BRC 1966 - Gentleman Brobots Club [RACE 6 P&Q]

[strop]

To add to this: I'm OK with the random element and I actually agree with killrob. One other problem is that the rule set of BRC is constantly evolving as the simulation continues to be refined, so we don't have anything even approaching a decent sample space. We'd have to run like ten parallel seasons ceteris paribus to see where the trends settle. In the meantime, it is actually reasonable to say the dice favour some over others given the small sample space. For those who are familiar with stats, if you ran an analysis on the results, you wouldn't be able to yield a significant nor powerful conclusion that the system itself was biased against a particular competitor... Only that the current events as they are, have been.

For those of us who aren't coming out on top, that's obviously frustrating, especially for those of us who are used to exploiting the predictable nature of Automation. That's racing, though. It's just even harder to navigate when there are sixty cars and just ten races.

As for me, I don't think I even made a car that could potentially win anyway, and that's a separate matter.

[Leonardo9613]

Robert, maybe I'm the one not being clear here.

Randomness is good, up to a point and if it is backed up by some kind of pattern. Right now, the choices we make when designing the cars do not affect the outcome of the races at all. Safe drivable cars commit driver errors which get them out of races. Unsafe, undrivable cars do make errors too, yet they are less likely to yield high damage rates. I would not be pissed if my car made normal driver errors, or if once it did it didn't go out of the race, but that doesn't happen. My data to be analysed is small and highly targeted at mine and norman's car? Absolutely. However, it is beyond me how our cars keep getting fucked. Say you, martin or absurdist have a driver error, you shrug it off, maybe go back to the pits, get it fixed and go back to the race. Me and Norman haven't had the change to do this yet. My car crashed with conan's in 50% of the events that have passed. I am in a terrible 30-something on the standings, because my car was thrown out of the race. Conan is on the top 10, despite similarly fast cars in qualifying, and I should be able to one up him because of my different tyre strategy. Anyway, on the two collisions I have had with his car, I have gotten out of the race, whereas he kept on racing, without even enough damage to warrant an extra pitstop, and that is despite his car having a safety rating of 1/3 of mine.

Also, saying I can't claim anything because I don't have data is pretty much saying that no one can never make assumptions and use logic. I do not have more data, but what I have is enough to prove that I'm right, or at least to make my suspicions valid.

[Absurdist]

I think we should wait to the end of the season, plot some data into tables and graphs and see how strong or weak the relation is in terms of the whole field for say, drivability and errors or safety and damage taking weight into account perhaps. And then also plot crashes vs starting positions or qualifying times. Frankly, I have an erection just thinking about all the uses for this excel-lent data.

.... I am, not sorry. So as I was saying!

Although there are definitely outliers (You, Koolkei, Poor poor Mer_at), I think that we should wait to see concrete data. It is certainly possible that the system is not really fair, which is ok, maybe Der Bayer can look at changing it. If the relation between drivability and errors is not really strong enough, perhaps it should be more so, or safety or whatever. It is definitely worth looking at. Der Bayer is basically a god I think, or maybe just a demi-god, but I am sure that if we can show statistically that it is not balance, which is possible, then he will respond to that fairly.

I will be happy to go and plot this data later in the season, as it is really interesting to me, not only in terms of this, but also how to build my car for next season.

I do however definitely see where you are coming from.

[Killrob]

Robert, maybe I'm not the one being clear here.

No, you are clear, and wrong, that is what I am pointing out.

Also, saying I can't claim anything because I don't have data is pretty much saying that no one can never make assumptions and use logic. I do not have more data, but what I have is enough to prove that I'm right, or at least to make my suspicions valid.

Nice strawman. That is not what I said and it would be a stupid thing to say.

"I am fully aware that the statistical sampling isn't big enough to provide an accurate analysis, but it is the only we've got"

and

"I do not have more data, but what I have is enough to prove that I'm right"

tells me that arguing anything with you on this topic is futile. I shall rest my case.

[Leonardo9613]

Direct hits at a flawed logic, which had a disclaimer that made the point still be concise. I have not called you stupid, dishonest or anything else, calling people names and undermining the person is bad behaviour and not sportsmanship. Also pointing out argument strategies doesn't make you win, that in itself is a fallacy. Maybe arguing here really is pointless.

And I am not wrong. It simply is too random, the inputs made on the design phase are meant for nothing. I deliberately made my car slower so it wouldn't crash as much or suffer as badly in those crashes. All for nothing.

Edit. I actually wonder why you called it a straw man. It is hardly like you hadn't made it your main point on the last two posts and I am pretty certain that my synthesis captured what you were trying to say with the dice example.

[findRED19]

Sooo basically, what we need is a racing series with 100 races, and to replay that series 10 times to see how the data compares.

At this point, we really just need to roll with the sim for now, and save discussing changes to the random code for later.

Boogity boogity boogity boys, Let's go Racing!

[Der Bayer]

Alright, first of all I give you the calculations behind driver errors and damage from crashes and errors:

damage in crashes: dmg ~ random / durability * opponent_mass / own_mass

damage in errors: dmg ~ random / durability
error chance ~ TrackSportiness / TrackTameness * (rain + constant) --> more errors in rain

with durability being the average of safety and avg. reliability.

TrackSportiness and TrackDrivability are not calculated by me but by Automation itself and can be looked up in the Trim Lua file if you are interested.

So you see, the current effects are basically completely linear and then multiplied with a random value. So when comparing a lot of crashes, damage taken by a safe car will be lower than by an unsafe car.

I do agree that some of you have been very unlucky, especially those in the very tight field between 10th and 30th qualifying position. Seeing the pattern during 4 races is not possible and feeling cheated is of course frustrated. I can assure you that you are not. I have run 5 test races to make sure that the simulation does not crash anymore. Each time I was in the top 5, once even 3rd. In the real race I was 20th. That sucks, but that's racing. The calculations for this season will stay the same, just for fairness reasons. Maybe you had your slice of bad luck already and the other cars will be hit in the remaining races.

For the future I want to revise the error/damage/crash calculations. In a first draft yesterday I revised the damage taken in crashes: First an overall crash severity/energy is calculated (randomly). This energy is then distributed to the victims: 33% of the energy to each car + divide the remaining 33% between the two cars. Then we know how much energy the car has to absorb and safety or reliability can come into play: total taken damage is then energy * (100-safety)/100. That should make it more comparable with still having the chance to be unlucky and out of the race in a safe car. But if you are and the opponent has a very unsafe car, he will highly likely be out, too. Overall damage can be reduced a bit that there is the chance for repairs more often. It's just a basic idea and will be tested for next season.
Random driver errors will not happen out of the blue as much anymore, because I want to implement a pressure/stress system for the driver. So if a driver is under constant pressure by a driver behind him or his own undrivable car, error chances will increase over time. On long straights, during pit stops and when driving alone he will be able to rest a bit and reduce the chance again. Error damage will be handled similarly to crash damage then, although I am still thinking about including "getting stuck in the gravel" for cars with very low offroad ability or at least make them lose more time in big errors.

A lot can be done, a lot will change, and I think it will be more reasonable in the future. Just don't take the whole thing to seriously.

[Riso]

TrackSportiness and TrackDrivability are not calculated by me but by Automation itself and can be looked up in the Trim Lua file if you are interested.

Note: the lua calls trackdrivability tracktameness.

[KLinardo]

Do y'all watch or participate in motorsports IRL? My lord this was excessive bickering. Firstly, Der Bayer, as a first time competitor I'm still in awe with how awesome this simulator is. Secondly, to those with grievances, it's a race with 69 cars in the same leaderboard. Crazy things, unpredictable things, and things you cannot engineer away will happen.

I have a relatively heavy and safe car. I got in 2 crashes in a matter of ~10 laps in race 3. I survived the one with strop and had my damage repaired (I think strop wrecked out) and then got into another one with Conedoger and both of our days were over. I gripped about it, but I didn't blame the simulator. I was deep in a big pack with less stable cars and someone made an error at speed on the front stretch, so we wrecked hard.

Disappointment happens. Mistakes happen. Crashes happen. I am sure that the simulator calculations will be updated and adjusted while we venture forth into this new possibility with Automation. It's the closest we're ever going to get the game to incorporate racing. All of the kinks that happen in between, well, it's just a racin' deal.

Until then:

Boogity boogity boogity boys, Let's go Racing!

I'm with him.

[koolkei]

^ditto

since i only joined when you guys were on the last 1 or 2 races of the last season's BRC.
i wanted to join, and i got an even better BRC. i am plenty contented.

although. i do wonder, in the top 10 standings, it seems like my car is the only one that has almost half drive to sport ratio, and only 1 out of 2 that has 0 safety.
yet so far, i have only damaged my car once, i could count the number of driver errors on one hand (5 iirc). it's like i'm either blessed by the god of luck or there's a halo above my car.
did my car broke the calculation somehow? am i just really super lucky? i really doesn't feel like it fits to be in the top 5.

[strop]

As I said before, only time will tell. And out of that, I don't even think ten races is enough to give us any kind of 'even' luck distribution. Like it or loathe it, life isn't fair, and complaining about a simulation that adds unfairness to what was previously fair that you want it to be 'more fair' makes sense... how? And I'm saying this as somebody who has also wrecked out and been crashed into despite making sure my car was more drivable than it was sporty and being almost identical in stats to some of the cars that are in the top ten (note: extremely rough guess, if somebody wants to spend time setting me straight on this record, please do so, I literally don't have time to properly check which is why I'm not more disappointed than I am). Do I think that this is enough for me to make a 'better BRC car'? Honestly... no. That's my one gripe. I'm none the wiser on how to make 'the winning car', but that's partly the lie of the land.

But think about it this way, you wanted a real experience, and that includes the high highs of unexpectedly good fortune, and the low lows of having horrendous luck

[squidhead]

I didn't make this argument about you, your car, or your ability, but simply pointed out that you have a car with no safety.

So your basic complaint was "Cars with less safety than yours should be punished more". Which is what... 60 cars of the racing pack, IIRC. I don't get your beef with 0 safety. The ruleset had a "recommended safety", which was not mandatory. Naturally a lot of us chose to disregard those recommendations, you could too if you wanted to. Everybody who's got 0 safety gets massive damage, a lot of cars are one crash-dnf cases, so, that's "too soft" for you or something? I don't see your point. I see a lot of people are having this season waaaaaay worse than you and are not complaining. And either way i do not actually care enough to get into a deep discussion on how unfair my car is or any car that has lower safety than recommended. Rules are to be obeyed, recommendations are to be considered, that's it, all the cars on the track are legal, you should get over it.

[koolkei]

^except one, and i forgot whose it was, the breadvan? the 12.8l jeep? or the 73hp car(what was it's name again?)

[TheBobWiley]

Depending how the calculations change, my current draft for 1970's car is looking at almost 1/1 drivability/sportiness and over 33 safety, will be interesting to see how much better it does crash/error wise than people who I assume will be going 0 safety, high sportiness again. Looking forward to the rules for 1970's and playing with turbos

If you guys want to complain about people not having crashes, errors, failures when they should have, bo back to 1945 and tell me how absurd Absurdist was

[Der Bayer]

Race 5 P&Q: Thruxton

Here are the results of the practice and qualifying sessions in Thruxton. The new CSV files have been added to the OP.