Surprise: AI research rates changed in V1.29

[cracker]

This afternoon Soren Johnson revealed that:

quote:

"another change I should mention is that tech cost is now constant across difficulty levels, so Deity so no longer be (quite) the tech race it has been [...]let me rephrase. In 1.21 and earlier, a tech cost the AI 6 and the human 10 under Deity. With 1.29, techs cost the AI 10 and the human ~15 under Deity."

Instead of exposing the multiple AI cost factors in the editor to allow the users a bit of control over these features, Soren decided to include this undocumented change in the V1.29 release in a further attempt to conduct unannounced human behavior experiments on the million plus users that have purchased Civ3.

I definately agree that the tech research rate for the AIs may have needed to be adjusted but this approach deserves public flogging.

The two biggest problems with AI inequality at the various difficulty levels are the tech research rate and then the Great Wonder production costs. Instead of fixing these and at least revealing the code controls to adjust these factors in the editor, Soren has just adjusted the tech research rate for ever AI in every game and then left this change undocumented for us to stumble across the difference in another "skinner box" experience.

This is also one of the most significant changes in the whole damned V1.29 patch since it effects every game and every scenario that could possibly be played. Yet, Soren's choice was to leave this little tidbit out of the Readme documentation AND to continue to hide the controls that might allow other people to verify potential fixes that could counteract any performance issues with the change.

I would like a published list of the unannounced changes to the game engine so that we can have at least some clue of why the game environment is changing. Some of these changes may be good but based on experience some of the changes will have unforeseen side effects.

 

[Bamspeedy]

"another change I should mention is that tech cost is now constant across difficulty levels, so Deity so no longer be (quite) the tech race it has been [...]let me rephrase. In 1.21 and earlier, a tech cost the AI 6 and the human 10 under Deity. With 1.29, techs cost the AI 10 and the human ~15 under Deity."

So, is this saying that no matter what difficulty level it is the AI researches at the same pace? Of course, though on deity libraries, universities, etc will be built faster and with faster growth it will have a little faster research. At least this would solve problems of players not getting to modern era on chieftain as tech costs for the human would be 5 instead of 10, am I correct on this?

Yes, these changes should be well documented! And they should be adjustable in the editor! What if this change had made tech researching a farce like it did in a previous patch (v1.17?), where it was useless to do your own research, and so much smarter just to buy them because of the tech devaluation as more civs learned that tech. If this was in the editor, we could fix it, but since this is the 'last patch' we would be screwed forever if something unbalancing like that happened.

 

[sumthinelse]

Originally posted by cracker
tech cost is now constant across difficulty levels, With 1.29, techs cost the AI 10 and the human ~15 under Deity."

The default "cost factor" for deity in civ3mod.bic is still 6. Does this mean the game ignores whatever you configure for the "cost factor"?

Why didn't he just change the defaults? Is it possible that he intended to change the defaults but it didn't make it into the patch? Are we sure the game really functions the way he thinks it does?

 

[cracker]

Its a bit tricky to interpret the sorenspeak, but I read the part of his comment that says V1.29 Deity now has the "AI at 10 and the human at ~15" instead of 6 vs 10, as meaning the AI is held constant while the human cost factor is manipulated at each level using the AI cost advantage as the denominator of the equation. The advantage is the same, it is just that all the human costs are inflated at any difficulty level above Regent.

Why do I feel like I am the only one posting coherent threads about V1.29 that do not include massive ranting from people who cannot even play the current game effectively without modifying it to fisher-price level rules. Arrrrrrrrrrrrgh!

If this patch disrupts things in any way like V1.17, there will be blood in the water.

 

[gonzo_for_civ]

Why do I feel like I am the only one posting coherent threads about V1.29 that do not include massive ranting from people who cannot even play the current game effectively without modifying it to fisher-price level rules. Arrrrrrrrrrrrgh!

Because, quite frankly, you are.

Although I don't think this one should really be that big of a deal.

10:15 (2:3) is a better advantage for the human than 6:10

reduced: 2:3 to 2:3 1/3

 

[cracker]

Gonzo,

You missed the little doohickey (tilde) in the sorenspeak. I am taking it to mean "((approximately 15)/10) = 10/6"

Technically 10/6 = 16.6667

We may never know for sure since this area of the game seems to be exempted from any documentation requirements. We would have to tie Soren to a stake and torture him slowly in order to get intelligible answers. COme to think of it, that option keeps getting more attractive each time a patch is released with undocumented and unavoidable sorenisms embedded in the code.

 

[chiefpaco]

I agree with you. It was a big shock when revealed. To me, this could be the biggest change they made, definitely not part of a BTW. There are many who don't really care about the rules and who do not study the game every turn. To them, it may not matter.

There are also many who pour over the details and are miffed over any detail not fully explained in the manual or readmes. To these people, I'm not sure Firaxis owes an explanation every time - there are some details which are best left for discovery. However, when changing the game in patches, I think it is wrong to not mention it. If not mentioning it was a mistake, ok. If it was intentional to hide it, that seems wrong to me.

As for the fix itself, I think it sounds brilliant. It may interest me to start a Deity game sometime. It was the precise reason I despised Deity games so much.

 

[Yzman]

Wait, so tech on deity is now just as fast as say on monarch or emperor?

 

[Qitai]

Good change. This would slow down the tech rate dramatically in deity. Hmm... Looks like mydisease's science beaker formula has to be changed.

 

[sumthinelse]

This doesn't do anything about the fact that you have to mod the AI to AI tech rate to have any kind of tech balance in deity. Why are the defaults set so that you have to mod the game to make libraries and universities make sense for the human player (at least when you have 16 civs in the game)?

Some will like what this change does. It seems to me that it takes away some of the flexibility we had in 1.21f.

 

[cracker]

I do not play Deity much because it just is not enjoyable for the most part. The primary reason I dislike the level is the AI cost advantage of 6 being applied to wonders. In rare cases, I can build 1 great wonders in the ancient or middle ages while the dominant one or two AIs will build them all. The game does not spread wonders around on Diety very well.

If I wanted to be tortured for someone elses pleasure, I would just do something quick and easy like fail to write down the amount of a check in my wife's checkbook or leave the toilet seat up.

Back on this topic, the fact that Soren changed the tech research rate coding demonstrates that he had the time and the knowledge to expose the multiple AI cost factors (one for growth, one for production, one for tech research, and one for great wonder production). Instead of providing the controls to adjust and verify a solution that might work in scenarios, Soren just changed a piece of the code to fix the problem as he saw it and them made that solution the hardcoded default for every application. More of the same behavior.

 

[sumthinelse]

I agree Cracker. I will go so far as to say that we are mature enough to decide what values should be applied to the AI research rate. It does not have to be hard coded.

 

[rangers85]

Originally posted by Yzman
Wait, so tech on deity is now just as fast as say on monarch or emperor?

The tech race will progress rather the same. Instead of before where the human remained the same on all levels and the AI gets cheaper researching (or more expensive for chieftan & Warlord), now the AI stays the same on all levels and it's the player who gets the slower research on higher levels or faster research on chieftan & warlard

 

[Qitai]

Trading is not the main reason why Deity tech goes so fast. The above is more the reason. Trading occurs regardless of level and usually, the tech leaders will trade no matter what trade ratio you use (since they will get 2 equally valuable tech which would means they would trade).

The impact of this change would more or less even out the tech progress rate on all levels subjected to impact of growth and production rate, which is more acceptable.

 

[sumthinelse]

Originally posted by Qitai
Trading is not the main reason why Deity tech goes so fast. The above is more the reason.

Both actually do contribute to the AI's advantage on Deity. I really don't know which is more important. Now that Soren has hard-coded the heuristics we can't change them to find out.

Trading occurs regardless of level and usually, the tech leaders will trade no matter what trade ratio you use (since they will get 2 equally valuable tech which would means they would trade).

Qitai, you are one of the people who write the best posts on this forum. Your knowledge of the game is really masterful. However, I am about to disagree with you.

Qitai, have you actually compared the AI tech progress in a 1.21 game with the AI trade rate at 100 vs. a game with the AI trade rate at 160? In my experience, even on regent level, when you change the AI trade rate to 100 it makes a big difference. I could go into more detail here, but the point of this thread is Soren's change. (I'm getting there, Cracker )

It may be that you and I play a different kind of game. I don't care about high score, I like other things in the game. I think you like to let the AI research ahead of you and then buy techs from them, and therefore you want the AI to be ahead of the human is research. I don't play that kind of game. I wonder if Soren's hard-coding of the AI tech research rate is based on the assumption that everyone plays the way you do. If so, he is wrong.

I saved a game in 1.21f where it would take me 8 turns to research Education. When I loaded it in 1.29f, now it takes 13 turns. I will still play the game to see what happens. I am not quitting CIV3. But I don't understand why this has to be hard-coded and undocumented!

 

[Qitai]

Well, you are right. I never tried the AI trade ratio of 100. So, I have to admit that I have made an assumption that I have no knowledge of. I guess you are aware that I never played anything below emperor (yes, the first game I played is already deity and . . . . . loss).

And yes, I agree for more editable attributes. And I do hate the non-documentation. I wasn't suspecting anything when I install1.17 only to have to revert to 1.16. Afterthat, I am more cautious on installing patches.

On this change, I think it is good. However, they could make it better by making the advantage/disadvantage of different level more editable, including, splitting food/production/science advantages/disadvantages up. That would allow players to selectively give AIs advantage base on the type of game they want.

 

[sumthinelse]

Thanks Qitai.