Is GT getting easier? - A small empirical analysis

[zocker_16]

@eran0004 : What you say is right again in several points, but there´s also something i see different.

Of course the skill of an individual is going into my datas but it´s effect is quiet small.
You see on the datas above there are between 10-30 people who are able to do a lap within a 0,5% Gap to the best time.
So if none of this 30 people enters one of the tests i´m looking at, the gap would decrease about 0,5%.

My basic assumption is: Theres a physical perfect lap which can´t be improved and can´t be achieved by a human driver.
A good driver is somebody who gets close to this perfect lap. The deviation is also like a square route. So the closer you come to this perfect lap the less players are able to achieve the time. Also improvements get smaller and the amount of time you have to spent rises exponential.
What i say is: It doesn't matter who´s the driver on P1 because there are several players who can do the same time +- a few tenths of a second. Like i said before this is an insignificant error.
This is maybe the main difference to what you say.
My assumption is that the average players time is based on luck, experience etc. so it´s variation is big.
The level of the top players is constant.

So on your example with the maths test. I have 3 students which are the best at maths.
All of the achieve 97-99 of 100 Points on a previous test (average of all students is 50% for example).
Now each of them gets a new test and they score 60, 75 and 96 points. To see which test is the hardest just look at their new results. Test Nr. 1 is hard and test Nr. 3 really easy.
60,75 and 96 points are the barrier and even the teacher would´t score more than 5 points over this.
It´s possible for every student to achieve 60,75,96 if they learn enough.
You don´t need to look on all other students to know this.
Because you know the deviation you can also estimate the average score of the other students. For example 25-35 points in the first, 32-42 in the second and 43-53 in the 3rd.
Now do the same with a bigger group of people so you have 1 student who scores 99 points on the standard test, 5 with 98 points or better , 25 with >97, 100 with >96...
So i do 2 new tests and the best students score 70 and 80 points, i know theres at least one of the top 100 in this test who scored >96 points on the standard test. I say it´s really significant that test 1 is harder than test 2.

Gold time on GT is like the amount of points you have to get to pass the test.
Example: Test 1
Top Score: 80 points
Score to pass: 50 points

Test 2
Top Score: 60 points
Score to pass: 25 points

So even if the top score in the second test is lower there will be more students who pass it because the relative gap is bigger.

Because it´s possible to restart in the game the factor of luck is small. Also the learning effect doesn´t matter because of the barrier and it´s possible to compare different "generations" of drivers. Each generation has their 96-99 point students.

 

[dahcmai]

I have a quick question since we're on the subject.

Is there a difference in the licenses when using a wheel as compared to using a controller with the Dpad and buttons?


The reason I ask is because I started with GT1 and just got used to the buttons and Dpad despite the analog joysticks being around. Even through GT3, I had all golds on the licenses. GT4 I admit I didn't play all the way through. I just didn't like that one as much. Then 5 came out and I got all golds again still using the buttons.

Now with 6, I had the fortune to own a nice G23 wheel and I find I'm having to put it away since I keep having trouble with them. Once I pick the controller up again, I can do most again without a problem. Some require the wheel, such as the cone slaloms and such, but most like the simple ones in A and B that require you just to make the apex of two curves I can't seem to do with the wheel. guess it's just me being so used to that pad, but that just seems silly compared to the control you get from the wheels.

 

[eran0004]

My basic assumption is: Theres a physical perfect lap which can´t be improved and can´t be achieved by a human driver.
A good driver is somebody who gets close to this perfect lap. The deviation is also like a square route. So the closer you come to this perfect lap the less players are able to achieve the time. Also improvements get smaller and the amount of time you have to spent rises exponential.
What i say is: It doesn't matter who´s the driver on P1 because there are several players who can do the same time +- a few tenths of a second. Like i said before this is an insignificant error.
This is maybe the main difference to what you say.
My assumption is that the average players time is based on luck, experience etc. so it´s variation is big.
The level of the top players is constant.

It depends on what you mean by deviation and variation. If we look at deviation as the difference from one position to the next, we find that the deviation is actually greater the closer to the top you get, while times further back are closer to each other. As shown by the data of the GT Academy UK online ranking:

Collecting the top only - essentially a spike - can create a really distorted picture of the whole, because the top time doesn't say anything about how the overall population did in the test. We know as a certainty that the rest of our sample performed worse, but how much worse? Is it a very steep curve down to the average time or is it a very shallow curve?

So it's really hard to say anything about the difficulty of the test, other than that maybe (if we assume that the skill level is near to constant in the top of every test, that the spike is always just about as big) things have changed for the very top players.

But strictly speaking, since the population has changed as well (some has left, some has been added but those who has stayed has also grown older with time) we can't really tell what has happened. In order to spot any changes in one variable, we need to make sure that the other variables remain constant. If we change two variables and get a different outcome, we don't know which variable caused it, or if both caused it, and in that case to what degree they influenced the different outcome. The game is one variable, the population is one variable. The equipment is one variable. The incentives is one variable (has it become more attractive or less attractive to rank high on these lists?). All needs to stay the same, so that the variable we want to test is the only one we change.

As an empirical analysis, it's hard to actually come to a good conclusion based on these data. It is interesting what it shows, but does it actually answer the question it is meant for?

Add a "... for top ranking players" to the research question and I'd agree that the possible errors are not too significant and that it can indeed show a tendency, although it's hard to say more precicely what is going on and how big the differences are. When it comes to determining wether or not the game has become easier [period], I still think that it doesn't really provide an answer, but rather that it creates more questions.

 

[MadMax]

@zocker_16 , I managed to dig some old GT1 & GT2 licence times from Racing-Line. (using wayback machine)

And we also have some fresher GT1 times listed here and some more at cyberscore.

 

[zocker_16]

Thank you for this graph eran0004 !
It´s something i´ve also noticed while looking on many rankings during the last years.
This looks disputed to the theory of small variation between top players.
But i think i found an explanation for this.

There are only a few people who can reach nearly perfect times, so density of drivers is low.
Density will rise with a higher lap time and decline after the average.
Thats why 1/10 can improve your position by a hundred places in many rankings.

Add a "... for top ranking players"

Yes this would be the right question but also less controversial and boring for a discussion

What i tried with my definition and the basic assumptions i made was exactly to eliminate several of these variables as good as possible. I also don´t see those numbers as perfect result. It´s the direction they show like you have said.
Change in population and high or less attractive top position was something already mentioned in my first post. It´s hard to estimate this effect. I think it´s not more than max. 1% or 0,6 seconds a minute, because theres also no ranking with P1 to P2 gap higher than 0,6 sec a minute on a spike.
But what you say: We create a lot of questions instead of answering the first one

Right now i´m working on Part 2 of my research. The question is: What is an appropriate gap for a gold trophy in time trails?
Based on GT5 online rankings of mygranturismo.com , because it´s the only database i know, which shows all times (over 100000 on some rankings).
So how many players are able to achieve a certain gap?

@MadMax : Thx for this rankings i´ll check if these are useful for a closer look

 

[sumbrownkid]

Maybe you're just getting better

I wonder if that's the case really. Because most games today add difficulty by either making your character weak, the enemy stronger to perhaps OP levels of strength, or a mix of both, so difficulty is achieved by nerfing your character. In racing games, the AI's were usually supplemented by rubber banding in arcade racers to give the illusion of difficulty.