen>Hill>JV>Frentzen in qualifying. I wonder how their "ML" handled this. 

[PlayboyRacer]

I wonder what happened to Hakkinen, Coulthard, Frentzen, Hill, Villeneuve?

Particularly the latter 3, since we had Frentzen>Hill>JV>Frentzen in qualifying. I wonder how their "ML" handled this.

Mansell? K.Rosberg? Montoya? Raikkonen?

Though I suppose not everyone can make a top 20 but still this list is gold.

[milestone 11]

Where's Gilles and Ronnie Alexa? "Good morning milestone, Amazon Machine Learning Solutions Lab asks who?"

[Broekschaap]

I wonder what happened to Hakkinen, Coulthard, Frentzen, Hill, Villeneuve?

 

Particularly the latter 3, since we had Frentzen>Hill>JV>Frentzen in qualifying. I wonder how their "ML" handled this. 

The machine was driving circle's on them, concluded that was not F1 but Indy and therefore excluded them!

[FortiFord]

Where's Gilles and Ronnie Alexa? "Good morning milestone, Amazon Machine Learning Solutions Lab asks who?"

 

It only uses data from 1983 onwards. 

[milestone 11]

It only uses data from 1983 onwards.

Exactly. In other words, it's a waste of bloody time. Did the world start in 1983, could have sworn that I have memories from before that. Maybe it was all just a bad trip.

[Spillage]

Yeah, it doesn't work at all. As good as he was, Heikki Kovalainen was not faster than Alain Prost.

[as65p]

Even if Gilles' and Petersons period was in there, it's stats based and their stats against teammates aren't as brilliant as legend would have you believe.

[DeKnyff]

I have seen along the last decade (including this forum and other motorsport sites) several attempts to "scientifically" rank the drivers of different periods. Many of those attempts, if not all, made by the same guy appearing under different nicks on several places. Well, maybe now he has convinced Amazon and Liberty.

 

I respect the amount of work that such an initiative brings, but the result is completely pointless, because it's all based on circular comparisons and it depends a lot on who your teammate is. For example, Leclerc would receive very few points for beating Ericsson at Sauber, but it doesn't mean he was driving any worse that when he beat Vettel. Also, drivers speed is not constant along their career.

 

There is no way you can "scientifically" compare drivers, it will always have a subjective component. In fact, that's one of the things which make this sport so attractive: you can discuss for ages if a driver won because of the car or because his skills and never reach a conclusion.

[ANF]

When will the machines learn what is a waste of time and computing power?

[Loosenut]

Here's my shopping list, I think you'll find it makes far better reading.. 

 

*bread

*milk

*tea bags

*tomato ketchup

*eggs

*beer

*beer

*toothpaste

*autosport

*bananas

*peanut butter

*phil collins cd

*choccy biscuits

*underpants

*beer

[noriaki]

The F1 Metrics guy at least makes a real effort, properly explains what factors the machine uses to arrive at the conclusions it makes (also displaying us data that is far more interesting than the actual ranking they produce) and recognizes his system isn't perfect. The criticism on him and comparisons to this are way unfair when these guys just pretend that the world started in 1983, that "fast in qualifying" equals "fast" and that this list is somehow "objective". 

[Heyli]

I have seen along the last decade (including this forum and other motorsport sites) several attempts to "scientifically" rank the drivers of different periods. Many of those attempts, if not all, made by the same guy appearing under different nicks on several places. Well, maybe now he has convinced Amazon and Liberty.

 

I respect the amount of work that such an initiative brings, but the result is completely pointless, because it's all based on circular comparisons and it depends a lot on who your teammate is. For example, Leclerc would receive very few points for beating Ericsson at Sauber, but it doesn't mean he was driving any worse that when he beat Vettel. Also, drivers speed is not constant along their career.

 

There is no way you can "scientifically" compare drivers, it will always have a subjective component. In fact, that's one of the things which make this sport so attractive: you can discuss for ages if a driver won because of the car or because his skills and never reach a conclusion.

I think the main problem is the headline. Just dont present it as a list of fastest drivers, but present it as a summary of data. I like those lists, as also based on data you can still see some interesting outliers (like Kovalainen in this case), which are then worth discussing maybe. 

 

At the end, it's just qualifying. It's true that drivers to push more in Qualifying, but of course there are also the cases where drivers purposefully focus less on Q to benefit them in the race. to then take any analysis on qualifying (putting aside that surely the analysis itself has some faults already) as a claim that a driver is fastest is just plain wrong. 

[monolulu]

Wonder if they used the same algorithm as the prediction of A-level results in the UK? 

[SonGoku]

If a clown on twitter made this it would be funny, but this is the official F1 organization, how embarrassing.

[Lights]

Aside from the obvious ridiculousness of Kovalainen, Barrichello, Fisichella etc... I had a chuckle at Button being there.

Don't get me wrong, Jenson was a great driver. But a banzai, one lap qualifying run wasn't a weapon of his. Whether it was a good car or not.

 

For most of his career, 10 out of 17 seasons, he was partnered by teammates that, according to this list, are in the top 11 of fastest qualifying drivers. 

 

Taking that into consideration it's not that ridiculous that he's on this list in 17th position.

 

Yes he might be slow compared to most of his teammates, but if those are all considered to be among the very fastest, then Button can still be faster than most other drivers that were not teamed up with the likes of Hamilton and Alonso.

 

The same argumentation holds for Kovalainen and Barrichello. We know them from being slower than Hamilton and Schumacher. But perhaps that's clouding your judgement of them.

 

It's funny how these mathematical lists are often so quickly ridiculed and dismissed.

 

Yes of course, your perception, that's the only correct version of reality, and make sure you don't let anyone change that.

[Baddoer]

Love how Hakkinen is not even in top 20. Hell, man did beat fastest driver in history in their first qualifying together.

Edited by Baddoer, 18 August 2020 - 10:27.

[ensign14]

That is literally THE worst list to do with Formula 1... in history.

And that's quite something.

There was a worse one.  It had things like Patrese above Prost.

 

And there was this one, which ranked Christian Fittipaldi as the 12th best driver ever.

[PlayboyRacer]

Love how Hakkinen is not even in top 20. Hell, man did beat fastest driver in history in their first qualifying together.

To be fair Berger did the same thing.

[NixxxoN]

I thought it was a reasonably list until I saw Kovalainen in 8th.

[PlayboyRacer]

For most of his career, 10 out of 17 seasons, he was partnered by teammates that, according to this list, are in the top 11 of fastest qualifying drivers.

Taking that into consideration it's not that ridiculous that he's on this list in 17th position.

Yes he might be slow compared to most of his teammates, but if those are all considered to be among the very fastest, then Button can still be faster than most other drivers that were not teamed up with the likes of Hamilton and Alonso.

The same argumentation holds for Kovalainen and Barrichello. We know them from being slower than Hamilton and Schumacher. But perhaps that's clouding your judgement of them.

It's funny how these mathematical lists are often so quickly ridiculed and dismissed.

Yes of course, your perception, that's the only correct version of reality, and make sure you don't let anyone change that.

I see what you're saying but I still don't think Button had searing one lap speed. And that's what this list is about. Wringing a hot lap.

If it was a list for the top 20 in history in mixed conditions, I'd have Jenson in there no question.

[absinthedude]

Why start at 1983? That rules out one of the *fastest* ever, Gilles Villeneuve. It rules out Jim Clark, Stirling Moss, Jochen Rindt, Ronnie Peterson.

 

But even taking the data they chose....no Mansell? Not the best all round F1 driver but one of the fastest....as was Keke. Where are they? 

 

Kovalainen in 8th just shows what a poor list this is. 

 

Poor effort f1.com, go back and redo from start.

[Fisico54]

Doesn't seem to take into account that especially in the 80s qualifying was not as important as now whereas race setup had more importance. Presumably was Prost is 20th and only he and Senna feature from the 80s.

[Beamer]

Too stupid to comment on. F1 and AWS are making a fool of themselves.

[Astandahl]

The GOAT is in the top 10

 

 

P.S I'd like to know if they included MSC "second career" in the dataset. Probably i suppose considering Rosberg place.

[Jazza]

Amazing! Apparently only 18 drivers in the last 37 years can get within half second of Senna. But the driver who actually could get well within that time and actually beat him (Mika) isn’t on the list...

[P123]

Here's my shopping list, I think you'll find it makes far better reading.. 

 

*bread

*milk

*tea bags

*tomato ketchup

*eggs

*beer

*beer

*toothpaste

*autosport

*bananas

*peanut butter

*phil collins cd

*choccy biscuits

*underpants

*beer

Shocking- I wouldn't serve you.

[femi]

Another joke

[as65p]

To be fair Berger did the same thing.

And Irvine to Schumacher.

 

To be fair, even if there was a hypothetical method to compile such lists with that multitude of factors involved and a perfect algorithm, i.e if it would be a scientifically perfectly legit list (which obviously is impossible), even such a list would still stir the same controversies among fans because our favorites are in our minds always placed to low and the drivers we hate placed too high.

 

It's just to get clicks, and people talking. Works every time, and you need no science to figure that out, just very basic knowledge of the human mind. 

[TheWilliamzer]

What a fraud that fella Hakkinen was...

[Jazza]

I would love to see the full list of this. If driver 20 is already half a second down, then what does driver 200 look like?

I wouldn’t be surprised if there are some big names further down who are 2-3 seconds per lap off because the maths is just BS.

[Celloman]

That is literally THE worst list to do with Formula 1... in history.

And that's quite something.

It's actually pretty good considering it was generated with machine learning. Add Hakkinen and remove Kovalainen and you'd have a pretty accurate list of top qualifiers.

[TheWilliamzer]

Where is Pastor? He was a good qualifier VS Rubens and Valtteri. 

Edited by TheWilliamzer, 18 August 2020 - 10:58.

[shure]

Nelson Piquet was once called King of Pole before Senna came along.  And now Sainz is faster than him, apparently.  And Kubica?  Er, ok.