Unpopular opinion: people that blame AI don't play on deity

When I was young toy grabber machines came out where you had to get a toy with the grabber. My dad explained how they are programmed to only grab once they have made enough money and I was shocked. There was a program about that and slot machines a few years ago, blatant money making machines.

If anyone thinks online poker is programmed fairly and does not manipulate the player, I have some bad news.
Even candy crush is highly manipulative to give the user a ‘better’ experience and encourage them to pay.
In the case of Firaxis it may be as simple as do not build rockets until T250 or There is less ‘encouragement’ to build cities beyond certain numbers.

I feel it is credible.
I feel as if many people are too caught up with the idea that companies are entirely driven by their consumers, however, in the case of AI I doubt there's enough going on there to put them in a position to make decisions like that. Maybe going out of their way to allocate the wrong type of people to it would better fit the state of it.

And yes it is possible that they prefer that the AI stays terrible. Ads added to the Civ5 launcher and the unlikely release of the DLL this time around shows their desire to make people constantly move on to their new products well enough.
A decent AI would counteract a content delivery business model as it may keep people interested in the previous type of content since the AI can actually play with it in a way that fits their content preference. Balanced and perfectly functional multiplayer can also keep people in previous iterations, but just as we know those two things don't exist in Civ multipayer, we know that multiplayer is simply to be expected. Decent AI on the other hand...
 
Last edited:
  • Like
Reactions: cvb
Do neural nets and game AI use the graphics processing unit over the central processing unit? I would have thought that the GPU was just for the visuals.

Neural nets are resource-eating monsters that need GPU to fulfill their calculation resource requirements.
 
Hope we can see positive signals towards DLL sources soon. "Soon!" ;)
Civ6 DLLs will never be released. They've made huge mistake releasing Civ5 dlls. Firaxis/2K will not do such mistake again.
 
This thread is still going eh...

At deity, we want the AI to be smarter and not as easy to exploit. This is much more preferrable to massive bonuses which skew the early game.
 
Count me in among the people who necessarily don't want better AI just for the challenge alone (though that would be a big and important consequence of it).
I would like AI that:
-Can actually accomplish things if it wars against me in the late game and not just send its units randomly across my territory
-Doesn't settle a line of *really* crappy cities right towards my territory when it has a *huuuge* swath of great land the other way all for itself, including two amazing locations next to a wonder (a bit specific but this just happened in my game).
-Doesn't eat up city-states as soon as it sees them (yes, I'm aware that they're changing some of this, we'll see how it actually works)
-Isn't a war-hungry monster in the ancient era so that you can almost count on it declaring on you if you encounter it fairly early on.
-Doesn't do weird stuff like avoid taking a city even if's sieged, surrounded and could easily be taken in a turn.
-Other smaller things like say "hey, don't settle near my cities" when it just foward-settled you horribly or say "hey, move your troops" when you have a scout in its land and the AI is sitting outside your borders with a million units.
-Doesn't settle *obviously* stupid cities that will loyalty flip right away. I mean, I can sorta appreciate it if it settles questionable cities sometimes, that can make the game more fun and dynamic IMO but when it settles locations where it's just impossible to defend...
-Doesn't build stuff like the Petra when it has one tile of desert in the city (which the Petra will occupy)
-Doesn't recruit Moksha if they are Kongo, even before another religion has even been spread to them (again, specific but it's just a good representation of the AI).

There are so many smaller niggles and oddities with the AI that I would want to be fixed. It's more about "I wish I could make it through a game without several instances of sigh-worthy behaviour from the AI" rather than "I want the most challenging AI ever".
I mean, I would *like* more challenge in certain aspects of the game as well. But most of all I would like less ".... what on earth are you doing...?" moments.
 
I would like to see a competitive IronMan game mode as a cheap solution to the lack of challenge for top end players. Since more investment in AI probably doesn't make enough ROI for the publishers.
A no-save mode like in X-com, where your win/loss ratio is tracked and linked to a competitive leader board in-game or on a website a la POE or Diablo leagues.
Since this is 'competitive' there must be fixed rules/settings for everyone.. and no mods obviously:cry:
Here is my take:
-Deity difficulty
-Retiring a game counts as a loss. Restarting a game counts as a loss (no fishing for ideal starts)
-Random leader
-Standard size/standard speed (all minor settings standard)
-Map type cycles between continents/pangaea/fractal/Island plates (order of each block randomised).

With these conditions I wonder if everyone will still be beating deity with their eyes closed and maintaining perfect win streaks? :lol:
Personally I would love something like this, anyone else? I am addicted to save scumming, I wish I didn't but I just can't stop myself.
 
... but why would that be a mistake ?
Releasing DLLs prolonged civ5 life cycle and created direct competitor for civ6. If Firaxis has plans for civ7 (I am sure they have) they will not release DLLs for civ6.
 
-Doesn't settle a line of *really* crappy cities right towards my territory when it has a *huuuge* swath of great land the other way all for itself, including two amazing locations next to a wonder (a bit specific but this just happened in my game).
R&F added a PlotEvalConditon called "Cultural Pressure", used as follows:
Code:
       <Row ListType="StandardSettlePlot" Item="Cultural Pressure" Favored="false" Value="1"/>
       <Row ListType="StandardSettlePlot" Item="Cultural Pressure" Favored="true" Value="-6"/>
I suppose the idea, in theory, was to boost score for plots that are free of the pressure and nerf the plots that have pressure. So, basically, prevent the behavior that you are describing.
I am pretty sure however that only the first condition is actually working (favored=false) and the second is ignored. The game provides partial information about plot scoring so it is possible to test it.
The above is independent from leader's affinity to do forward settling, 'cos some could do that anyway.
 
  • Like
Reactions: cvb
The AI is still super dumb at deity level. It just has a ton of bonus to compensate for its stupidity. It's like playing chess against a 4-year old kid that starts with 3 queens.
 
Releasing DLLs prolonged civ5 life cycle and created direct competitor for civ6. If Firaxis has plans for civ7 (I am sure they have) they will not release DLLs for civ6.

It still seems irresponsible to me. If they are unwilling to improve upon the AI, then they have a responsibility to release the code so others can do so.
 
Unpopular opinion - huge cheating bonuses are no substitute for an actual AI.

That's a pretty popular opinion, but an unfeasible one. I do wonder if the people making this complaint have played any other 4xes. I'm playing Stellaris at the moment. Even Civ VI's AI makes the Stellaris one look as dumb as a brick, and that's a mechanically simpler game. Every Civ game has used bonuses rather than differences in the AI to scale difficulty. Even the classic Master of Orion is unplayably easy because the AI has no idea how to design ships and its tech progression is appallingly slow (unmodded, that is).

The problem is not with the concept, it's with the execution. Civ VI, and to a lesser extent Civ V, is simply not capable of making effective use of its bonuses - so it gets big bonuses but the game barely becomes more difficult. The fundamental problem it comes back to is that Civ VI is simply too easy on all levels - if it is made appropriately challenging at the highest levels, why does it matter whether the game gets there through AI improvements or bonuses?

People who complain that the AI is 'cheating' sound as though they're complaining it's artificially difficult - if you find that to be the case, your issue is different from mine and I think from most people who are complaining about the AI. Our issue is entirely that the game is not difficult enough, 'cheating' or no 'cheating'.

I would like to see a competitive IronMan game mode as a cheap solution to the lack of challenge for top end players.

Do many people reload old saves if things go wrong? I essentially play almost every game as though it were Ironman - I just don't find it interesting reloading things I've seen before to make different decisions, and generally only ever do so if something goes wrong due to a bug.

-Deity difficulty
-Retiring a game counts as a loss. Restarting a game counts as a loss (no fishing for ideal starts)
-Random leader
-Standard size/standard speed (all minor settings standard)
-Map type cycles between continents/pangaea/fractal/Island plates (order of each block randomised).

Sounds essentially like my default - though I favour shuffle maps over fractal (which seem to produce pangeas more than anything else in my experience) and I always set map size to Huge in the absence of a random map size option (which I eternally hope will eventually be implemented). Does anyone have any clear idea whether difficulty scales linearly with map size, or whether standard may actually be more difficult than Huge (due perhaps to fewer AIs getting in one another's way, and because culture and religious victories are easier with fewer civs)?

Releasing DLLs prolonged civ5 life cycle and created direct competitor for civ6. If Firaxis has plans for civ7 (I am sure they have) they will not release DLLs for civ6.

Sorry, but I don't buy this at all. Civ V remains popular because it's a good game, but it hasn't substantially impacted Civ VI sales - Steam top sellers/played lists indicate that Civ VI is one of the platform's best-selling games of 2018 and has overtaken Civ V in players.

A wide variety of factors extend Civ V's appeal - Civ VI was frankly an atrocious game until the release of the last expansion, it lacks a diplomatic victory feature, its AI remains worse in the base game (so completely ignoring DLL-derived mods), and Civ V has been on sale frequently. Even though most of the current Civ V players seem to be people who have had the game for a while (its sale figures are much lower than its play figures for 2018), it still sold well enough to get a 'bronze' ranking on Steam, so it's still accruing new players.

You're drastically overestimating how many people are likely to use AI mods, I suspect - any of the above factors, and certainly all combined, are more likely to influence Civ V's popularity than the availability of the source code. And Firaxis is hardly likely to be unhappy with a situation where Civ VI is a top-seller and they're also still making money from new sales of Civ V.
 
Last edited:
Releasing DLLs prolonged civ5 life cycle and created direct competitor for civ6. If Firaxis has plans for civ7 (I am sure they have) they will not release DLLs for civ6.
I don't think that's the case (ie lost of income because civ5 is a big competitor to civ6), but only 2K can get those numbers.

From the general impression on the forum here, there were a lot of people not moving from civ4 to civ5 because they didn't like civ5 game's design and they could mod civ4 all they wanted with its source code, yet civ5 was a huge commercial success, and at the same time civ4 still brought them money thanks to the prolonged life cycle.

I bet it's the same with civ5 vs civ6, and that they do plan long term (ie still getting money from civ6 even after civ7 is out)
 
OP: noone is asking for an AI like stockfish on chess. Just one that is better then the current one.
Which is much like a toddler who has first been given a pc mouse for the first time (without bonusses).

I have not read through all 7 pages, just the first one. I still think civ 6 will get 3 expansions in its life cycle.
I think firaxis will upgrade the current AI for second expansion accordingly to make us of its new features.
And not add much else, Perhaps add a decent aircraft overhawl as talked about on one of the later lifestreams.
I think this is what is going to happen.

What i am more excited about is if modders could be given the tools to upgrade the AI by firaxis. So a community can form for fixing the AI with better tools. DLL code.

Civ 5 AI with mods is a lot better, do i have only played civ 5 for like 1200 hours. I continued on civ 4 for years after civ 5’s launch.
 
Releasing DLLs prolonged civ5 life cycle and created direct competitor for civ6. If Firaxis has plans for civ7 (I am sure they have) they will not release DLLs for civ6.

We cannot just disregard plain incompetence as a factor, that would not be realistic. Let's be honest: they never could replace Soren Johnson, period. In that sense, if this is a factor, then releasing the dll for modders to show what their engine can really do puts them to shame. Nobody likes that; most people react bettering their performance, but some people react by hiding the evidence.

That's a pretty popular opinion, but an unfeasible one. I do wonder if the people making this complaint have played any other 4xes. I'm playing Stellaris at the moment. Even Civ VI's AI makes the Stellaris one look as dumb as a brick, and that's a mechanically simpler game.
...
Does anyone have any clear idea whether difficulty scales linearly with map size, or whether standard may actually be more difficult than Huge (due perhaps to fewer AIs getting in one another's way, and because culture and religious victories are easier with fewer civs)?

Two good points. Let's be fair, Stellaris 2.2 messed something up with the AI, that is not the "normal" Stellaris AI which provided a good challenge before. When they fix what they messed with the patch, we can judge it.

As for the influence of settings in difficulty, I think it's a combination of factors in either direction; i.e. map size regarding one of the biggest equalizers of the game, espionage. The bigger the map, the harder it is to catch up or disrupt other civs due to the fixed maximum number of spies in game. OTOH, the bigger the map, the less contention in space between civs, thus the easier to establish a good baseline empire.

What is obvious to me, and more than in any previous iteration, is that the game is balanced (and barely for that matter) only for Standard size and speed.
 
We cannot just disregard plain incompetence as a factor, that would not be realistic. Let's be honest: they never could replace Soren Johnson, period. In that sense, if this is a factor, then releasing the dll for modders to show what their engine can really do puts them to shame. Nobody likes that; most people react bettering their performance, but some people react by hiding the evidence.
That's a hilarious conspiracy theory that can be debunked by basically any professional programmer's Twitter feed.

Not all code is nice.
Not all code is clean.
Not all code is as fast as it could be in a theoretical ideal world with an in-house, modern set of tools and no deadlines.

The necessity of DLLs should be relegated to the past anyway. Ideally, more companies should invest in creating or making internal tools accessible for their fans. The main blockers to this are a) using licensed tools (costs money, can't distribute), b) time on skilled programmers that could be doing literally 1,001 other things and c) pushback from legal and / or other bodies relating to the creation process (and IP rights). For example, Relic Entertainment historically have worked very closely with Games Workshop, and have said on more than one occasion that GW directly limits what can be done when it comes to modding their (40k-related) titles.
 
That's a hilarious conspiracy theory that can be debunked by basically any professional programmer's Twitter feed.

Or is it?

Twitter feeds are now proof of competence as opposed to cold, observable end product performance? :crazyeye: I would say THAT is hilarious.
 
Or is it?

Twitter feeds are now proof of competence as opposed to cold, observable end product performance? :crazyeye: I would say THAT is hilarious.
You're talking about the personal reaction of people feeling ashamed or upstaged. You're literally inventing a theory that the DLL won't be shared because the programmers at Firaxis (rather, than, say, the legal team at 2K, who own everything Firaxis do) will be jealous that other people can do a better job.

I've got a decent Tweet for this occasion, actually:

https://twitter.com/uxresearch/status/1079148047257395200
 
  • Like
Reactions: cvb
You're talking about the personal reaction of people feeling ashamed or upstaged. You're literally inventing a theory that the DLL won't be shared because the programmers at Firaxis (rather, than, say, the legal team at 2K, who own everything Firaxis do) will be jealous that other people can do a better job.

Nope, I am not inventing anything. I am just stating that nothing can be ruled out, including the possibility that incompetence is a factor and that such factor may influence the willingness of some to share tools that could amplify the observability of said possible incompetence.

The only way that incompetence as a factor can be ruled out is for someone to clearly show that, in fact, incompetence with regards to AI programming in civ 6 is an impossibility. Twitter feeds will not prove that, not to me at least, especially when I have "only" their final product to judge.
 
You're talking about the personal reaction of people feeling ashamed or upstaged. You're literally inventing a theory that the DLL won't be shared because the programmers at Firaxis (rather, than, say, the legal team at 2K, who own everything Firaxis do) will be jealous that other people can do a better job.
I would not rule out this as a factor. Cutting angles is probably common practice in todays games industry. Take Endless Legend. Some guy just brutally decompiled its DLLs. Well... it came out Victory Quest is not actually solved by AI, but simply resolved by random dice roll. Wow... And this is only one of many "shortcuts". One could only imagine what we would find in Civ6 core files.
 
Top Bottom