Is the Definition of Reject Obsolete?

Aug AFormula One, like all sports, has been subject to evolution throughout its history. The rear-mid engine revolution, the rise of ground effects, the commercialisation of the sport under the influence of Bernie Ecclestone and the #F1Sprint are just a handful of examples of the drastic change competitors in Formula One have to adapt to.

One of those revolutions was the metagame shifting towards reliability in the noughties. As part of “cost-cutting” rules, teams were tasked to make sure engines and gearboxes could last multiple races. The teams managed to meet that challenge successfully, reducing the number of technical issues across the board and accelerating the trend that was already beginning in the seasons before: the guiding principle of Formula One car design became the classic phrase “To finish first, first you have to finish”.

Of course, this plays a crucial role in determining who is a reject and who is not a reject. When drivers ahead of you are less likely to retire, it means you are less likely to take one of the positions required to rid yourself of reject status if you lack the outright pace to take it by force.

That point has led to some discussions on the GPR forums and the GPR Discord in recent years on whether the criteria for being a reject (which are, for the most part, based on the criteria used on the former site) are obsolete and do not do the current state of the sport justice. They were, after all, designed for a time when only six cars scored points and the number of retirements was much higher. Is there a need for a new definition of criteria to not punish modern drivers for driving in an era where there are just no opportunities to escape reject status in the way Arturo Merzario, Bob Anderson and Christian Danner were able to?

This column will analyse whether it is harder to not be a reject these days and eventually determine whether a change of the definition of reject is advisable or not.

Since hyper-reliability is the main argument for changing the definition of a reject, we obviously need to determine how reliability has changed.

Average number of technical retirements per race (1987-2019)

Average number of technical retirements per race (1987-2019)

This graph shows, as its title implies, the number of technical retirements per race from 1987 to 2019.   Note that these numbers do not actually reflect the amount of non-classified cars as driver or team errors that resulted in retirement are not counted. This will cause objections, but those will be addressed later in the article.

The first instinct is to look at that graph and consider the point proven and the case closed. Technical retirements have been trending downwards massively and so it is much harder to lose your reject status, requiring a change of the definition. That assumption, however, leaves out a crucial factor: grid size.

Simple math test: you can enter a race with a 30-car grid with 15 technical retirements or a 20-car grid with 6 technical retirements. Assuming your car’s relative competitiveness is identical, in which race is finishing in the top 5 more likely?

It is fairly obviously the 20-car grid. Therefore, it is important to account for grid size in determining whether good finishes are less likely because of reliability.

For that purpose, GP Rejects presents a new stat: Expected Finishers.

To calculate it, the number of technical retirements in a race is subtracted from the normal number of cars starting a race in that season. To demonstrate the calculation: the 2021 British Grand Prix had one retirement for technical reasons (Sebastian Vettel, overheating issues) and normal grid size in Formula One in 2021 is twenty cars; 20-1 = 19. The expected finishers for the 2021 British Grand Prix therefore were 19. That is the reason why the starting date of 1987 was chosen in the earlier graph, as that was the first time in Formula One history grid sizes were pretty much equal for all races.

Average number of expected finishers per race (1987-2019)

Average number of expected finishers per race (1987-2019)

This graph, showing the expected finishers per race from 1987 to 2019, demonstrates that the actual impact in increased reliability is not as severe as the sheer reduction of technical failures would suggest. While there is a trend of increased expected finishers, that increase is much more subtle. For the most part, the number of expected finishers correlates with the number of teams on the grid. The major spike in the early 2010s was a direct production of the development of the number of cars. The spike from 2010-2013 was in large part related to the arrival of HRT, Virgin and Lotus; their demise led to a drop-off, before the birth of Haas caused another minor spike. To emphasise how marginal the increase is, the final data point (2019: 18.6 expected finishers/race) is equal to the expected finishers/race in 1996. 

Of course both of these numbers do point to an increased difficulty based on competitor unreliability. However, there is a simple way to truly prove whether the standard is too harsh: finding out whether the number of drivers who have managed to lose their reject status has decreased. Of course, this test also requires accounting for grid size: logically less drivers can lose their reject status if less drivers are on the grid, period.

Number of drivers that entered races and drivers who lost their reject status per season (1989-2019)

Number of drivers that entered races and drivers who lost their reject status per season (1989-2019)

As this graph demonstrates, the number of drivers that lose reject status is dropping, but is doing so virtually in parallel to the general number of drivers competing in a season, which is getting lower. If anything, the number of drivers is shrinking more than the number of drivers losing their reject status per season, albeit with a hardly noticeable difference.

So the number of relevant retirements is down, and drivers still lose their reject status as much as they did 30 years ago. With all those numbers, it is safe to say that the reject criteria decided by GP Rejects is still valid. Conceptually, Formula One is still close enough to the time period when the concept of a Formula One reject was established and the author sees no way it could realistically change to the point a different definition is needed. It would require NASCAR-sized grids with current levels of reliability before the author would suggest needs to shift its approach and even in the budget cap era, that will not be happening.

Still, contrary to what many people believe, “wrong” beliefs do not just randomly pop up in people’s minds. So the question of why some readers think the reject system may not work as well as it used to is worth analysing.

Of course, one factor is a simple psychological aspect. Even if a Formula One driver were to read, it is pretty safe to assume that they would not be very concerned with their reject status in the heat of an actual Grand Prix. Their main interest is getting championship points for themselves and their teams. With points being handed out to the top-10 now, there seems to be a bigger temptation to be satisfied with e.g. an eighth place, which does not help lose reject status, but adds four valuable championship points. That luxury did not exist when the definition of “reject” was created: P8 did get you nothing, so there was a drive to push for P6 (the first reject-relevant position) until the bitter end. However, the author believes that concern is overstated: while drivers can afford to back off now, their instincts will always tell them to shoot for more and I am certain that even before the expansion of the points system, teams told their drivers to hold station when the position(s) ahead were unassailable and just hoped for something to go wrong in front of them.

Aug E.Another factor is the relative lack of anarchy in modern Formula One. As much as Liberty Media would like us to believe otherwise, in modern Formula One the “engineered insanity” is pretty much absent.  This statement may seem absurd, given as we just had a race where over one quarter of the field was eliminated on the opening lap, but it is still true. To take the most extreme example: the 1996 Monaco Grand Prix had 19 expected finishers. It is well-known how many cars actually saw the flag at the conclusion of the two-hour time limit. Nowadays, such races just could not happen: the poor wet weather tyres provided by Pirelli paired with the liberal use of red flags causing interruptions until the track is functionally dry would prevent such scenarios.

One point that cannot be ignored is the consistency at the top of the grid. For all this column talks about retirements and chaotic races, the most reliable way to lose one’s reject status is to be in a top car. From 2000 to 2019, Ferrari ran nine different drivers including the substitute stints of Luca Badoer and Giancarlo Fisichella. In the two decades before the turn of the millennium (1980-1999), Ferrari employed twice as many drivers. As driver lineups across the grid are subject to less fluctuation, drivers stay in the sport longer and the reduced grid movement results in less chances to move on to a better team to get those elusive top-six finishes.

AUg FHowever, the author feels the largest reason for why some members of the GPR Community feel like there is a need to change the definition of “reject” is that driver standards and car performance is higher across the board. As mentioned two paragraphs above, many races in the 80s and 90s that had a high number of expected finishers still saw a single-digit number of finishers because a noteworthy number drivers lost control of their vehicles: either because they ran out of talent or the car was such a terror to handle. Even the worst cars on the grid in recent years are superior in grip and downforce to classic championship-winning cars like the McLaren MP4/2; the less said about comparison to actual reject cars of the time, the better. Even Sir Lewis Hamilton and Max Verstappen would struggle to achieve anything worthwhile (if they even finished the race) with, say, an AGS JH21C.

Reject status, as GPR and the author are quick to mention, is not a statement of a driver’s ability or lack thereof. However, it feels like there are more rejects who were good or even great drivers by Formula One standards than ever. Even in the 90s, there were such cases (e.g. Eric van de Poele), but in the modern era, even upper-tier talent like Sébastien Bourdais, Pascal Wehrlein and Jean-Eric Vergne can end up with reject status if luck or the timing of opportunities do not go their way.

Even though it just seems (even to the author) just not as likely for a reject to get lucky and sneak into the top-five, the author thinks that the given numbers prove that the reject definition is still perfectly valid and would advise against any changes.

However, eventually there will be a new generation of fans and hopefully GP Rejects (or a potential successor site) and its spirit will exist for a long time. Perhaps a future generation of Formula One fans will judge our assessments as wrong and forge their own definitions. After all, just as sports is subject to perpetual evolution, so is sports fandom.

So is life, period.

And that’s a very beautiful thing.

Sources: GP Rejects, StatsF1

Image Sources: WilliamsF1