“ are rfactor2 physics broken” video

Yea you right, physics is the last thing a simulator has to worry about especially when there are exploits in it.
Graphics, U.I and competition system is what a simulator is for after all.

 

It seems to me that you're twisting his words into an argument that's easy to destroy, in rhetoric we call that "the man of straw", basically you're taking his words back but you're changing their meaning into a ridiculous idea that's easy to contradict in order to give the impression that you're right.

He didn't say that physics wasn't important for a simulation, he just said that the situation doesn't require physics to be the top priority. The exploits you're talking about, if they really exist, which is not demonstrated, are so marginal that they don't justify stopping all development to correct them as a priority.

We all understood that according to you it was more than important, it was a "sine qua non", especially now that the competition is coming, but it seems to me that you are adding a lot to it and that it is not really obvious and that it doesn't seem to give a decisive advantage.

If we had a proof of the advantage of this "exploit", it would undoubtedly be important to take care of it but the video that you showed or the opinion of Ermin, who I remind you is only a sound engineer (with all due respect of course), in no case a famous driver or an expert in automobile physics, prove absolutely nothing.

 

If the incomes come from graphics, U.I and competition system, yes.
And then improve physics.

 

Especially that winning .7s between two laps, everyone can do it without even changing the setup, just take 2 or 3 turns better and you can win those 7 tenths. We only have two laps to compare, not an average of 20 laps...just 2 laps, no kidding...

These "demonstrations" impress those who know nothing about statistics or scientific demonstration: it's neither revealing nor reproducible, basically it's...nothing. Just a guy doing a better time on one of the laps he shows us, great, I'm really convinced...

 

Well after all these years if you still don't believe them then you are either living under a rock or simply ignorant.
Why don't you give them a try yourself, that will open your eyes.
Don't just watch videos i posted, you got net right ? search some alien footage and setup yourself.
It takes simple observation skills and common sense to see this vs real life difference.

 

Do these "aliens" show their setup, possible aids, telemetry, the consequences of these "exploits" in their direction, etc?

I'm probably ignorant, I don't know if you know the Dunning-Kruger effect (on which one should not be mistaken, it is not a means of judging others but of judging oneself), I'm probably in the valley of humility in terms of driving, perhaps a little further away but I also went through this phase "I know everything, I am the best and I judge the experts without difficulty". It's up to each one to gauge and judge himself.

I am also and above all a scientist by training, the observation you are talking about only leads me to hypotheses, which must be verified and validated...and there, sorry but you didn't do it, neither did I, so it remains what it is: a hypothesis, nothing more.

I tested this famous "exploit" by voluntarily inducing an understeer at my front end to verify: on the one hand I was not gaining lap time but on top of that, I found myself with unbearable vibrations in the steering wheel, you can probably use a lap to make a good time but clearly not over a whole race.

The phenomenon of the rear anti-roll bar is not an exploit in my opinion, it behaves as it should, in any case I don't see anything crazy in the behaviour. You explained to us in this subject that you didn't see why the car would hold better by releasing the anti-roll bars, we explained why you were wrong but it seems to me that you never admitted the explanations.

I repeat: I'm not saying that there is no possible exploit, I'm just saying that the elements brought are not proofs, at best they are doubts that need to be checked.

 

it is not hypothesis nor a doubt.
It seems you will not acknowledge it still. Doesn't matter anyways. we will see if we get a physics update or not.That will be perfect proof.

 

When do you expect physics update? And by it what do you mean? Update for rf2, or particularly some content? For rf2 I wouldn't count anything about physics in 2021, for particular content I'd only expect stuff that would make cars less challenging therefore "improved".

 

Of course they are, for the moment and until proven otherwise, only hypotheses and doubts. You have proved nothing at all.

And frankly, if there is some kind of evolution in physics, would that be proof that you were right in your opinion? Then the game will never get better without you ending up thinking you were right? Seriously?

Physics isn't perfect, we all know that, it can't be. The question was never whether it was, the question was whether there was some feat that gave you an unfair advantage over others, it's still not the same thing.

You systematically assume that you are right and you refute any idea that questions that idea, it is called cognitive dissonance. Just try to see things from another angle (and don't come and tell me you've already done it, I won't believe you for a second) and admit the idea that you might not be 100% right, then you'll have to prove your hypothesis, if you succeed then yes, you can say you're right but for the moment nothing has been proven. There is just "reasonable doubt".

 

How can we prove that a physics update solves an exploit if there are no telemetry data to prove the exploit?
The physics update could be made to improve something or to add any functionality.
I'm not telling that the exploit doesn't exists but i can't tell that exists because it's not well documented. A video isn't a document that has technical data to base on.

 

I think we're all going about this discussion the wrong way and it's going nowhere. I also think you're not a "troll" who just wants to criticize for the sake of criticizing.

So I suggest that we create a test protocol for each of the points you raised earlier in this topic. For each of these points we imagine a method to affirm or deny a possible problem and we submit our results to S397.

Some of the people here are professionals in the field, they know what to expect in a given situation, such as the two people who did the GTE BoP analysis document. I'm sure these people would help. I am also in contact with Aristotelis Vasilakos (developer for Kunos in charge of ACC physics who I find did a monstrous job), I am convinced that he would help without difficulty to understand and indicate what should be expected from the behavior of the vehicles...in short, people of good will, there are many of them and since your goal is to improve rFactor 2, let's improve it but let's be organized.

I don't know what everyone thinks about it but it seems to me that it is doable.

 

"not proven" even though whole leagues of people using it and aliens ...... . ok, i'm done. i don't need to give you guys proof, this is not my findings, it has been in this sim for years.

 

I really hope you are wrong on this point.

Also concerning offline (AI), even if I understood correctly that online is a priority for S397, I hope nonetheless that they will not neglect their thousands of offline players.

Everyone should get satisfaction at least on 1 point.

 

Again, I'm sorry if I'm not clear, English is not my mother tongue so I try as much as I can but I'm not as precise as in French.

I'm sorry if you take it wrong, that's not my goal, my goal is to explain that in order to make things move faster, we can do some of the work. You are observing a phenomenon that seems abnormal to you and therefore you draw a hypothesis from it. All I'm trying to explain is that it's not enough, you now have to work on confirming or invalidating this hypothesis...and that hasn't been done yet. If we can clearly establish that there is a problem because the result obtained does not correspond to the expected result, then we could present this problem to S397 and they can deal with it.

So far we have not proven anything. We have to check before we can move forward.

 

AI is more my area of expertise even if I don't know enough to create an AI powerful enough to drive a vehicle "as a human would" but I know enough to say that as long as what we call "AI" is actually just a script, we won't have any human behavior coming from it, not a chance.

NVidia is more or less opening the way for a new form of AI, a real one this time, by moving the heavy calculation of creation and allowing its cards to use this AI with a dedicated chip, which dramatically reduces the performance impact for the user. As an example, this is what DLSS does (even if the impact on performance remains more than noticeable). Sooner or later we will have video cards with a dedicated processor for this type of calculation, we could even add another dedicated card for that.

The idea would be to have an AI that would "see" what's in front of it and act accordingly, so we could imagine an AI that would modify its setup alone, that would discover a circuit alone and finally drive it alone, without anything being "scripted".

"Impossible" will say some people...and yet, we are really not far from it anymore, I wouldn't be surprised to see the first versions of such an AI arriving before 2025, or even earlier.

Anyway, the current AI in rFactor 2 is just a script following a serie of points with some variables to adjust inputs...nothing more.

 

Your idea is good but unfortunately i don't think S397 will do anything.

Here is my thought->

The unrealistic behaviour shown by driving and setup etc can be applied on all the official DLC modern physics cars.
This made me think that , it can't be possible that all the cars have same problem, yet they do. What are the possibilities of that the data in all the cars provided my manufacturer is wrong? i'm talking about all the major DLC car ,so the probability of wrong data in ALL CARS is highly unlikely.
I.E it is very unusual that all the cars have wrong/bad data and their exploits are similar.
What am i getting at ? it means that despite having correct data the core physics engine is not executing them properly i.e the problem is in core physics NOT individual car.

This is why S397 is either not bothering with it or taking this LONG time.

 

This or it is in the tire file that the problem lies. If they share the same tire model, maybe there is a flaw in it ?

How to test it ? I mean the test is obvious but what should we expect and what do we get ? This is the hardest part actually, we have to agree on the experiment and show results.

Then, to know where the problem is exactly is up to S397, we can't do this for them and I agree when you say we have no guaranty they'll do anything...but at least, we would have done our part.

 

We don't have official data to cross check and hence we can't prove we are correct.
We have our observation, common sense and experience to know if things are sort of realistic or not.

IMO in RF2 things are quite clear on what's wrong because IMO they are easy to cross check with our experience, IRL observation etc BUT not easy to prove on paper.

 

Not saying the physics are perfect or that there are no areas of improvement. I indeed just think it is not the only thing in their roadmap. I vaguely remember that S397 was working on an update to the Oreca tyre model.

There's some stupid things in the sim like where having lowest tyre pressures is always fastest. That makes tyre pressures less interesting strategically than in ACC.

I don't know about detaching the anti-roll bar - whether it is the same in real-life - but I've never heard about it from my iRacing teammates. But indeed, it is also something I do on many cars.

Just saying Ermin's video is quite bad because he only does two laps while longer stints should also be taken into account. For instance, imo the Ferrari is super fast in the first two laps or so and then it abruptly drops in pace and stays there. I don't know about the vette. Just seemed to me and he said this - that he scrubbed the tyres less and was on the throttle later, saying what he calls "rotating the car on throttle". That is also a thing in real-life. To me he just underdrove the car.

Not saying that everything he talks about is wrong. Just saying it isn't new. But in his testing there's too many variables to say anything about lap times. The video is superficial. I'd rather have something like what the guys did with the GTE fuel. That's actually useful.

But again, I wouldn't go as far as to call it "exploits" or "hacks" or saying all of rF2 physics are broken. That's just unfair which brings me to my earlier point - which is that rF2 always had the reputation of having state-of-the-art physics. Doesn't mean that ACC isn't becoming a contender, but again indeed also consider the scope of rF2 with all the different cars and the sim being quite old.

To me it just seems like S397 is putting down pieces of a puzzle needed to generate more income. Competition system, new UI (which is more about flexibility in functionality) and eSports. They explained that narrative perfectly well in the last few months in the devblogs. I think that's what people misunderstand when they act like S397 doesn't care. They do. It's probably in the pipeline somewhere, but not something they are putting all their resources on. Why would they?

 

As everybody share the same physics, it is at least fair. Just play the game the way to be fastest. Like in reality btw. .
There are hundreds taking part in competitions, why should s397 start changing things based on crying of one or 2 users? Shall they allways start changing models, which takes much time and without knowing overall outcome whenever a user starts crying out loud and long enough that something is not real, with no real experience, no competitive pace and refusing to spend time himself to bring solid proofs, except for spending time on crying?