Spectrum Computing

Since we now have magazine scores in ZXDB, as a result of the ZXSR integration, would it be worth while to calculate an average score for each game, from which Top 10s/Top 100s could be created?

I'm open to that idea [mention]Rorthron[/mention].

Of the magazines indexed in ZXSR I can see most used a score out of 10, apart from Crash which used out of 100 which is simple enough. Were there any other scoring methodologies?

I do wonder how useful this would be, as ZX Computing didn't seem to provide a score (Certainly with the review of Avalon I looked at).

It would also only be based on the magazines in ZXSR.

I'm not against it, just trying to see the whole picture.

Your Sinclair initially had out of ten then later changed to a percentage.

Sinclair User started with out of five, then out of ten (I think) and eventually a percentage.

Crash was always a percentage and I think all of the others were either out of five or out of ten. Apart from Ace - they bizarrely had out of 1000.

[mention]Rorthron[/mention] ,

Thanks to the code from Einar, the average magazine score now appears in the main game information section.

Peter

Thanks! That was quick.

Is there any interest in a Top 100 page, based in these scores?

Renegade 3 - 79.675%
Hypersports - 78.4%

Not completely accurate at the moment as there are review scores in ZXSR which aren't currently shown alongside the magazine review entries on the game pages.

8BitAG wrote: ↑Mon Sep 14, 2020 8:10 am Not completely accurate at the moment as there are review scores in ZXSR which aren't currently shown alongside the magazine review entries on the game pages.

Thanks for the feedback. Could you give me a couple of examples and I can advise which bug thread to add too.

Cheers.

PeterJ wrote: ↑Mon Sep 14, 2020 6:30 am @Rorthron ,

Thanks to the code from Einar, the average magazine score now appears in the main game information section.

Peter

Great! For display on the pages themselves I would round the average score down to 0 decimals, 82.080% - the .080% is just noise IMHO.

Thanks Stefan,

Maybe one decimal place would be a good compromise

With a top 100 average score done based on reviews we may end up with a bunch of games with 100% (if they've got 5/5 or 10/10 in a couple of mags).

I think this is where the old-style crowd-sourced anonymous ratings were generally more accurate in giving you an idea of how good a game is. But it might be an interesting exercise to see what happens if someone can be bothered to do it.

Don't forget to include ...before any calculations...



[Edit: before anyone mentions it, I don't know SQL so the syntax might be wrong]

Excellent feature!

I just checked some classic games:

Sabre Wulf: 90,3%
Knight Lore: 90,57%
Pyjamarama: 87%
Cyclone: 83,80%
Highway Encounter: 85,25%
DT Decathlon: 70,50%
Jet Set Willy: 96,25%
Manic Miner: 96%

I think they're reliable scores as behind them there were the main magazines. Obviously we may have different opinions on each one, but it's good as a reference.

Good work.

Ant Attack - 59%

Morkin wrote: ↑Mon Sep 14, 2020 9:57 am Ant Attack - 59%

There was a late re-review of 22%, which drags the score down.

For similar reasons Jetpac scores 60%. None of the original reviews had a numeric score, so the only rating is a later 6/10 re-review, which claimed it was "dated".

There does seem to have been a journalistic need to lower scores sharply in re-reviews. Seriously, 22%? It's just attention seeking in my opinion.

The average score is not perfect, especially for early games, but I think it's a good benchmark, nonetheless.

Rorthron wrote: ↑Mon Sep 14, 2020 10:04 am

Morkin wrote: ↑Mon Sep 14, 2020 9:57 am Ant Attack - 59%

The average score is not perfect, especially for early games, but I think it's a good benchmark, nonetheless.

Agreed. Not perfect, but still useful as long as the information is taken in context.

Replying to the comments from [mention]Morkin[/mention] .

No promises here (and no timescales), but I'm building a specification for [mention]Ivanzx[/mention] to send to his friend for building a scoring system.

The hardest thing (in my eyes) is checking to see if users are logged onto the forum from the main site to allow them to vote

What do people think of the voting system on Lemon64? This is what I'm thinking of asking for.

Thanks

Edit: We won't be adding user reviews as these would need manual checking before publication and we don't have the capacity to do this.

PeterJ wrote: ↑Mon Sep 14, 2020 10:25 am
What do people think of the voting system on Lemon64? This is what I'm thinking of asking for.

Thanks

Edit: We won't be adding user reviews as these would need manual checking before publication and we don't have the capacity to do this.

I think you're referring to this:



"You must be logged in to vote" appears before voting.

That's the idea I have in mind. It even shows which score by each member. No anonym votes, no trolling.

It'd be great.

Yes, [mention]Juan F. Ramirez[/mention],

I think it looks great too. I will give others a few days before replying to Ivan.

I like the transparency of Lemon64's voting system as well.

Hi

I love the average rating feature. But, I'm not very fond of including re-reviews, simply because those new reviews were published when a certain game was re-released as a budget title, normally several years later. In that case, as far as I know almost every single game was rated worse than the first time around, so because of that a lot of games, even some classics, will have severely decreased ratings.

By the way, what about those magazines not using an overall rating, just a graphical or numerical score in different categories? For example C&VG did that for a long time, Microhobby too (later on changed for percentage), also ASM (Aktueller Software Markt).
Is there any way to incorporate that 'unorthodox' scores to the magazine reviews section and therefore to the average rating itself?

Here you have some examples:

Hi [mention]Audionautas[/mention],

Thanks for the feedback. This is not a perfect science and excluding re-reviews get too complicated.

Just treat it as a guide. Nothing more.

Note that only reviews stored in ZXSR and a having a numerical score are included.

Rorthron wrote: ↑Mon Sep 14, 2020 10:04 am There does seem to have been a journalistic need to lower scores sharply in re-reviews. Seriously, 22%? It's just attention seeking in my opinion.

Yeah, generally mags did seem to be quite harsh on the older games in their later re-reviews. Though I noticed that YS originally gave Renegade III 79% before boosting it later to 82%...

(I'll shut up about Renegade III now...)

PeterJ wrote: ↑Mon Sep 14, 2020 11:28 am Thanks for the feedback. This is not a perfect science and excluding re-reviews get too complicated.

In the future perhaps we can have an algorithm that only matches reviews close to the publication year, this way most of the re-reviews can be filtered.

I agree that this is only a guide, a reference as to how the game is, specially for obscure or unknown games that we find for the first time.

I think this must go 'step by step'. Let's just introduce it and see how does it go. In the near future improvements can be done.