I do not like the depot review system, or more specifically, how people are using it.

My latest 3059 review has nothing to do with the actual gameplay, or even the game itself. It more looks like a forum question, or something that would just need to be cleared up in an e-mail.

In addition, the latest Ants! reviewer didn't even play my game, but reviewed it anyway. Again, something that could have been cleared up in an e-mail or forum.

Anyway, I am frustrated that my game got a 2 star rating because I chose not to support Alt+Tab (for very specific reasons). I instead programmed a very quick and seamless save/exit and restart feature to replace Alt+Tab.

Bottom line is, I would MUCH rather see people with single issues or questions be directed to me as an e-mail, memo or forum topic before deciding to score my entire game on it.

PM the reviewer and ask him to actually test the game properly and change the review. If that doesn't work PM Matthew and ask him to remove the inappropriate review...

I wouldn't get so sour over those reviews, they don't seem that inappropriate to me.. the game didn't work for the one guy, so he gave it a bad review (seems justified to me), and the other guy had problems playing it as well.

They aren't professionally written, and everyone knows that

I hate it when game developers these days purposely causing errors when the games lose focus. That bothers me so much, and I don't even get any pop-ups.

His complaint was that he was unable to do just that

I'd do the same thing with a piece of hardware. If I installed it and it didn't work, it's getting a low score.

OK, say version X didn't work for player Y. Y then leaves a bad review saying X doesn't work. The programmer then fixes that bug in version X+1, but that bad review for version X still ruins the overall score, even though nobody will ever experience that problem again.

Well, I have a specific reason for not allowing switching of applications. I have used those trainers a million times, memory hacking programs blah blah blah, and I just didn't want it to make it too easy for a player to cheat in my game by switching to a memory hacking program. I know a cheater will eventually find a way to hack my game if they wish... I just don't want to support it. Instead, I make the game quickly save and quit, so it can exit as quickly as an Alt+Tab.

Being able to change and update reviews would be great, so old reviews wouldn't get stuck with newer versions (especially if the review is based on a bug or feature etc.)

For a quick fix, how about requiring reviews to include a version number? Then once the depot version changes, just put a symbol next to the review that means that review was for an older version.

Yeah, it should get a low score; hardware is released to hundreds of thousands of people and is hard to fix or update as quickly as software. It just sucks on this Allegro site for software we can update in a days time. I fixed that bug with Ants! about 1000 years ago, but that 'review' will always be there.

If you ask me, these are NOT reviews, but bug reports. I do not believe they belong where they are.

AFAIK if a person reviews again, it overwrites the previous review.

I haven't heard from the last reviewer yet; I hope he responds. I fixed his bug (the freezing), and added a warning message on the title screen. I hope he can leave a more appropriate review.

I don't think I'll support Alt+Tab for reasons above, I believe my save/exit and restart feature is sufficient.

Maybe the ability to mark a review as "applying to a previous version" when an update is done? And, optionally, notifying the admin (ML) that a review is no longer applicable because the bug was fixed? Or perhaps even a weighting system of some sorts so really old reviews on a game when it was back in beta doesn't destroy the rating of a 1.0 release?

I know these are just a few ideas, but I think they are feasible; especially the admin notification and previous version warning on reviews.

I second this.

I also 'second' the suggestion.

I think the notifying would cause ML's PM box be full quite quickly

Why not the project "owner" could remove those which present innacurate data. Tho then there is this risk of manipulating "negative" comments away.

To avoid manipulating, let's reviewer will have one week to update his review and the it will be deleted.

Well, so you punish 98% of your players for the 2% of cheaters?
And 100% of the players will hate it (both cheaters and normal players).

I removed the two opinions in question, although I'm not in 100% agreement. I have only done so in consideration that they were both in related to specific bugs that may no longer apply and the users making the review were notified.

I do strongly recommend that when reviewing a project in regards to a bug that you first contact the developer.

I will never allow a developer to selectively remove opinions he does not like. However, I have made some changes (if they ever make it live), namely that opinions are linked to versions, and the developer may chose to delete all previous opinions when creating a new version.

Though, that can be abused. They may just tack on a new version ".0.0.0.1b" so that they can kill all the comments.

Perhaps he could just make sure the values are within proper ranges. That's (partially) how punkbuster works.

So? They will lose all their reviews. If they have never gotten a good one, then yes they could keep deleting the bad ones. That won't help them to abuse the system in the long run.

I like that idea

How about a fixed time limit on a review, of say 3 months? Reviews older than this will not be counted for the overall score and not visible without extra effort (another click).

The reviewer could log on every month and press the "I still hate this" button if they want their review to stand indefinitely

Good point. However, I'm worried that that 2% of players can ruin some of the sense of accomplishment in getting a high level character, if you could have just otherwise cheated fairly easily... might as well make those 2% work for it... i just would like to do what I think is a small change to help keep 3059 a little more 'valid.'

Hrm, there really are no real 'proper' ranges for my game... you can keep improving your player to a pretty insane level, and I don't want to mess up any of their games. I've seen this attempted in other single-player games, and it usually doesn't work very well, or I just kept tweaking numbers until it worked.

Thank you very much. However, I noticed the overall ratings are still effected.

Do the reviews publically say what version they are linked to? Anyone see any of these new reviews?

There might be some old reviews I still want on my front page that still apply...

I think Matt's new system sounds pretty good. Although, in my situation, I would not have liked to remove some of my other reviews (which are good quality) for 1 inappropriate review.

I think anybody around would realize if somebody suddenly begins releasing versions as soon as he gets a first negative review.

As for the trainer stuff, I can write you one that will hook into your application and prevent the ALT+TAB key combination from reaching your program

Now, seriously speaking, I had never thought about creating a trainer for some Allegro game. I think even I have never recompiled code with a new set of parameters to make the game easier. If ever, you do that for commercial games. Believe me, nobody would try to hack your game unless you get over a million downloads and Microsoft, Electronic Arts or Nintendo offers some good money for it

Eh, I guess it has to do with my background with memory hacking programs like ArtMoney. You can easily track down variables and change them with any running process, but you had to switch applications back and forth to search for changes in values. I thought the easiest way to prevent this was to prevent Alt+Tab at some level. I didn't want to lose the 'switch out' feature, so I coded an entire quick-save-exit and restart-at-same-point feature.

Bottom line is, I wanted it to take a little more than firing up ArtMoney to cheat at my game...

Jeremy, you could comprimize... quick save upon ALT Tab, clearing all values, and Quick load upon Alt tabbing back into the application. That way the values are not in ram when you alt-tab, but people can easily switch back to the application.

Hrm.. interesting.. but I don't know how I could possibly 'clear all values'... there some are insane amounts of values... sounds like it would take quite a bit of work.. my compromise was taking away the Alt+Tab but programming in the quick-save-exit and restart feature.

Clear all values? It is rather easy, just set them all to zero... Or better yet, set them to a number generated by a random number generator to really throw off people.

Seriously, if you are going to go to the trouble to write all of the values to disk, what is just an additional line setting the value to zero, or some other randomly generated value?

I have it save it all to disk, because the game needs to remember everything when the player turns off the computer or whatever. Thank you, it sounds interesting, but I don't think that is the solution I am looking for.

I have not used memory trainers myself but i guess they work by checking the memory for certain values (which the user identifies in game) until they find the variable memory locations. Then, when the memory locations are found they let the user change those values at the located addresses. Or are they more sophisticated?

If not, why don't you try out a couple of ways yourself and see if you can fool (most of) them.

Suggestion1:
For example you could try storing each variable in another form (using the binary operator XOR for example to invert it, it is pretty fast). Or using something like x = x + 123 % sizeof(x). As the value will be changed it might be harder to look up.

Suggestion2:
You could try storing each variable in multiple locations (L1, L2, ... , Ln). WRITEing first checks that L1 ... Ln are all equal, if not someone has cheated. And then writes the same value to all locations. When READing you read from an arbitrary location Lx where 1 <= x <= n. Even if they can look up all locations they need to change them all, and all before you do your next WRITE or the cheating will be detected.

Suggestion3:
Suggestion1 && Suggestion2

Just some ideas, the above will slow down your program to a certain extent. But i wonder if it will ever be noticable (read as: i think not).

IMHO, if somebody wants a memory trainer, let him use a memory trainer.

I agree. I have never understood why to cheat in a game anyway? It spoils your own fun. It must be some 1337 thingy that passed me by.

Except it defeats the purpose of an internet ranking system.