AI Teching Too Slow? It May Be Spying

[ratrangerm]

I've noticed a couple of people have observed that the AI's tech rate is too slow, but I think there's a reason for that... the espionage slider.

In several games I've played just to get the feel for BtS, I've noticed that, in the early part of the game, I would be in contact with four civs, and manywould have a higher ratio of espionage points on me, despite the fact every civ was in contact with everyone else. You only get four espionage points from the palace, so how is it possible that some AI civs would have a higher ratio of points on you... and a pretty high ratio at that (more than 4 to 1)?

The answer: The espionage slider. Certain AI civs are going to put more of their money into espionage early, and to do that, those civs have to take away from their science. There's no other way.

As an example, one game had Tokugawa, Genghis Khan, Justinian and Hatshepsut as neighbors. I put one espionage point toward each of them, and I knew all of them had contact with each other because their units were often found next to each other as I explored.

But Toku and Genghis had much higher ratios on me than Justinian and Hatshepsut... in Toku's case, it was more than 4 to 1. I concluded that Toku must have been using the slider to accumulate additional espionage points. Which, of course, means he puts less into tech and, therefore, he's going to tech slower.

If there is something in the game that keeps the AI from teching too quickly (and from my experience, I tend to doubt that), it's probably because the programmers were trying to balance out the espionage feature. Don't forget that the AI gets it as well, and why should the AI get an easier time of deciding how important espionage is to it while getting a boost in tech?

As I play games, I plan to take a closer look at the espionage feature. I had not been a fan of spies from past games, but the new espionage system, I have found, is more useful than ones in past games, so I'll be looking at it as I play to figure out how to best utilize the system to one's advantage, and figure out which civs tend to favor espionage more than others do.

 

[InvisibleStalke]

Unfortunately generating lots of EP and just annoying your neighbours with the occasional sabotage isn't going to compete with investing in tech, so any AI that does this is going to suffer.

I can't believe they do it that much though - I am usually out spying the AI without using the espionage slider at all - due to prioritizing courthouses and other similar buildings and running the occasional spy specialist. I think its probably a combination of espionage and not being able to successfully manage their early expansion maintenance costs.

 

[Elrohir]

That's probably why the AI techs so slowly now. I play on Monarch, and I'm actually considering going up a level, because the AI is teching so slowly now - in every game I've played so far, I've gained a tech lead before Liberalism. Every one. I used to only be able to do that about half the time in Warlords.

Honestly, the AI needs to use espionage more intelligently. (Especially since with just a few specialists and the proper buildings, it's possible to have espionage leads on ALL the AI at the same time without using the slider much) Sacrificing science for the sake of spies is not the answer.

 

[uncarved block]

Yeah, it sure looks like Espionage is a big part of the explanation for the slower tech rate. If you look at the EP graph, you can really pick it out after a hundred turns or so. If the program is being fair as well, you have to believe that they're trying to gain EPs for all of their neighbors, and not just the human player-- even on a Standard size map, that's a sizable investment. Tack on the fact that the most aggressive AIs when it comes to spying are also unit spammers, and I can see a lot of games where my Riflemen are fighting hordes of Macemen on the way to taking down AI cities.

It's not just the slider, either. In may last game, there were patches of turns where it seemed like I was catching a Spy every turn for five, ten turns at a whack. That's a lot of hammers, especially since over half those Spies were caught around the same border city . . .

What's the answer, though? If the AI tones down spying too much, it will create another exploit for the human player, and lead to even more bone-headed decisions as the AI makes choices without access to important info like comparative Power and Production. Tough choices, that's for sure.

 

[futurehermit]

I think the AI also does a very poor job at setting up sufficient #s of commerce cities.

In warlords even financial civs didn't build enough specialized commerce cities. Too many farms, etc.

Early on techs are cheap, but as they get more expensive you either need more scientists (which the AI doesn't handle well) or cottages (which the AI doesn't build enough of)

We didn't notice this too much before because they got huge cheat-bonuses at higher levels. But now those bonuses have been removed and the AI NEEDS to get cottages down in numbers to compete. And my hunch is that they are not.

 

[Snaaty]

I think the AI also does a very poor job at setting up sufficient #s of commerce cities.

In warlords even financial civs didn't build enough specialized commerce cities. Too many farms, etc.

With the latest patch, the AI does surprisingly well in warlords... ...even the warmonger cottage up most of their empire during peacetime... ...which means that even civs like Mongolia can compete in endgame

So I think it´s mainly a BtS problem... ...simply too many options for the poor AI who wasn´t innitially designed for that

 

[Unconquered Sun]

Maybe it is the other way around - human tech too fast?

Perhaps Firaxis expected us to alot a steady % to espionage...which noone is really doing. If so the fault was in betatesting.

 

[madscientist]

I am happy to move from a warlords Prince to a BTS Monarch. Bit for the really good players who find the AI teching too slow why don't they simply make a rule to run the espionage slider ALWAYS at something like 20%. Seams like one of the simpler fixes.

 

[LlamaCat]

I agree the tech pace is due to the espionage slider, I'm 100% convinced of this in my own experiments or games, which are easy to set up the way ratrangerm has done. you just have a limited group of AIs that have all met each other and the key here is to examine at the very beginning of the game, before courthouses are available.

I think they made a mistake in coding the AIs to move up their EP slider, especially so early. maybe certain personalities are more prone to do it than others? I would be interested in seeing some eventual War Academy articles with detailed results.

you will also notice that if you have only met one AI and they are actually behind you in EPs, it probably means they have met other AIs already that you haven't met. that's the only reason really why they would be behind you because they normally seem to get ahead.

but of course it's easy to catch up in EPs later with courthouses and other means, since humans will outplay the computer in this fashion. but I generally don't move up the EP slider because I have yet to find a need to do it to win.

 

[futurehermit]

Either espionage needs to be more effective justifying spending $ on the slider or else the AIs need to it much less often.

Also, I am playing a game now where I am going pure food/production with 0 cottages and am comfortably nestled in amongst the tech leaders without anyone being too far ahead or behind. It also allows me to keep up with their production potentials making warfare easier. It's still disorienting seeing opponents without longbows at 1200AD let alone liberalism, but I figure what the heck as long as they aren't too far ahead and I'm not too far ahead it seems like a balanced game.

Playing with cottages, especially with a financial civ (I love the new financial civs: dutch, mayans, etc. ) just makes teching too easy imo.

 

[MrCynical]

There are a couple of civs (e.g. Mansa Musa) that tech at a good pace still, but most of them seem too weak to be accounted for just by the espionage slider. I rarely move it from zero, and I seem to be generating comparable amounts of EPs. Might be that I prioritise courthouses more than the AI, but even so they can't be spending that much on espionage.

My suspiscion is still that it's the ludicrous numbers of (usually outdated) units that most AI's build. The support costs must be very significant, and it would fit with the fact that it takes a while for the AI tech pace to really fall off - it's the time till the go way past their free unit cap. It can certainly be amusing to tip these AI's into Pacifism one way or another and watch their economy crash, but possibly they need a little work on this.

 

[ratrangerm]

I do think it tends to vary by the civ. In my latest offline game, one that I'm trying to play until the end (but with the bugs people are reporting, who knows if I'll make it to the end), I had Ragnar, Hannibal and Catherine as neighbors, with Churchill nearby on a smaller continent (map style was big and small).

Ragnar was clearly putting a lot into his espionage slider... his EP ratio was higher against me. Churchill, on the other hand, was teching pretty well... he beat me to the Great Library, and when I researched Paper, I could see his research and he was able to get it in the same amount of turns I did. And I had a good chunk of my cities cottaged up.

Cathy seemed to focus her EP mostly on me, but over time, my ratio went past hers and she was teching pretty well, until I decide to invade her. And Hannibal got a good EP ratio lead on me, but was behind in tech, although he's starting to catch up now thanks to his Financial trait.

It definitely seems that Aggressive civs like to up their espionage early. Leaders who are more likely to be peaceful, though, would rather tech and aim for a Cultural or Space victory.