March 16th, 2018, 00:18
(This post was last modified: March 16th, 2018, 00:20 by haphazard1.)
Posts: 5,606
Threads: 47
Joined: Mar 2007
"Stupidest possible bug" is being rather unkind to Firaxis. Sure, it is a tiny thing with an outsized impact, and the effects should have been spotted and corrected. But I have seen stupider bugs. Heck, I have created stupider bugs during my career.  I expect just about anyone who has spent any significant time in software development has...although one would hope they did not slip through to production release, as this one did. But stuff like this is incredibly easy to have happen and to miss, particularly when two interacting components (the AI code and the leader info files) are being created by different people at different times, as I suspect was the situation in this case.
Thorough, carefully constructed QA is very, very important. And it rarely gets enough emphasis or resources.
Sorry for the rant, but my "have some sympathy for the over-worked software devs with a thousand tasks to finish by ship date" button got triggered. Back to hoping this will make a difference with the AI.
March 16th, 2018, 10:26
(This post was last modified: March 16th, 2018, 10:27 by RFS-81.)
Posts: 851
Threads: 22
Joined: Aug 2011
I agree that it would be wrong to call the developer who made this typo stupid. But even smart programmers can produce stupid bugs. If we want to play the blame game, I blame lack of QA, and low expectations for game AIs.
EDIT: I'm not calling the testers stupid either, there were probably too few, with too little time.
Posts: 9,706
Threads: 69
Joined: Dec 2010
I think you are being too much optimistic about the impact of this change. Civ 6's AI problem is not that it doesn't value its tile yields like it should, but the fact that they simply don't know how to win the game.
March 16th, 2018, 11:32
(This post was last modified: March 16th, 2018, 11:33 by haphazard1.)
Posts: 5,606
Threads: 47
Joined: Mar 2007
(March 16th, 2018, 10:26)RFS-81 Wrote: EDIT: I'm not calling the testers stupid either, there were probably too few, with too little time.
There are ALWAYS too few testers, and too little time allocated for testing. Always. It is inherent in the development process, since the devs keep changing and fixing things right up to the ship date, which keeps requiring more testing, cycle, repeat.
Why yes, this is a bit of a sore point with me. 
Anyway, I am afraid Ichabod is correct and this will have minimal impact at best on the AI's performance. But hopefully it will help at least a little.
Posts: 5,357
Threads: 53
Joined: Oct 2010
So, now that the issue is known, did they fix it yet? or are they going to wait 3 months until the next patch?
March 19th, 2018, 08:32
(This post was last modified: March 19th, 2018, 08:35 by Rowain.)
Posts: 8,244
Threads: 30
Joined: Jun 2004
(March 16th, 2018, 10:49)Ichabod Wrote: I think you are being too much optimistic about the impact of this change. Civ 6's AI problem is not that it doesn't value its tile yields like it should, but the fact that they simply don't know how to win the game.
Thats nothing new in the civ-series.
No civ-AI no matter what civversion does really know how to win the game. They can win if you give them enough time but a lot depends on luck for them. But it was never an intention to create an AI that can really consistently win.
EDIT. The bug of course needs to get fixed but I wonder if there won't be alot of unhappy people if the AI suddenly acts a lot smarter than before.
Posts: 174
Threads: 10
Joined: Apr 2013
RFS-81 Wrote:That's almost as funny as the urban legend about Gandhi going crazy with nukes because of an integer wrap-around. unsigned integer. the agression factor in civ1 is an unsigned 8 bit value. India has an agression factor of 1, while adopting Democracy substracts 2 from the ag factor.
btw, Jagged Alliance 2 also suffers from an overflow issue - if the player chooses double stealth promotion during character creation and afterwards increases the agility attribute of the char, then the stealth attribute will overflow
me on civfanatics.com
An ideal strategy game would tone down efficiency challenges, while promoting choices and conflicts
No gods or kings. Only Man.
Posts: 3,750
Threads: 13
Joined: Dec 2016
Looks like there's another typo-related bug in the AI. Apparently the AI's "Medieval Strategy" requires the Classical Age as a condition, so it follows a blend of its Classical and Medieval age strategies during the Classical Era and then doesn't have a strategy throughout the Medieval Age.
https://forums.civfanatics.com/threads/a...or.630469/
March 20th, 2018, 14:55
(This post was last modified: March 20th, 2018, 14:58 by RFS-81.)
Posts: 851
Threads: 22
Joined: Aug 2011
Oh, so it does have a strategy prior to the Medieval Era? I hardly noticed! More seriously, could that be the reason why they keep around medieval units long past their expiration date?
EDIT: Eh, this seems to be about priorities for faith, gold, production, etc. again.
@Krill: Speaking of unsigned integer bugs: In MOO1, AIs could somehow lose more ships than they have...and those numbers have a few more bits than 8. I've never seen that myself, though. Maybe it was fixed in an official or unofficial patch.
Posts: 5,606
Threads: 47
Joined: Mar 2007
(March 20th, 2018, 14:55)RFS-81 Wrote: @Krill: Speaking of unsigned integer bugs: In MOO1, AIs could somehow lose more ships than they have...and those numbers have a few more bits than 8. I've never seen that myself, though. Maybe it was fixed in an official or unofficial patch.
It has been a while since I have seen it, so I think it was fixed in one of the patches. But I have had the "negative fleet bug" happen to me. If I remember correctly, it was tied somehow to the AI scrapping ships; they would end up with a fleet of 32000  ships due to the wraparound. Instant completely unstoppable death fleet, unless the design just happens to be something terribly obsolete.
|
