Ratings

From Sega Retro

How will this average feature work: will it need to be updated manually when a new score is added? I know MediaWiki has a basic system that we can use to automate everything, but how will this extend to multiconsole and Template:Num?
Hell, we should probably automate the magazines that don't use a single number...- Andlabs 16:33, 14 August 2011 (CDT)

I just calculated things by hand - I think there was an attempt at automating things when we first made this template but it didn't work out. Besides, if you're working with multiple consoles it'll probably get a bit confused. It's tedious, but hunting out old magazines in the first place probably takes longer.
More of an experiment to see if it would look nice really. Tons of pages use this template and it would take a while to go through them all and add avereages -Black Squirrel 05:03, 15 August 2011 (CDT)

testes

Sega Mega Drive
25 Sega Retro Average
Show review scores
Sega Saturn
50 Sega Retro Average
Show review scores
Sega Dreamcast
75 Sega Retro Average
Show review scores
IBM PC compatibles
47 Sega Retro Average
Hide review scores
Publication Score Source
DC-UK 53 №11
DC-UK 2: DC Harder 15 №101
DC-UK with a Vengeance 85 №501
Live Free or DC-UK 35 №7
A Good Day to DC-UK 97 [1]
iOS
86 Sega Retro Average
Hide review scores
Publication Score Source
DC-UK 53 №11
DC-UK 2: DC Harder 15 №101
DC-UK with a Vengeance 85 №501
Live Free or DC-UK 35 №7
A Good Day to DC-UK 97 №321

or is this too much like Metacritic

the idea here is that it would be one per system and you'd be able to show/hide each list, thus reducing clutter. Probably needs tweaking. Probably not easy to convert into a proper, working template either. But yeah -Black Squirrel 16:20, 9 December 2011 (CST)

If we do go with that one, there might be a way to automate taking the averages but it might involve an extension and might not work with the average at the top; I'd have to look into it - Andlabs 16:35, 9 December 2011 (CST)
I really like this, but we also -really- need it to do averaging automatically. If we have to get an extension to do it, so be it. - Scarred Sun 14:35, 15 December 2011 (CST)

7

2

5
If you see 7 2 5, then we have both ParserFunctions and Variables extensions, which should be all we need (as far as I can tell)
Maybe we can also make it slightly fancier by providing things like an automatic GamePro score calculator that one can show the original score as a popup, but IDK if/how that would work - Andlabs 16:50, 15 December 2011 (CST)

ok so we have ParserFunctions but not Variables - Andlabs 16:55, 15 December 2011 (CST)

OK now that it works

I went ahead and added a collapsible element on the template (I had wanted it to show "out of x reviews" but it looked kind of assy that way.) My question is: how do we want to handle rescored games? That is, a game gets reviewed in one issue of a magazine and then gets a second review (usually for a buyer's guide or something.) My gut is to go with the first recorded review, as that stays truer to the time and truer to the goal of getting all reviews of that time. - Scarred Sun 23:54, 17 March 2012 (CDT)

I already made a "cvg_1" for second Computer & Video Games reviews. I wouldn't like to not have second review scores recorded, but I'd be fine with not counting them towards the average. - Hivebrain 00:20, 18 March 2012 (CDT)
Perhaps something after the template, to indicate that a rescore happened?
Sega Saturn
52 Sega Retro Average
Based on 2 reviews
Publication Score Source
Honest Reviews 100 №120
Irate Gamer 4 №14
The following rescores later happened:
Magazine New Score Notes
Honest Reviews 10 №124 revealed that the CEO of Honest Reviews's publisher Good Reads had been paid by Acclaim to spin a positive review in the hopes that would lead to some good publicity as the game had been mercilessly torn to bits by every other major publication at the time. He personally ghostwrote the review and snuck it in the galley proofs reportedly mere minutes before they went to the press. Upon finding this out, the original reviewer quit and sued Acclaim and Good Reads; Good Reads's CEO resigned shortly thereafter.
Irate Gamer 52 №13 accidentally swapped the review breakouts for this game and Time Travel Salesman Simulator 3.
...on second thought that could get unwieldly, but explaining the situation might be helpful - Andlabs 00:35, 18 March 2012 (CDT)
They don't usually give reasons for rereviews. Just stick them at the bottom of the list like this:
Sega Retro Average 
Publication Score Source
90 №123
82 №45
Computer & Video Games (rescore) 100 №125
Sega Mega Drive
86
Based on
2 reviews
Sega Retro Average 
Publication Version Score
Sega Mega Drive
N/A
Based on
0 reviews

Ratings

- Hivebrain 01:16, 18 March 2012 (CDT)
I'd average the two scores. So if CVG gave the game 78 one month and 80 the next, we'd list it as 79 and provide two sources.
The problem is, you can't always treat the first review as the most accurate. Some magazines review the import version of a game before the localised one. -Black Squirrel 05:20, 18 March 2012 (CDT)

On another note, I'd like to figure out a way to have the "based on # reviewS" show up on the collapsed box without it looking gay. "Oh hey, this game has a 58 review, it must be meh... wait it's based on one review :/" - Andlabs 15:10, 18 March 2012 (CDT)

Non-listed magazines

How do you add ratings for magazines not listed, such as Zero?

http://en.wikipedia.org/wiki/Zero_%28video_game_magazine%29

Like I just did - see the history -Black Squirrel 05:26, 21 May 2012 (CDT)

I understand now, thank you.

Spam protection filter

"The text you wanted to save was blocked by the spam filter. This is probably caused by a link to a blacklisted external site.

The following text is what triggered our spam filter: display:non*"

Now I can't revert to the previous edit either as it also contains "display:non*". Could someone fix this please as the template is now displayed wrong on all game pages. (*=e, I can't post it here either as the Spam protection filter continues to block it)

Thanks for reverting my edit, but Spam filter still prevents me from editing this page, I'd like to add Games-X magazine so that I can start adding the reviews from this magazine.

Aaargghh ... now I can't post a link without spam filter asking me to tell cats from dogs, but there's no submit button to enter the answer and the pictures are also obscured by the black Sega Retro bar at the top of the page. This is frustrating --Pirate Dragon 15:10, 6 August 2012 (CDT)

EDGE and Page numbers

How are we placing page numbers in the _source= fields? And just to confirm: for Edge's x/10 ratings we just multiply by 10? - Andlabs 16:18, 27 October 2012 (CDT)

№101, p55
Vol. 1, №66, p23
I just extended the num template a bit. Just convert any scores to a straight percentage, so yes multiply by 10. - Hivebrain 20:02, 27 October 2012 (CDT)
Thanks - Andlabs 20:45, 27 October 2012 (CDT)

High Scores

So guys, how feasible do you think it is to create a Metacritic-style "high scores" table, i.e. aggregate all these ratings templates so we can compare Mega Drive games against each other or whatever

I've tried to do this in the past, but it's a really awkward mess. If you use DPL you can grab all the ratings templates from pages in a category, but it won't differentiate between consoles and you can't pick apart values (it will always render the table in full, console icons and individual review scores in-tact - you can't just have it print the average score). Also Cloudflare hates you if you use DPL too often

Supposedly there is syntax available that will let you generate lists depending on template parameters, but it's pretty horrible and woefully under-documented. Any ideas? -Black Squirrel 14:33, 23 February 2013 (CST)

2016 update

Not sure this is populating Category:Rating without PDF source - it's likely a quick fix but I'm not sure if we want a more elegant solution? Do we want to build this functionality into the Num template, in case we need a PDF source somewhere else?

p.s. do any testing on Template:RatingsTest - because there's so many pages relying on this template, updates kinda cripple the server in the short term, as mediawiki goes through every page using this template and updates it. It takes hours to move pages en masse to and from categories - that's why I didn't notice the above wasn't happening - the category seemed full and it would take an age to referesh.

It's also not flagging up non-magazine publications even though I'm using Template:OnlineRatingTest. I don't know why (yet) - Mediawiki syntax is really hard to read. -Black Squirrel (talk) 11:52, 25 June 2016 (CDT)

The way it works now, any edit to the magazine name templates will also lag the server, which is something to keep in mind. - Hivebrain (talk) 12:53, 25 June 2016 (CDT)
I'm not sure it works like that.
I think, when you update a template, Mediawiki cycles through any pages that link to that template. There's thousands of (big) pages that use the Ratings template so that takes a while (though why it's given such a high priority as to lag the user experience is another matter).
The MagazineName template is only used by a small handful of pages (which are themselves templates), so that doesn't take very long to cycle through. Something like MagazineNameUK is technically only linked to once, so the update is instantaneous.
I've just added something to MagazineNameJP which will be getting called indirectly from thousands of ratings templates and 700+ magtables, and it hasn't caused an issue... unless there's a delay -Black Squirrel (talk) 13:19, 25 June 2016 (CDT)
Okay there was a delay and it killed the wiki for half an hour. That's not good -Black Squirrel (talk) 14:07, 25 June 2016 (CDT)

2018 update

Had a couple of ideas:

Sega Retro Average 
Publication Score Source
80 [1]
Arcade
80
Based on
1 review
Sega Retro Average 
Publication Version Score
Arcade
N/A
Based on
0 reviews

Ratings

I'm thinking we no longer need to use Template:num anymore for references. Magref will do - it helps circumvent some of the problems with magazines that don't have numbers. (i.e. we wouldn't be doing "№1998-01" anymore), and it's less for editors to think about. There is a slight issue perhaps with real page numbers not lining up with how the magazine counts page numbers... but I almost feel is if that "offset" should be baked into the cargo table somehow. If we even care.

Secondly there are more magazines than I originally thought that don't give an overall review score. EGM is a known example where four people give judgement (which we currently average), but loads of Japanese magazines (like Famitsu) do it as well. While I still think the Sega Retro average score is a useful thing to have, I think we can afford to show our workings

e.g.

  • Low-G-Man for the NES:

File:EGM US 013.pdf

I think we should probably say the score given was "6, 8, 6, 5". We then convert to "62.5" internally and use that as part of the average.

Likewise if a magazine uses a star system, show the actual stars:

★☆☆☆☆

Wikipedia does this, but they don't calculate an average.

so

Publication Score Average Source
EGM 6, 8, 6, 5 63 [1]
Something else ★☆☆☆☆ SegaRetro SealOfQuality.png 20 [1]

if the magazine gives ones of those awards like "crash smash" or "C+VG hit", we could support that too. Bit excessive, but it represents what the magazine actually gave the game.

This system would mean editors don't have to try and work out what the numerical meaning of "B-" is, because we could build it into the template. I've avoided putting in EGM and GamePro scores for this reason - I can't be bothered to do the maths -Black Squirrel (talk) 13:49, 5 December 2018 (CST)

Sega Retro Average 
Publication Score Source
57 [2]
Expression error: Unexpected < operator. [2]
Expression error: Unexpected < operator. [2]
Expression error: Unrecognized word "b". [2]
{{|L}} Expression error: Unexpected < operator.
Based on
4 reviews
Sega Retro Average 
Publication Version Score
{{|L}} N/A
Based on
0 reviews

Ratings

This was my plan (sans the extra column) ->
this method isn't going to work though, because the new Star/LetterRating templates have been evaluated prior to this ratings template being called. So in this example you get 75 (B) for the first magazine, and then nothing.
This will need more thought. On the plus side, a bit of tweaking and users won't have to do any maths when adding a score. -Black Squirrel (talk) 13:27, 6 December 2018 (CST)
Instead of "joypad={{StarRating|4|6}}", you could have "joypad=4/6", then the ratings template detects the "/" and uses StarRating accordingly. That should let you get the rating value as it's needed, and it'll make it easier for people to add ratings.
One thing that slightly bothers me about star ratings is the way they're scaled, so you only ever have a rating between 20 and 100%. It might make more sense to have one star equals 0%, two equals 25% and so on. This is just a minor quibble that can be dealt with later. - Hivebrain (talk) 14:16, 6 December 2018 (CST)
I suspect some means of explicitly telling the template what rating system is in use will be required - there's at least one German magazine that I've come across that uses the country's academic system (which handily I can't find right now, but it's definitely there), and UK magazines like ACE which rate out of 1000.
One thing about the star system which I've not implemented are half stars. I uploaded the graphic to Retro CDN, but I can't think of any magazines off the top of my head which use half stars, so I didn't bother implementing it (I'm almost positive there's at least one though). It means 1/5 can't be 0%, because you might get 0.5/5.
I did think once of using some fancier maths, i.e. adjust the scale relative to the lowest (and highest?) score a magazine ever gave a game. So if a magazine never gave a score below 20%, 20% would be treated as "0". While it would theoretically remove a bias, it falls apart a bit if the magazine didn't review many games. -Black Squirrel (talk) 14:49, 6 December 2018 (CST)
IIRC Metacritic keeps records of how much a publication deviates from average scores. It's not very scientific but it lets you know if a publication is unusually positive or negative compared to others. We could do that... somehow -Black Squirrel (talk) 14:53, 6 December 2018 (CST)

I was going to make vectors, but that took too long:

GamePro Rating 10.pngGamePro Rating 15.pngGamePro Rating 20.pngGamePro Rating 25.pngGamePro Rating 30.pngGamePro Rating 35.pngGamePro Rating 40.pngGamePro Rating 45.pngGamePro Rating 50.png

GamePro ratings. Don't know what I'm going to do with them, but they're here.

GamePro Rating 50.png GamePro Rating 40.png GamePro Rating 40.png GamePro Rating 40.png GamePro Rating 50.png
GamePro Rating 50.png GamePro Rating 40.png GamePro Rating 40.png GamePro Rating 40.png GamePro Rating 50.png

It's a bit unsightly at lower resolutions. This works out as (((5/5)+(4/5)+(4/5)+(4/5)+(5/5))/5)*100 = 88 (or 70.4 if you adjust for the fact no games could get less than 20%).

experiment etc. -Black Squirrel (talk) 13:16, 17 December 2018 (CST)

Sega Retro Average 
Publication Score Source
Expression error: Unexpected < operator. [2]
Expression error: Unrecognized punctuation character "[". [2]
{{|L}} Expression error: Unexpected < operator.
Based on
2 reviews
Sega Retro Average 
Publication Version Score
{{|L}} N/A
Based on
0 reviews

Ratings


MeanMachinesMegaGame.png I spent far longer on this than I probably should have -Black Squirrel (talk) 06:40, 21 December 2018 (CST)

Broken Ratings Template?

When expanding the "Sega Retro Average" box, most of the reviews no longer display. Is there a new way that reviews should be added, or is this a known issue that will eventually be fixed? If the latter, is it worth adding reviews now, or waiting until the issue is fixed incase reviews will need to be added in a different way? Pirate Dragon (talk) 15:16, 28 January 2020 (EST)

Still add reviews - it's important that the information is saved, even if we can't see it
Although you might just need to go Actions -> Purge Cache (I really want the wiki to do this automatically... but I want lots of things) -Black Squirrel (talk) 15:28, 28 January 2020 (EST)
Purging the cache doesn't fix it, but I'll add them for now anyway. Before I add thousands of ratings, what's the preferred format for the source? "magref|example|99|99|printpage=99"? What's the difference between page=99 and printpage=99? Sorry for all the questions, it's been a while since I've added these and a lot has changed in the meantime. Pirate Dragon (talk) 16:55, 29 January 2020 (EST)
Page is the page number of the PDF (so it links correctly)
Printpage is the page number(s) printed in the magazine. So if you had a 4 page review starting on page 99, it would be:
|page=99
|printpage=99-102
Also the PDF numbers and the numbers printed in the magazine aren't always in sync. -Black Squirrel (talk) 17:05, 29 January 2020 (EST)
Got it, thanks. Pirate Dragon (talk) 17:08, 29 January 2020 (EST)

2020 thoughts

There's a column in Sega Magazine that's stuck with me:

File:SegaMagazine UK 06.pdf

"on the plus side, most magazines will tell you what version they reviewed"

They do.

We don't.

And while a lot of the time it doesn't matter, it's perhaps something we should log if we can, especially if a magazine reviews a game twice.

So for Streets of Rage 3 which does vary between regions

  • an overall aggregate score for all versions
  • JP version score
  • US version score
  • EU version score

because the crappy PAL conversion is bound to be treated differently than the original Japanese one


I've had another thought since 2018 too - now that we have non-Sega magazine pages, perhaps we could merge in Sega Retro:Ratings. A template/cargo table or something that defines the rules for reviews per publication.

I think we might be overcome a few errors this way. For example, a magazine that never scores below 1/10 means that every game is worth at least 10%, which means there could be a postive skew on results.

I like the idea of just having to put in the rating the magazine gives (e.g. "15", "B", "1,2,3,4", "good") and have it calculate something behind the scenes. Perhaps with fancy icons (see GamePro above) -Black Squirrel (talk) 09:32, 19 April 2020 (EDT)

Oh and also, some means of indentifying whether we think a magazine is reviewing a prototype. That's not likely to be an exact science, but it would help inform users about reviews that could be garbage. -Black Squirrel (talk) 09:35, 19 April 2020 (EDT)


Okay
Given that this template is broken anyway, it might be wiser to go back to using Template:Magtable and Template:Magrow for storing review scores. This would mean we don't have to maintain two separate lists - whenever a rating is added, a row in the magazine articles sub-page is added. This would also give us room to show GamePro-esque ratings and any "awards", without it intruding too much into the main game page. Also it means the scores are re-calculated less often which will be good for performance.
File:SegaMagazine UK 06.pdf Sega Magazine (UK) #6: "June 1994" (1994-05-15) Review: Something

A review by Sega Magazine. Reviewer XXX says that the music "sounds like dying cats" and playing it is "like sawing off your own leg", before awarding a final score of 7/10.

Version: NTSC-J final Rating:
GamePro Rating 50.png GamePro Rating 40.png GamePro Rating 40.png GamePro Rating 40.png GamePro Rating 50.png
Awards:

MeanMachinesMegaGame.png

Calculated score:
70

[Methodology]

the Ratings template would just be used to show ratings, rather than store them -Black Squirrel (talk) 14:13, 19 April 2020 (EDT)

I like the sound of that. I wonder if it's possible to take it a step further. If/when we get the External Data extension, we could list the contents of a magazine on its file page on CDN (which would store the data in a cargo table), then pull that info over here for both magazine article tables and ratings. - Hivebrain (talk) 15:10, 19 April 2020 (EDT)

"I think we might be overcome a few errors this way. For example, a magazine that never scores below 1/10 means that every game is worth at least 10%, which means there could be a postive skew on results." But very few magazines that score out of 100 ever scored any games below 10, so I don't think it really would bias it to a noticable amount. Unlike computer games, console games generally had a filter from the manufacturer preventing any truly broken games that would score so low getting released. Assuming that 1=0 seems to overly complicate matters. If we also did that to 1-100 scores then it would throw everything off by 1. That's just my opinion anyway.Pirate Dragon (talk) 08:25, 20 April 2020 (EDT)

I won't be doing 1-100 - when you're working in percentages, 0% is about as likely as 100%. In fact that might be the case with a 0-10 scale too.
The Sega Retro average is probably best viewed not as a "0 = broken, 100 = perfect" scale, but more how good games are relative to all the other games for a console. And there is a bias at the moment (you can see it) - there is only one Mega Drive game rated below 20%, which means 1/5th of the scale is essentially unused. Half the library is sitting between 70 and 90% which isn't very useful - "everything is great!".
It could actually be a numbered list (e.g. "this is the 55th best game on the Mega Drive"), but then the next question would be "why?" and you'd have to show your workings. Also it breaks down if you don't know how many games there are - 55/55 is worse than 55/1000.
But it's all up in the air - I may not even get this system working -Black Squirrel (talk) 14:10, 20 April 2020 (EDT)
Smash T.V./Magazine articles
Special:CargoTables/ratings
Things to do
  • Default rating system per magazine, so we only need to specify when they do something strange (might have to do something with dates in CVG's case since they swapped it out for 1-5)
  • Each magtable is going to have to specify a format. Which maybe it should have been doing anyway.
  • Make layout not suck
  • Make Template:Ratings read from cargo table
  • try not to think about the conversion process
-Black Squirrel (talk) 17:42, 20 April 2020 (EDT)
Okay I don't trust myself - I'm going to set magazine defaults to 0-x rather than 1-x for the time being, and I'll explore positive skews at a later date -Black Squirrel (talk) 08:19, 21 April 2020 (EDT)
We currently have average ratings colours from red (1-20%) through to green (81-100%). As you point out, only one Mega Drive game is in the "red zone". Maybe it would be better if the colours represented which portion of that list they fall in, ie bottom 20% of list = red, rather than just 20% or less rating = red. I'm not sure how practical it would be to do that though.Pirate Dragon (talk) 09:00, 21 April 2020 (EDT)
GamePro Rating 10 1996.png GamePro Rating 15 1996.png GamePro Rating 20 1996.png GamePro Rating 25 1996.png GamePro Rating 30 1996.png GamePro Rating 35 1996.png GamePro Rating 40 1996.png GamePro Rating 45 1996.png GamePro Rating 50 1996.png
GamePro Rating 05 1997.png GamePro Rating 10 1997.png GamePro Rating 15 1997.png GamePro Rating 20 1997.png GamePro Rating 25 1997.png GamePro Rating 30 1997.png GamePro Rating 35 1997.png GamePro Rating 40 1997.png GamePro Rating 45 1997.png GamePro Rating 50 1997.png
A chore to clean up, but a pleasing result. Too bad there's another set that emerges around 1999/2000 -Black Squirrel (talk) 16:49, 23 April 2020 (EDT)
GamePro Rating 05 1999.png GamePro Rating 10 1999.png GamePro Rating 15 1999.png GamePro Rating 20 1999.png GamePro Rating 25 1999.png GamePro Rating 30 1999.png GamePro Rating 35 1999.png GamePro Rating 40 1999.png GamePro Rating 45 1999.png GamePro Rating 50 1999.png
Fun fact - Paint.NET doesn't save backups. -Black Squirrel (talk) 14:07, 24 April 2020 (EDT)

clash

Clock.svg
This task is unfinished
This is a wiki task that requires more thought or peer review. An important issue that needs to be solved, which will appear at the top of Sega Retro:Todo until this message is removed.

If this template appears above Template:Creditstable, creditstable doesn't work (e.g. Sonic R). There'll be a variable that needs clearing, probably. -Black Squirrel (talk) 06:18, 25 October 2021 (EDT)

Adding new ratings systems

I'm sure there's a simple way to do this, but I'm not up to date on it. L Ordinateur Individuel scores 0-5 stars per reviewer, with up to 9 people reviewing each game. Currently we seem to max out at 7 x 0-5 scores for a review, so we're missing the full scores for some SG-1000 games.Pirate Dragon (talk)

"Simple" if you know what you're doing. And I wouldn't blame you for not knowing, because no matter how you write out Mediawiki syntax, it ends up being confusing and messy
Template:ConvertRating/Multiple is the place to look - it was hard-coded to stop at 7 because... I hadn't written conditionals for values greater than 7 -Black Squirrel (talk) 17:49, 29 August 2022 (EDT)
Thanks!Pirate Dragon (talk)
  1. 1.0 1.1 1.2 Computer & Video Games, "September 1994" (UK; 1994-08-15), page 64
  2. 2.0 2.1 2.2 2.3 2.4 2.5 Consoles +, "Septembre 1994" (FR; 1994-0x-xx), page 122