You know, Leader just made a good point regarding the Vor's constructors. He'd just demanded their entire treasury (i.e. a N for 1 month tribute) Did it credit it back to them double? That would explain where they got the cash to crank them out. If that's the case then, well, it is a rather unfortunate and major 1.0 bug, and that would explain where the accusation of cheating came from.

~SDC~

The accusation of a cheating AI is pretty bad when the developer says it isn't, and you don't have incontrovertible evidence. He's also backtracking on other things, saying he didn't mean it the way most people would interpret it. Personally, I hate because Brad is spending time there posting, instead of giving us more goodies here (just kidding, Frogboy :))

Grifman

Geo: No, the bug didn't work that way. It only works in the X for Y months which, based on how people play, doesn't affect most people.

It only comes up when you're trading X for Y. Treaties are handled differently. One is handled as tribute, another is handled as a lease.

~SDC~

Brad: OK, cool. Would've just been awful if after all this, turned out it was a bug.

~SDC~

Brad: actually, just occured to me. Leases were bugged. The AI apparently knew this (re: your statement in OH NO that it was accepted trades because they were good for it). Doesn't that mean the Vor might have been leasing Constructors or even money from other races left and right, confident that they were getting a larger treasury for doing so? Seems that if you leased money for money, that'd benefit both sides...would the AI have been "smart" enough to offer this kind of deal if it calculated it as advantageous?

~SDC~

Oy...kinda sad that someone is allowed to write a review that doesn't even understand the product he's reviewing! I prefer sites that do both a 'professional' (whatever that means) review and has user reviews. I feel you get a better idea from comparing 'professional' vs 'joe everybody' reviews.

Geo: Leases and Tribute are different.

~SDC~

Brad: well, part of it is obviously a difference in terminology here. What's Tribute and what's Leasing?

Regardless, my original point stands. You flipped a sign, and had a way for a race to mint free money, whether it was via "tribute" or "leasing" doesn't matter. Moreover, the AI's evaluation routines apparently used the same calculation, and -knew- they were getting free money. I'm asking if it's at all possible it was exploiting that, hence being able to buy stuff in short order. Whether that exploit happened by X for Y or X for 1 is a non-issue.

(That said, I'd still like to know what you mean by "tribute" and what you mean by "lease", just for future conversations.

~SDC~

Geo: This is in the context of his review. The bug we fixed today didn't have anything to do with what he experienced. That was my point.

But even if it did affect it, the question then is, should that lead it to get a lower score than MOO 3 and HOMM IV in the AI category? Isn't that really the bottom line? HOMM IV got a 4.5 star AI score. GalCiv 2 star.

He has responded here:
http://forums.avault.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=3;t=001071;p=3

And it's pretty plain that he won't admit to any error. We're all just misunderstanding him apparently.

~SDC~

Brad: not trying to defend Richard--he's being a pompous ass, pardon my Drengin, and you certainly didn't deserve a 2 star AI. I'm just having a hard time seeing how a (supposedly) broke race can get a couple of Constructors in a few turns, even via leasing them--it'd need a down payment, and I'm wondering if it got that down payment via the bug in question. I'll hush, though...I'm just glad you found the bug.

~SDC~

AI players can tell techs just as players do.

Unfortunately, it is difficult to tell what exactly happened in his game. Remember, he is also claiming how these various terrible events happened in the first 50 turns too even though that is not possible.

What if the Drengin gave the Yor money or ships? We don't know.

What I do know is what the code says. The AI cannot build ships without money. It is impossible to tell what other sources of revenue the AI had. Unless he specifically traded ships/tech to the AI for X per turn for Y months, he wouldn't be affected. And given that he didn't seem to understand leases, it's pretty unlikely he would have done that in the peace treaty.

Odds are, he did what most everyone else would do, they asked for peace, and he demanded all their stuff including their treasury. That part works perfectly.

~SDC~

I'll second that motion. Rating HOMM IV and MOO3 as having a bettar AI is pur nonsense. I easily beat HOMM IV and MOO 3 was snore fest. However i am still slugging my way trying to beat this game and having fun at it. It may see simple but I like its elegance.

Its valid point the even if CGs AI cheated, as Brad noted, it still would make it the norm. Hell even the great Sid Meier's games the AI cheats more shamelessly than most others; and they do not even have the degree of compleity as GC in a lot of ways. the very fact that GC is 1) playable 2) challenging 3) a nicely involved (but not speadsheet complex) .. is in itself a great accomplishment.

Bottom line. I have never seen a Avault review worth much. I dont think many seriosu gamers pay much attention to the mass appeal sites. Trusting Avault for reviews is using Martha stewart's opinion on which car to buy.

dawg

Well, they closed the thread on Avault. But this was said by the editor:

"Brad investigated the save file on his end and sent us the info from the debugger. Low and behold - we were wrong (yes, we're not afraid to say it). What was actually taking place inside the game was a rather intense set of trades and diplomatic negotiations that gave the outward appearance that the AI was "getting help."

In my opinion, there was no way Rich could have interpreted this differently based on the information the player gets in the game; and so any comments that he didn't spend enough time playing, etc. are uncalled for. Rich did make one mistake in this area of the review - he didn't factor in the Diplomacy AI as should have been done. Instead Diplomacy was scored as part of the Gameplay section only (it should be scored in both for various reasons).

Rich is working on a rewrite to the GalCiv review's Intelligence & Difficulty section based on what we now know. This will be published sometime on Saturday (4/5)."

So two points. Point one being that the point Brad and many players were making had nothing to do really with how much the reviewer played.. or how skilled (or unskilled) he was at this game, but rather that he didn't bother to follow up and ask the devs about the game 'cheating' when they felt this was the case. They were wrong for assuming and posting the article without asking.

And second.. with a revamped AI section, the game should get a proper 4+ star rating on Avault, like it should have all along. I'm hoping for a 4.5.. LOL

Anyways.. kudos for fans for sticking up for this game, and to Brad for sticking to his guns. Avault is a well respected site, and a corrected score may go well to spreading the joy of a game that is GalCiv.

fleabitfox

W00t!

- Ash

~SDC~

I don't think this will change the overall score. But fringers are crossed.

Though either way, the main thing was getting the "AI is cheating" at low levels thing out of the way.

Ultimately, in many respects, GalCiv is a just a big graphical eye candy thing wrapped around 6 AI engines. If the AI isn't any good, then there's not much point to play the game.



~SDC~

Well, I thought Brad defended himself quite intelligently. It's obvious that the Rich couldn't back up what he said, and instead chose to fight everything that we idiot readers keep thinking he said (sigh).

I'm glad that there was the savegame transaction to prove them wrong . I'm sure I'm guilty of some "fanboy" bias, but I'm a bit urked by anyone who says that the galciv AI isn't any good.

To be honest, if it were *any* other site, I wouldn't have posted. I tend to take the view that it's bad taste for developers to get into public spats about reviews.

But since I have been writing for Avault for 2 years, I had a pretty high expectation that every step would be taken to make sure the review was fair.

While I can't say that I'm completely happy with the final result, I think Brian has done more than most editors would do by re-examining the issue.

I find the comments surrounding the savegame transaction quite fascinating, actually. To have someone outside the game describe the diplomatic manuevers of the AI at such an early stage of the game as "a rather intense set of trades and diplomatic negotiations" is interesting, to me. I always ASSUME the AI doing stuff, but since it's hidden from the player, you can never really tell how MUCH it's doing. To have someone with such obvious reason to downplay the AI's accomplishments make that statement is quite a powerful confirmation of just how intricate and deep this AI is.

- Ash

~SDC~

Actually some of this is in the debug.err file. Not much of it but some of it.

The major players make heavy use of trading with the minor races. The Yor had two minor races right near by and had traded tech with them for money which it was using to build constructors. It built 2 constructors in 10 turns. Note that it wasn't 10 constructors in 10 turns. That's because it was taking 4 or so turns for the money from its trade agreement to add up to enough to pay for the lease.

That said, I truly don't think the AI cheating or not cheating is material. What matters to me, as a gamer, is whether the game is fun. And for me, fun means that the game plays by the same RULES. Hence, if it took giving the AI a 25% economic bonus even at the lowest leels to make the AI competitive, I would have done so. The only reason why it doesn't get bonuses is because it wasn't necessary to. Most humans just are incredibly inefficient with their resources and don't make use of trade very well. But if it had required cheating with money bonuses to make the AI more fun, I would have done it in a heart beat.




~SDC~

"What matters to me, as a gamer, is whether the game is fun."

Of course. Why I find the AI not cheating thing to be such a big deal is that as a fan of the art/science of games, the fact that you've created an AI that can be competitive and fun WITHOUT cheating is a tremendous accomplishment. It's not that GalCiv is more fun than, say, the Civilization series; it's that I know you've achieved the same amount of fun through less nefarious means.

You COULD have given the AI economic bonuses, but the fact that you didn't need to do that seems, at least to my relatively uninformed mind, to be a landmark technical achievement.

- Ash

~SDC~

So I'm wondering if Gamerankings will update the review if the final score changes.

~SDC~

I'm sure they will if someone tells them it changed...

- Ash

~SDC~

@Frogboy: Well, GalCiv IS more fun than Civ ever was! And speaking of Civ (esp. CivIII): This has to be the Great Grandmaster of cheating AI's. Man, Civ's AI uses cheats GALORE on high difficulty levels!

~SDC~

debug.err

Hmm...there's some interesting stuff in there, such as:

Debug Message: AI: Drengin Empire blackmailed Torian Confederation for $ 1316!

Hehe, cool.

~SDC~

Wow. What fireworks.

Normally I'd be pretty irritated if a developer had a public fight with a reviewer. I'd simply ignore them both, and the game.

But... I've been programming (commercially) since 1984 (whoo, I'm dating myself). And I wrote a couple of game, although none of them got published.
I do recall writing a very simple strategy game, and struggling with the AI. I didn't want it to cheat, and spent a LOT of time reading through research papers, testing ideas, fiddling about... and failing to create something challenging in the end.

So if somebody writes such a sublime AI as Frogboy did, which is kicking my teeth in at higher levels, and it does so without cheating - well, that's astonishing.
I can fully understand why he's so upset, and I admire his patience for even answering that pompous, unpolite reviewer.

Go, Brad! And let's have a look at the updated review later tonight