Our ReVolt Pub

Hello dear people,

seeing that our buddy Zach is willing to make changes to improve Re-Volt Zone, please give your opinion about which grading scale you prefer the most.

the Classic 0-10 that was in RVZT?
or the 5 stars-based system introduced recently?

Don't forget to check the other Poll here.

Having a 5 star scale encourages me to vote. It's easier to overcome and requires less unnecessary thinking. If I understand right the actual average is calculated with decimals anyway. If I wanted to express what I think of the track more precisely than that, grades would usually not suffice anyway no matter how big a scale. Thank goodness we have comments!

urnemanden @ Jan 27 2014, 11:33 PM wrote: Having a 5 star scale encourages me to vote. It's easier to overcome and requires less unnecessary thinking.

+1 for this. A five-level ranking is actually enough for most of this kind of rating. I'm quite convinced it's even more accurate at the end, since you can define words that guide all users toward a similar grading method, like eg.

1 star: to avoid, bad work, too much default
2 stars: quite bad, some good points maybe, but not really interesting
3 stars: average, not amazing, but a really descent job already
4 stars: good job, already better than the average
5 stars: excellent, ie. a master piece

More levels for rating makes no real sense to me.
Of course you can (should!) use half-stars to display the average score.

Displaying the number of votes next to the starts would also shows the "meaningfulness " of the rating. Like:

I am surprised to see the different opinions we can have or what different things we expect when we users are browsing the same site. I really didn't expect to hear things like Stars being more accurate or encouraging you to vote in comparison to the classic numbers.

When I want to calificate something as an user, I would choose a number rank. That is more natural for me as an user. Stars definitely don't encourage me to vote or give a mark at all. I don't find it practical.

It's interesting to see these differences sometimes which I ignored. In RVL everyone chose Numbers, and now I see you 2 saying this which makes me think.

Phantom @ Jan 28 2014, 02:19 AM wrote: I really didn't expect to hear things like Stars being more accurate

The reason I think so it that it leads you to define some specific criterion per grade. And you can define (and remember) 5 different sets of criterion, but you can't with 10.

I use a five stars system to rate my MP3, and I'm perfectly fine with it.
A french popular movies website use this for rating and this is both user friendly and efficient (they display both stars and the rating value from 0 to 5 with 1 decimal).

Numbers as I can score things more specifically, and it allows tracks to be properly ordered by grade.

nero @ Jan 28 2014, 09:56 AM wrote:Numbers as I can score things more specifically

Though I bet you wouldn't give the exact same score if you had to do one month later without remembering the previous one.

and it allows tracks to be properly ordered by grade

What stars allow doing exactly the same way.

A point that some seem to forget is that their own single rate is not significant. What will be relevant is the average of all users rate. And the fact that you were able to rate the graphics 7.352 / 10 instead of 4 stars over 5 will make such a small difference on the final rating that it is pretty useless.

Anyhow, stars or number, it is exactly the same thing internally. It's just about using a more convenient interface for most users. I still wonder why we have to go to such discussion over such a simple thing.

My thoughts exactly jig. ;-). As you said, its the average, and yes, partial stars are used to display the average grade. Fewer options lead to more consistent grades between different times and users.

I've noticed, that although a 5 rating was the majority prefered choice, no one (well, very few) actually rate's anything.

I thought the point of rating's is to get an idea of the car/track before download.

so from a statistical point of view, afaik it is in principle illegal to do an arithmetic mean on ordinal variables / fixed scale things (though used a lot in practice).

another option is to pick the mode (which score was given the most). that would be less prone to spiky rating distributions (fx 80% voted 2/5 but one guy voted 5/5).

jigebren @ 28 Jan 2014, 02:14 PM wrote: And the fact that you were able to rate the graphics 7.352 / 10

In RVZT, that's actually 7.4

What I miss from the old system, is not only the fact that that is what I was accustomed to, it's the fact that nobody really rates anything. Especially cars.

Don't ask people to rate anything if you've deleted 10 years of ratings and reviews in a website update. People won't spend time in re-reviewing and re-rating old things again if they feel that it could be removed from existance again.

I know that I'm entering this discussion a little late but since I just saw it pop up again:

Imo number/star scales are the wrong way to go here, there are just too many factors for cars that you can't rate objectively (most specifically handling and feel).

But the thing that bothers me more is the "Top Graded" listing. Since people tend to give a full rating for cars that they like rather than being a little more critical about it, a lot of cars end up with the same rating (that and the fact that there is actually not much space for the rating to vary).

However this ends with a lot of cars disappearing in the back even though they were top rated. I don't know what the other ordering factors are for RVZ after the grading but I think it would be very fair/useful to also store/display the number of people that rated this car and list these cars higher than some new entries where only a few people voted 5 stars.

And a "Top Downloaded" listing would also be very useful (of course the whole thing goes for tracks as well).

Kenny @ 27 Feb 2015, 10:23 AM wrote:Imo number/star scales are the wrong way to go here, there are just too many factors for cars that you can't rate objectively (most specifically handling and feel).

No star, no number... but what do you suggest then??

At some point something has to be chosen, and the simpler is often the better. That why 5 stars is very well appropriate for this case (one for the modelling job, one for the handling, I don't remember if the overall one is computed for the 2 others of it it has to be entered as well in the current RVZ).
And if it had to be changed for something else my next choice would be a 3 stars rating (I don't like - neutral - I do like ).

Kenny wrote:Since people tend to give a full rating for cars that they like rather than being a little more critical about it, a lot of cars end up with the same rating

True. Maybe a more elaborated code could compensate for that. For example an user that gives only 1 or 5 stars should have less* weight than a user that use all rates. I bet this would not be that complex... But I doubt Zach has still time to update RVZ, hasn't he?

EDIT: (*) I had written "more weight", I obviously meant "less weight", thanks }!{enR for pointing that out.

Kenny @ 27 Feb 2015, 02:23 PM wrote:Imo number/star scales are the wrong way to go here, there are just too many factors for cars that you can't rate objectively (most specifically handling and feel).

But the thing that bothers me more is the "Top Graded" listing. Since people tend to give a full rating for cars that they like rather than being a little more critical about it, a lot of cars end up with the same rating (that and the fact that there is actually not much space for the rating to vary).

jigebren @ 27 Feb 2015, 07:31 PM wrote:

Kenny @ 27 Feb 2015, 10:23 AM wrote:Imo number/star scales are the wrong way to go here, there are just too many factors for cars that you can't rate objectively (most specifically handling and feel).

No star, no number... but what do you suggest then??

Well, actually this rating by KW is the best one I saw ever. It would be nice to implement it at RVZ (maybe also only allow to submit the rates that has been calculated by this form), same for the cars.

Kenny @ 27 Feb 2015, 02:23 PM wrote:However this ends with a lot of cars disappearing in the back even though they were top rated. I don't know what the other ordering factors are for RVZ after the grading but I think it would be very fair/useful to also store/display the number of people that rated this car and list these cars higher than some new entries where only a few people voted 5 stars.

And a "Top Downloaded" listing would also be very useful (of course the whole thing goes for tracks as well).

Agree with that.

jigebren @ 27 Feb 2015, 07:31 PM wrote:

Kenny wrote:Since people tend to give a full rating for cars that they like rather than being a little more critical about it, a lot of cars end up with the same rating

True. Maybe a more elaborated code could compensate for that. For example an user that gives only 1 or 5 stars should have more weight than a user that use all rate.

Well, actually in most of the cases "an user that gives only 1 or 5 stars" is a person, who doesn't think much, just clicks the buttons, while "a user that use all rate", in my opinion, did a proper testing.

}!{enR @ 27 Feb 2015, 04:12 PM wrote: Well, actually this rating by KW is the best one I saw ever. It would be nice to implement it at RVZ (maybe also only allow to submit the rates that has been calculated by this form), same for the cars.

Yeah, I had something like that in mind as well. Of course for cars it would be points like:
- how well is it balanced in comparison to stock cars regarding handling (its underperforming/similar to/outperforming the original cars)
- how well is it balanced in comparison to stock cars regarding speed (its underperforming/similar to/outperforming the original cars)
- how easy is it to lose control in turns (not only spinning out but also flipping over)
- how well does the AI deal with the car (always behind, can keep up, always in front)
- are there minor glitches with the car like hull problems (too big/small hulls), vertex problems (missing vertices or similar), other problems (like electro/battery effect not being displayed properly or bein unable to fire weapons (keyword bikes))
etc...

Of course if these things together would sum up to another rating then the necessary information is lost again so it would be useful to be able to see details on how many votes each option has (where the overall score could be viewed as a short summary).
I know that all of this is probably too much to do and rather unlikely to happen but since jigebren asked, I replied

But I doubt Zach has still time to update RVZ, hasn't he?

Well would he be willing to give other people access/control over RVZ? (not talking about me but generally other people)
I think regarding custom content its actually the most important place nowadays and as such it would be in the community's best interest to improve it in every possible way.
But thats just not going to happen if all relies on just one person who pops up once in a while (not saying anything against Zach but its a fact that many people would like to see some kind of imrovement on the site and nothing is being done about it).

Kenny @ 27 Feb 2015, 10:59 PM wrote:I know that all of this is probably too much to do and rather unlikely to happen

I don't see a really big problem with it. (well, maybe I am blind... ) The half of this system is already implemented, so I think people just should discuss the key points that should be rated and add a bunch of columns into the database...

Just a pure suggestion;
The car/track automatically receives a 2 star rating (across the board) when an individual (IP logged) first downloads, then if they wish they can come back and adjust their rating depending on their personal opinion.
In conjunction, an item's overall rating receives a percentage boost/bonus depending on number of individual downloads (global popularity).
Because the IP is logged any subsequent downloads by that idividual would not effect the overall rating or your own rating.

So at the very least everything receives a 2 star rating plus popularity bonus without anyone actually individually rating.

Unfortunately it will end up in most of the tracks been rated as bad, because most of the "individuals" are not coming back to vote. And also it will automatically lower a total rate of the track, because there will be more people who just downloaded the track, rather then those who actually voted. So it probably will be better to have an unrated tracks/cars by default...

}!{enR @ 28 Feb 2015, 06:53 AM wrote: Unfortunately it will end up in most of the tracks been rated as bad, because most of the "individuals" are not coming back to vote. And also it will automatically lower a total rate of the track, because there will be more people who just downloaded the track, rather then those who actually voted. So it probably will be better to have an unrated tracks/cars by default...

Did you understand or even fully read what I actually said ?

It would fit me well I usually have to download many times the same tracks because of reinstalling windows, yet it would rate it just the first time I download, that way our assets don't get underestimated

@MM: Oh, I re-read what you wrote and I got your idea now! It sounds quite good.