Quote from Gunn :

The fact that real stewards make errors of judgement sometimes only strengthens my opinion that an automated system would be inadequate.

That fact has very little to do with what this topic is about. IRL we dont get to pit a driver for misbehaviour , in LFS we do and can and need a well constructed solid set of rules and sub-rules to carry out that decision when its called upon to do so.

E: jesus h, if I only knew how to write scripts n' stuff I might know wtf im goin on about!

Quote from Gunn :

No I don't think it's an overstatement. And I do believe that any automatic system would need to work perfectly to be of any real use. The fact that real stewards make errors of judgement sometimes only strengthens my opinion that an automated system would be inadequate.

OK, thanks for the reply.

I'm not trying to pick a fight here, it's an honest question: do you have a background in computing / AI / cognitive science? I seem to remember you're a psychologist, is that correct? I ask because I'm surprised by the strength of your conviction on this point and I want to know how to interpret it.

Also, a hypothetical: let's say that an appropriately trained and incredibly altruistic person was willing to watch every LFS race on your favourite server and to hand out penalties, live, for any infractions such as unsafe overtaking. Let's say this person did the job as well as anyone could hope to, and that they were completely unbiased. Would you want them watching your races or not? In other words, is some of your antipathy towards an automated system because you wouldn't particularly want a human system either?

Cheers.

Quote from jtr99 :

OK, thanks for the reply.

I'm not trying to pick a fight here, it's an honest question: do you have a background in computing / AI / cognitive science? I seem to remember you're a psychologist, is that correct? I ask because I'm surprised by the strength of your conviction on this point and I want to know how to interpret it.

Also, a hypothetical: let's say that an appropriately trained and incredibly altruistic person was willing to watch every LFS race on your favourite server and to hand out penalties, live, for any infractions such as unsafe overtaking. Let's say this person did the job as well as anyone could hope to, and that they were completely unbiased. Would you want them watching your races or not? In other words, is some of your antipathy towards an automated system because you wouldn't particularly want a human system either?

Cheers.

I am already a part of such a human system and yes, a human system can be very successful. I don't know why you are taking it all so seriously, I'm not. Logically an automated system can not work fairly. I've already arrived at that conclusion after much thought. That's how you can interpret the strength of my conviction. I'm not passionate about it at all, on the contrary, I'm not going to waste my time on a lost cause such as this. It's as simple as that really, sorry that you took my opinion so much to heart, I can't help you with that though.

The 'human' system is obviously going to be a better judge of what is happening and the causes of an accident or whatever.

The problem is that requires watching replays and dealing with it after the event, meaning if someone is causing problems for drivers then you need to be able to successfully kick/ban them and then save the replay to send to the relevant governing body.
But if half the track doesn't see an event or votes to kick the wrong person, the guilty go unpunished and the victims get to continue being victimised.

An automated system is the only real-time way of trying to sort people out, but if such a system was to be added it would need extensive testing and a lot of variables to check against.

Let's just hope that the final damage model sorts them all out instead.

Quote from Gunn :

I am already a part of such a human system and yes, a human system can be very successful. I don't know why you are taking it all so seriously, I'm not. Logically an automated system can not work fairly. I've already arrived at that conclusion after much thought. That's how you can interpret the strength of my conviction. I'm not passionate about it at all, on the contrary, I'm not going to waste my time on a lost cause such as this. It's as simple as that really, sorry that you took my opinion so much to heart, I can't help you with that though.

No big deal, I'm happy to agree to disagree on this. I would point out that you surely wouldn't have put "much thought" into it if you weren't in the least interested in the issue. I would also suggest that under many logical world views the human brain is but another automated system so I think there is room to disagree with your core premise. But you have obviously drawn a firm conclusion and more power to you.

Quote from Dajmin :

Let's just hope that the final damage model sorts them all out instead.

That might do it; would be great if that's how it works out.

Quote from Dajmin :

The 'human' system is obviously going to be a better judge of what is happening and the causes of an accident or whatever.

The problem is that requires watching replays and dealing with it after the event, meaning if someone is causing problems for drivers then you need to be able to successfully kick/ban them and then save the replay to send to the relevant governing body.
But if half the track doesn't see an event or votes to kick the wrong person, the guilty go unpunished and the victims get to continue being victimised.

An automated system is the only real-time way of trying to sort people out, but if such a system was to be added it would need extensive testing and a lot of variables to check against.

Let's just hope that the final damage model sorts them all out instead.

An automated system can not differentiate between an offender a non-offender in every case. That would make it unfair. The other huge downside is exploitation of the system. Either way, innocent drivers get penalised by the software, it is unavoidable. Even in real life penalties are not given by default just because a rule was broken or contact was made. Software can not make the tough decisions nor hand out penalties even-handedly. Software can not scrutinise. As for drivers voting to kick people off, such a feature has no place in organised online racing.
An automated system is a system that is not capable of scrutinising incidents nor applying penalties fairly, nor even identifying true offenders and victims. An automated system can be exploited by disruptive people, the very people that are supposed to be "sorted out".

Not sure if this is known, but in LFS you "give out" a yellow flag when you leave the pits mid-race.........

An automated system will never work, too many variables involved.

surely when the damage system is done heavy contact will be penalised simply because car is damged?

Quote from kieran20 :

surely when the damage system is done heavy contact will be penalised simply because car is damged?

What about the tagging and blocking that goes on?

Quote from CSU1 :

What about the tagging and blocking that goes on?

well if its deliberate those should be banned. if its an accident, sorry is said and we move on.

I fully agree with GUNN on this one. I do not thing the systems, logic and data systems are in place within LFS to even come close to building a system that is accurate enough to be seen as a fair and equitable system. I am not sure but the numbers I see form looking at $myinfo make me believe that the number of "Yellow Flags" is not the number of times you have found yourself in a position to send a yellow flag for the nuumber of cars that have recieved the yellow flag packet because of you. In a 20 car race for a single incident this could be as high as 19 "Yellow Flag" tags for a single spin or pit out.

This is an area where LFS needs many improvemts before we can get a reliable penalty system for things like contact and full flag rules.

The yellow flag indication is the number of yellow flags your car has caused even if no-one else is there - as per INSIM specification, at a maximum of 1 per sector - as per STCC-SRA system specification.

Before I do anything regarding implementing modifications to the licence field I think I will make some kind of $reportyellow command that turns on reporting of when a yellow is registered and then people can report any glitches, and see for themselves just when they are throwing yellows out.

I tested the system quite extensively, and whilst it does have limitations, it does seem to work - but i'm not prepared to wager money on it and that isn't only because I dont have any . The data the system shows is only useful when looking at the extremes. Some people have lots and lots of laps with only a handful of yellows - some people have more yellows than laps, there is a third category of inbetweens.

On the whole there is three distinct groupings when plotted, which suggests it is driving style / racing methodology that is influencing the result.

Quote from Becky Rose :

On the whole there is three distinct groupings when plotted, which suggests it is driving style / racing methodology that is influencing the result.

This thread is worthless without graphs.

Gimpster, you and Gunn seem to feel very strongly about your side of this issue. What would you say to a system that banned drivers who were in Becky's group of high yellows:laps ratios, the people who seem to be taking the piss from a racing point of view?

Quote from jtr99 :

Gimpster, you and Gunn seem to feel very strongly about your side of this issue. What would you say to a system that banned drivers who were in Becky's group of high yellows:laps ratios, the people who seem to be taking the piss from a racing point of view?

It's not my side of the issue, it's the logical outcome of thinking about all of the possibilities. Becky's proposed system isn't about applying penalties as incidents occur and therefore does not fall inside the scope of this discussion.

Quote from Gunn :

It's not my side of the issue, it's the logical outcome of thinking about all of the possibilities.

I'm sorry, Gunn, but repeatedly insisting that you've done all the hard thinking and have reached a firm conclusion that constructing such a program is impossible (and note that you've said impossible, not just "really hard") does not an argument make.

Quote :

Becky's proposed system isn't about applying penalties as incidents occur and therefore does not fall inside the scope of this discussion.

No, fair enough. But it points to the possibility of thinking laterally and coming up with automated systems for policing driver behaviour
that are perhaps more accessible and more reliable than we might have imagined.

Quote from jtr99 :

I'm sorry, Gunn, but repeatedly insisting that you've done all the hard thinking and have reached a firm conclusion that constructing such a program is impossible (and note that you've said impossible, not just "really hard") does not an argument make.

That's a relief, I'm not trying to argue.

I still feel it is not an accurate reflection of a racers ability. I have a question for you though to try and put this in to peerspective.

How many times have you been the victom of an incident that threw a yellow flag?

Got me that number is high. I would have to guess that in the last year 3/4 of the yellow flags I have thrown have been because I was knocked out of a race by some one elses mistake. Whether that was getting punted off track when I was braking early to avoid the car in front of me, or being pushed off track by a barge passer, or throwing myself off the track to avoid contact with ongoing incident.

Now do these incidents mean I am a bad driver? No yet that reppresent a majority of the data that Becky's system is using. If she then takes the step to ban people with a higher yellow flag count and I have had a string of bad luck and been the victom in a lot of recent incidents then I coudl be unjustly punished with no way to protest as the data is not in context.

One part of good driving is to know situations that could end up wrong and avoid them, there is many situations where you can be a victim in yellow flag incident, but making another choices there would never been that incident.

For example once while braking I saw from mirror car coming too fast and I stop braking and hit throttle, messed my corner entry completely and got partly to grass, but could carry speed and avoid accident.

Good driving is also to learn know racers around your car so you can estimate probability of them making error and need for evasive actions.

So I would say even driver is victim, amount of yellows do show something about how good driver is.

It is very very small amount of yellows where system would been done unfair marking to stats for me, maybe one or two can be called so, but it is really so low amount, that I would not say that it would matter anything and defineatly would not be enough to get wrong side of ban barrier if there ever will be one (I haven't heard from such thing?).