Repo rating system: 5 stars or binary ratings?


_wb_

Microbe
Staff member
Joined
Apr 5, 2012
Messages
5,390
Age
42
Location
Brussels, Belgium
The 5-star rating system of the repo is not a very good one imo, see e.g.


http://davidcelis.com/blog/2012/02/01/why-i-hate-five-star-ratings/


Perhaps it's better to change it to a binary system like YouTube, or something ternary like this:


-1 : don't like, recommend not to try this


+1: like, recommend


+2: like very much, strongly recommend


where instead of taking the average of all ratings, the counts of each category are shown, e.g. with a red bar the size of the -1 votes, a dark green bar the size of the +1 votes and a light green bar the size of the +2 votes (or just let the +2 votes counts as two +1 votes). The total size of the bars could reflect the total number of votes (but maybe not linearly).
 
-1 : don't like, recommend not to try this


+1: like, recommend


+2: like very much, strongly recommend

Ternary systems such as the one you recommend are not good because they are biased towards positive scores :)


A good rating system should be unbiased. That goes without saying.
 
Last edited by a moderator:
Why unbiased? The software we hope to get on the repo shouldn't be neutral on average, it should be good on average :)


If something really stinks, it will be clear enough when it gets only dislikes and no likes. No need to introduce "strongly dislike" imo.
 
I like _wb_'s system. I could imagine it for the repo, too.


Instead of the stars you could have a youtube like bar with the colors red, green and gold for don't like, like and like very much.


In that case a absolut rating (like 4.7 stars) woudln't be necessary (to shown), as Youtube don't print the difference between the up and down votes, too.


You could just use this internal difference (or another formula based in this ternarity) for the sorting. ;-)


@Znirk: Had to think of, too. If I vote with stars, I vote 1 for things I don't like, 4 if I like and 5 for thing I like very much.
 
Last edited by a moderator:
If you're interested in money, you can always sell your software. The Pandora isn't the best platform to do that anyway, and this will not change unless it has several orders of magnitude more users. I suspect most of the Pandora devs do what they do because they enjoy it, as a hobby.

If your interested in hardware you can always buy your hardware with your price money. I don't see your point there.
 
Yes, with money you can of course buy hardware, but psychologically it is not the same. I consider coding stuff as a hobby; getting paid for it doesn't feel right, but getting some prize as a token of appreciation is something else.


Also, if the hardware is stuff that the OPT sells with some profit margin, they can give it away cheaper than having to give the equivalent amount of money; the difference would be their contribution to the prize, while the community donations go to the production cost of the prize. Maybe they can even avoid having to pay taxes on the prize hardware (or at least pay less taxes).
 
Why unbiased? The software we hope to get on the repo shouldn't be neutral on average, it should be good on average :)

Because a scoring system has no meaning if you bias it. Basically you cannot rely on it because it gives a disproportionate visibility to the positive votes. Then let's rate all application 5 stars and not put any negative value anymore, while we are at it.


Or if you assume all applications are good by nature just because someone is capable of compiling a pnd, then let's remove all rating systems :)


A rating is a form of feedback. Feedback can be negative as well as positive. What is the problem with accepting very negative feedback if you accept very positive feedback in the first place?
 
You can make it -2, -1, +1, +2 if you want, but I doubt there will be much use of -1 then, because people don't tend to think something mildly sucks: either it sucks (call it -1 or -2, whatever), or it doesn't. Distinguishing between "nice" and "very nice" is useful imo, but distinguishing between "mildly sucks" and "sucks bigtime" is not so useful.


Anyway, with a (-2, ) -1, +1, +2 system you can display the votes as a bar like Ziz suggests, and for sorting/ranking purposes you can just add all the numbers together; this puts stuff with more votes (as long as they're positive) higher in the ranking, which makes sense. With 1-5 stars you cannot do that because even 1 star ratings are "positive".
 
With 1-5 stars you cannot do that because even 1 star ratings are "positive".

Have you ever used Amazon ? They have 1-5 stars system as well, and it works perfectly for me. I clearly understand that 1 means "crap".

I was talking about adding ratings together to get an overall picture. With a 1-5 system you have to average the ratings, not add them, to get something meaningful. And then you get the problem of how to sort, because 4.9 star average with 500 votes is clearly better than 5 star average with 2 votes, so just the average is not enough.
 
With 1-5 stars you cannot do that because even 1 star ratings are "positive".

Have you ever used Amazon ? They have 1-5 stars system as well, and it works perfectly for me. I clearly understand that 1 means "crap".
So does 2 and 3 stars. I only buy stuff with 4 or more stars.


As Znirk posted:
star_ratings.png



It is MEANT as "1 star is crap", but it is not used in the meant way. ;-)
 
Last edited by a moderator:
The ´rating system´ could be spared for judges maybe?


It could be bad for new programmers to get rated low on a first attempt, even if his soft is simple. Such won´t encourage much I guess.


I think a thread on the game/app is enough on getting proper feedback. IMHO, ratings only work when dealing huge amount of users from different areas and, since we are a niche thing, I see no reason for such.
 
You see Dreamcast is possible,then my Coderfriends lets programming and start the New Competition.


So many Softwaregoals


- Dreamcast Emulators


- OpenGLES to OpenGL Wrappers then you can port normal OpenGL Games on Pandora


- PSP Emulators(but i think this Emu isnt realy possible)


- PS2 Emulator(wohah cool Atelier Iris on Pandora)


- Space Trader Games like Privateer,X Beyond the Frontier or Elite 3 First Encounters


- New OSses :)


- and naturaly self programmed Games with many many Gamefun :D


a new Competition Sounds good B)
 
For Dreamcast emulators, you can't code them for a competition wih a limited amount of time.


There is too much involved in this.


Drkraziel could release and polish his emu for Pandora though...


On openglwrappers: As poke said in the other thread, there alread are wrappers, but they won't work for everything.


You can already use nanogl or jwzgles if you want to try some. Unfortunately you have to learn how to include an includefile and add a sourcefile to your binary to use them...


Well or do the conversion yourself. There is a good tutorial, which covers a lot of stuff to get started with this on the wiki. It is not that hard, even for noncoders like me.


Or even do the conversion and let a wrapper handle stuff like displaylists for you like sebt3 and I did for pollybegone.


And Ps2 emu? Come on...
 
Last edited by a moderator:
And Ps2 emu? Come on...

Ps2 emulation is obviously not possible on Pandora 1 / 1.5 but should be doable for Pandora 2.


I would add to Ingoreis list also:


-Mupen 2.0 (hope Stradian work will be continued by other devs)


-Mame 0.106 with ASM cores (this would rock!)


-Daphne SINGE (Mad Dog Mac Cree on the Pandora would be so cool)
 
And Ps2 emu? Come on...

Ps2 emulation is obviously not possible on Pandora 1 / 1.5 but should be doable for Pandora 2.

You realize current high end desktop computers struggle with PS2 emulation right?


There won't be that major of a jump in ARM technology by the 2014 estimate of when the Pandora 2 comes out.
 
Back
Top