MOO3 rebuttal

I liked Bruce’s article on how the loss of the imperial focus points changed the entire playing field in Master of Orion.

One of the challenges in a system like that, and probably a reason why it got taken out, is that games have to tell players when they are doing something intentionally. Otherwise, players might just think it’s bad AI.

To use Bruce’s article as an example:

On the planet where the player was planning to go to war with the Klackons, I, emperor, set a given planet to be military oriented in preparation. The viceroy, however, not being aware of my future plans, changes it to an amusement mark.

But for this to work, in the news report it needs to actually explicitly say:

“Xaleath Elkin, Viceroy of Calibus IV, has decided after much local pressure, to refocus the planet’s priorities away from military production to focus on trade instead. A spokesperson for Mr. Elkin said ‘With all the trade we’re doing with the near by Klackons, my constituents demanded we capitalize on our trade position’.”

To go further, IFUs could take the view that each viceroy has to keep its local population happy. That THEY are voted in every X turns and hence they are under local political pressures. As long as this is spelled out, players will be much more likely to see the IFUs as a game mechanic.

At the risk of making our game look bad (good thing no reviewers hang out around here…) we went very late into beta without realizing that the AI wasn’t actually giving feedback to the player on what it was doing and why.

Back in the December builds, GalCiv’s AI would detect your build up of ships and then attack. Well, to the player, the AI just randomly declared war.

So we began putting in AI messages to the player letting the player know that it KNEW what the human player is up to. Stuff like “Your starbase in sector 4, 1 is clearly a build up for war. Given that you attacked us just 3 years ago after doing something similar, we demand you hand over this base now or we will destroy you…”

It’s letting the player KNOW why the computer AI is doing something that is half the battle. Heck, as a coder, it pains me that the AI sees the player building up for an attack or trying to culturally dominate but has to roll a random() to see if it will actually do anything about it (based on intelligence) in the name of “fun”. ;)

Anyway, the point is, in games where AI is such a key factor, the game has to make sure the player knows why an AI player made a particular decision. Not only so that it knows that an action isn’t a bug or just randomness but because it makes the game feel more alive (IMO).

One of the things i’ve seen people say about MOO3 is that people don’t know why a computer player did something.

The IFU design would have required vastly more explanation by the game for AI behavior.

Phew, sorry to be so wordy!

Good point about the viceroy AI needing to inform you that it was “consciously” disobeying your orders, and not just buggy. I think it was Andrew Mayer who brought up a similar point in the thread Erik started. It was funny watching the official MOO3 forum Kremlinologists try to come up with explanations for the broken diplomacy AI when people first started reporting it. They would ask the person who reported the problem to list the diplomatic/military/economic conditions present at the time of the alleged bizarre AI behavior, and then if they couldn’t concoct some halfway-plausible explanation out of these, they’d accuse the player who reported it of missing something or just not paying attention. Or even better, that it was “obviously racial.”

Very true. No matter what the game is, there will always be people who will rationalize anything.

I posted here last week about early in the GC beta the AI players in the United Planets voted randomly (literally ulVote = rand()%5). But some people would come up with all sorts of reasons to explain the AI’s strange behavior.

That’s why I think having the AI explain why it thinks something is important. That why the player KNOWS that something was intentional and not just rand()%5. ;)

Players are still insisting the diplomacy AI isn’t broking but behaving as it should. They say it’s spies at work. That in itself is funny. “It’s the spies, I’m telling you, the spies! You can’t see them, but they’re there!”

Excellent point, Brad.

Games are about interactivity. Interactivity is about feedback. Without feedback, it might as well not happen. Without sufficient feedback, it might as well be random.

And God bless them all.
The design of Dogz was based entirely on that premise.

There’s a lot of power in that impulse, but you have to use it wisely and you have to focus it in the right places.

Your Power Pill

You guys are touching upon something I’ve always felt about computer strategy games: the player needs to know the underlying code. That is to say, the player needs to know all the formulas, all the numbers, and all the percentages involved in everything that happens. This is why board war- and strategy games are generally superior to computer war- and strategy games. In order to properly strategize, one needs to know all the factors involved, and in board games these things are all necessarily known to the players. In computer games, on the other hand, they’re hidden and kept secret. All we’re ever told in the manual is that such-and-such thing positively affects this, or that this other thing over here negatively affects that. We’re almost never told how much these things affect other things, or, more to the point, what the precise calculation is that’s used to reach the results we see in the game. This is a major shortcoming of computer games as they’re currently marketed, imo, and it’s why I really refuse the title “strategy game” (when I’m being literal and cranky) to most computer games. They’re simply not pure strategy games in the sense I grew up with.[/i]

Features like that cater only to the ultra-hardcore, and can be hard to justify spending limited development time on.

I disagree. That takes away from the mystery and unpredictability. It would be really abusive.

I think that the AI should communicate with the player, as it does in GalCiv. Lemme tell you, the AI is not happy when you start building heavily armored starbases in sectors it controls.

Not for me- I hated playing board games with people who would take half an hour to optimise an attack which would exactly make 4:1 odds. As if Zhukov counted every tank in his battles. People need enough information to make decisions, not enough to make it a cheesy number exercise.

That is to say, the player needs to know all the formulas, all the numbers, and all the percentages involved in everything that happens.

It depends on how much realism you’re going for - if you’re a commander in the military you don’t know that your M1A1’s do 120 points of damage with each shot - all you know is that they’re probably the best MBT in the world, with a good crew. You don’t know the crew’s Morale number, or the gunner’s accuracy rating. All that stuff is way out of your control.

Also the reason all those old table top games like SL and ASL were that way because there wasn’t anyway to hide the calculations from the player. I’d rather have a game that’s realistic then a game I can metagame into the ground.

I’m with Mudpuppy on this one. I like the fact that in Civ, I know if I can fill up all the wheat icons in a city’s “grow box” that the city will increase in size or that if I build the Colossus each of my squares will have an extra unit of trade. I like knowing that if my station in RR Tycoon has 4 villages in its radius, it will demand mail. I think the Sid Meier games work well because they follow a set of clear rules. Not all the players need to understand the rules to have fun, but for me at least, having rules I understand improves the game experience.

What I don’t like is games where the rules aren’t clear and I can’t really tell when something has an effect. For example, suppose you’ve switched from a long sword to a battle ax in an RPG. Did it make a difference? Is the ax faster? Does it do more damage? The game should tell you what the difference is between a long sword and a battle ax, and ideally it should tell you quantitatively. In some games this information is not presented, and I think that definitely makes the game less enjoyable.

As Kevin said, “Without sufficient feedback, it might as well be random”. For me if I can’t tell that the game is following a clear set of rules, I have no way of knowing whether or not my decisions are having an impact.

Features like that cater only to the ultra-hardcore, and can be hard to justify spending limited development time on.[/quote]

I’m not that convinced this really takes a lot more development time. The rules do need to be written into the games code in any case. And I think coming up with clear and simple rules (ala Sid Meier’s Civ rules) actually makes for better games). If the rules are actually included in the printed documentation, then sure, it will cost more money (and potentially may scare off some players). Presenting the internal rules in some form whether printed or not will incur some cost, however, if the developers are willing to discuss the rules used in an online forum, in most cases, I suspect some fanboy will be more than willing to put them together into a FAQ.

Strong agreement. I just wrote a long-ish post (in the Spoiler Cell thread) about that very point, as it applied to the design of the Thief games.

I have to agree here. I’ve resently played some Civ 3 since I needed a 4x fix badly and MOO 3 didn’t seem to deliver. IMO the game is great, but one of the things that bugs me is culture. It might have been implemented quite good, but it doesn’t give you much indication on when a city decides to defect. I can visit my city and see that the city will grow in 50 years, and that my fighter will be finished in 8 years, but the game neglects to inform me in any way that the city will defect to my rival the very next year! Frustrating and thank God for auto-save. Although it isn’t completly random, the lack of information makes it very hard to predict (in stark contrast, as you mentioned, to all the other easily accessible information).

I’ll echo this one too. When I look at screenshots from strategy games that I potentionally want to buy, I always look for stats. I need those numbers to figure out the game. One of the games that I have played recently that lacked stats in favor of text description was Medieval: Total War. What is Good Armor? What is a Strong Charge? What does two armor upgrades constitute, and is it better than upgrading to a higher unit type? Sure it might be more realistic, but its a game and I need those numbers to make my decisions. Strangely enough you had stats for the different leaders though.


KODP takes the other road here.

It does model IFP in that there are a very few things you can directly control for your clan. IIRC You control two things each season/turn and when you control the second one, time advances to the next season/turn.

They also kept a great deal of the game opaque from the user community and added in a good healthy dose of randomness. In small doses, It made things seem inexplicable, and that caused many people to give up on the game before the dose got large enough to hook the player.

On the whole, computer gamers didn’t seem to much care for those choices. It would have been interesting to see them used again.

With all this discussion of qualitative vs. quantitative nature of data presentation I’m beginning to think that it’s a damn fine measure of the fundamental difference between PC and Console titles.

But on either platform, beyond a group of hardcore addicts and programmers, few gamers want or need explicit access to the underlying models. What you’re arguing for here is something that would cause the games to sell significantly less, even if it makes you happy.

Abstracting data to the user in an effective way is something that has moved gaming forward into the mainstream over the last 20 years.
It’s fundamental to the success of every Sims game since Sim City.

Just to back up my argument here, I’ll point out that you can’t take the quantitative argument to it’s logical extreme, because then you’re no longer playing a game, you’re writing code.

One thing that Reiner Knizia does very well in his boardgames is create an abstraction model that is information rich. It’s also dead dog simple. I’d argue that the type of games you favor would do better to follow that model.

Your Power Pill

It depends on how much realism you’re going for - if you’re a commander in the military you don’t know that your M1A1’s do 120 points of damage with each shot - all you know is that they’re probably the best MBT in the world, with a good crew. You don’t know the crew’s Morale number, or the gunner’s accuracy rating. All that stuff is way out of your control.

No you don’t know damage points. But you know a lot more than just that you have a good tank. You know that in general one APDS round will destroy any armored unit it hits which isn’t a tank, and you know for which enemy tanks a hit on the turret may not in fact penetrate, and so on and so on. That kind of data may or may not be quantified in probabilities for a tank commander, but it’s there.

Somehow that knowledge should be conveyed to a player in a strategy game. Damage point and armor numbers may or may not be appropriate for a given game, but players should have more basis for their decisions than semi-mystical induction based on observed behavior, which as has been noted tends to result in utterly incorrect inductions due to random fluctuations of game RNGs.

While I agree that is true for the sim type game I don’t believe the same can be said for the strategy game. Some else up there said, and I agree, that the details are needed in order to make critical decisions for the strategy genre.

I thought Bruce’s article was excellent and very thought provoking.
I canceled my MOO3 Pre-order so I’ll probably never know what a mess the game was. However, I’d like to put in a good word for IFPs in a game.

I spent most of High-Tech career sitting 3 cubicle from Intel’s legendary CEO Andy Grove so I got a close hand look at how Imperial Focus Points work in real life. Simply put focus really matters to leadership.

For a 4x game I think IFPs potentially solve some very important problems.

  1. The increased length of turns due to micromanagment

  2. The exaggerated effects of economies to scale

  3. Increase the competitiveness of the middle game

  4. Allow the player to do more of what he thinks is fun.

  5. Now I realize that most games have governors and some are actually decent. I even use the ones in CIV once I think the game is won and gradually automate more settlers… But the temptation is to micromanage to much for the simple reason, because micromanaging is more efficient and there is no penalty for doing so. In real life there is a penalty for micromanaging, you may not be as good as you think, and you will certainly make your workers unhappy.

One of the problems with MOO3 and IFPs was that players were spending more times on turn not less because they were analyzing how best to spend them. Any IFP should include reading reports/status screens as an action. After all Andy Grove could have had 50,000 weekly status reports emailed to him… As a side benefit if you don’t know that a Governor changed your R&D allocation for farming from 8%-10% you won’t upset! However the player should be allowed to drill down as deep as they want but at the cost of IFPs. If you have 25 IFPs to spend each turn than the end turns will be roughly the same length of time as the middle turns.

  1. Everybody knows that if Player A has 10 planets/cities or Player B. has 5 planets/cities that Player A will be produce 2 twice as much and research twice as fast. This explains why GM, IBM, NEC and China have all the technology breakthrus and are super efficient :wink: . Of course there are huge productivity drags associated with any large organization so in fact 10 cities isn’t twice as good as 5 cities. Now CIV3 tried to model this by increasing corruption and decrease the time it took to research a technology that was already discovered. The lack of penalties for a large empire was one of biggest flaws I think with MOO2.

Now there a lot of reason why there is inefficiency in an organization, but one of the main reason is that leaders have to delegate to people who are either dumber and/or less clear on the goals. An example: Early in Civ my city produces either 3 shields and 2 food or 4 shields and one food. If I am building a warrior that costs 10 shields, I want to micromanage the workers so for one turn I build 4 shields and 2 turns I build 3 shields. The warrior gets built in three turns and I have optimized production.
Now when I delegate the production job to the Governor I can tell him to either emphasis food or production but it won’t be smart enough to switch each turn like I did. Now if switching workers cost an IFP it would be something I would do in the early game but something I would quickly delegate in the later game.

  1. By forcing the player to rely more and more on the inferior AI, it gives the player a slimmer advantage as his empire grows larger. To me the most interesting times in 4x games is when the outcome is in real doubt.
    Any technique the extends that period of time is good for me.

  2. I have a love hate relationship with micromanagement. I love it because it makes it possible to pull a losing position out, and because it leads to high scores. I hate it because it simply gets dull and makes games go on much longer than the should. I also hate it because sometimes the perfectionist in me feels guilty for letting the computer screw up decisions and start micromanaging again. IFPs would remove the temptation for me to micromanage and if they actually made games faster that would be a good thing for me. :D In my case I would probably choose to spend my IFPs handling the space combat tactical detail and probably ignore plantery production micromanagement.

You work(ed) in RNB??? I worked for him for a brief time and the other executive staff as their dedicated tech. I moved on to Folsom as our ARIT (US) lead. Working directly for the execs was fun and at times a little scary. :) I agree with your assessments as well. As the ARIT Lead I often have to pick and choose my areas of focus (just not enough time in the day), but I always had enough data on what needed to be micro managed and what I could leave to site leads.