Messages

1

Discussion / Re: Feedback and impressions after 1 campaign

« on: January 23, 2012, 12:12:30 am »

By the way, research is exactly a good example of where a time-only approach leads to discovering the "one best choice" strategy. If I know when aliens are going to bring out those big guns, I can figure out exactly how many labs and scientists I need to keep up. Then there is no arms race, if I am certain I can get Plasma Rifles/Heavy Lasers (preference based) in time for the medium armour wielding aliens. A less predictable campaign has a much higher chance of giving me that ohshit moment when I find myself outgunned.

I'd also say it's important consider independently the two factors of alien escalation, that is quantity and quality. Lots of aliens with crappy equipment can be tough, fewer aliens with great equipment can be tough. Currently, the escalation happens at a similar rate for both. You get more aliens and they're better equipped. And in the late-game, they wield the best equipment and come way too often. Maybe some circumstances should only lead to an escalation of alien equipment without greater numbers.

Of course the issue is difficult, you have to not only find a curve for escalation that would be okay, you also have to maintain some pacing balance between the number of missions and the strategic-level events like production and research. Even a production time of 2 days will feel extremely slow if you're doing 7-8 missions per day.

2

Discussion / Re: Feedback and impressions after 1 campaign

« on: January 22, 2012, 11:10:09 pm »

Hey BTAxis - good point there, and I see how one can come to that conclusion. My response to the point that a campaign which is time-dependent only provides a better challenge is that time isn't the only factor in what determines a challenge. For instance, I also experienced a point in the game where all missions had become pretty easy. Sheevar and particle weapons had not yet appeared, while my main assault team had increased quite considerably in skills. Most tactical missions were a quick bag-and-tag operation until that certain amount of time passed and the aliens escalated again.

I'm very strongly against a strategy game coming down to an easily obvious "best strategy". In fact, I'd say a quality strategy game is one where it takes a long time to figure such a strategy out, even for the developers. But I believe that predictability is an enemy of strategic depth. If I know in advance, with decent precision, the approximate date when the aliens might wield certain weapons, that makes it easier for me to figure out the optimal strategy as the sheer amount of variables is reduced. Further, in the specific case of this game, the lack of a dynamic component in the campaign makes me feel like my efforts do not have any impact on anything. What's the point of PHALANX if the only thing I accomplish by doing missions is gaining favour from the planet's nations, but the aliens themselves do not seem to be bothered or slowed down by my operations against the aliens. It would intuitively make more sense if the aliens were building up a presence on Earth, and their success at doing so impacts how much of a threat they are.

Of course, this is a fundamental design issue. There's nothing inherently and necessarily wrong with a time-only approach, even if I prefer a more dynamic one. It's secondary to the issue that the amount of spawns should feel right in any case, which it doesn't in the current version.

3

Discussion / Re: Feedback and impressions after 1 campaign

« on: January 22, 2012, 10:06:13 pm »

Well that's interesting. It differs a bit from the way I'd have envisioned it to be done.

A general note on campaign balance is, I think it'd make more sense to first get the pace of UFO spawns / missions right, and then adjust payoffs / happiness / etc. as needed. First the amount of spawns has to be right, since otherwise the players get frustrated - and then it doesn't matter at all if the mathematical balance of economics is perfect. For what it's worth, as I reached the late game, I didn't care about income from selling UFOs, I had more than enough money, and I could easily complement my income by selling some weapons. Doing a late-game mission nets you a nice bunch of plasma weaponry, alien armour and assorted miscellaneous items that sell for quite a price.

Now, what I'd expect from the spawn rate is for it to be dependent on my previous success against the aliens. So if I'm doing poorly, the aliens eventually appear in overwhelming numbers, I can't keep up so I lose through attrition, as the nations no longer support me. But if I am doing well, I would expect the aliens to spawn less intensely. After all, they're presumably here with a goal. So their missions are important to them, not just to make sure the PHALANX soldiers exercise - and in that case, I would expect that killing their missions would make life subsequently harder for them.

As such, my immediate idea would be to keep a "success score" for the aliens. Goes up if they complete a mission successfully and down if PHALANX troops arrive at a mission site and clean it out, or if UFOs get shot down. This score would also be a moving thing - that is, not remembering what happened a long time, but rather applying to only a more recent amount of missions. And then the current UFO/mission spawn likelihood is affected by this score. Here's a rough example of what I'm thinking:

Mission success is kept for the last 10 missions. The 11th mission's score will then overwrite the score of the 1st (least recent) mission. A simple circular data structure familiar to programmers. Now, let's suppose a UFO that's shot down en-route to mission gets 0 success, a UFO that lands on a mission but PHALANX troops take it out gets 25, while a mission that gets completed undisturbed gets 100. This is simplistic - in reality, it'd make sense to at least add weights to those numbers based on UFO/mission type (a failed harvesting mission for the aliens is much worse than just losing a Scout!). So let's suppose of the last 10 missions, you shot down 2 UFOs, handled another 6 on the ground, while 2 were completed by the aliens. Thus, their success score is currently at 0 + 25*6 + 100*2 = 350.

Now, next time we check whether to spawn a UFO/mission, modify the probability (as a percentage) of it spawning by (success score - 500) / 10. So in this case, it'd be modified by (350 - 500) / 10, or -15%, as in, the chances of a UFO spawning are 15% lower. (I imagine this as an absolute change, say, the existing in-game formula gives 60% chance to spawn, and then this modifies it down to 45%, but of course I haven't actually checked what the probabilities in-game look like). With these numbers, you might be able to decrease the probability of a spawn by as much as 50% (if you shot down 10 most recent missions), or have it increased by 50% if you failed to do anything about the last 10 missions.

That's the general idea - the numbers are mostly for the sake of having an example. Anyway the first thing to do would anyway be to reduce the exponential rate of UFO spawns, because it should under no circumstances be as crazy as it is in the current late-game.

4

Discussion / Feedback and impressions after 1 campaign

« on: January 22, 2012, 05:21:34 pm »

Hello everyone!

First, a quick introduction. I'm an experienced strategy player, mostly of the Civilization series, and an online community dweller. I've also been involved for a number of years in testing and development of games. It has, by now, become a habit to look for balance and AI issues whenever I play a game, but first and foremost, I just enjoy good strategy games, so I was happy to try out UFO:AI, being a fan of the original X-COM. Following are my impressions after playing a campaign in the stable version (2.3.1). I do my best to be constructive in my criticism, and of course an advance warning that yes, this is a bit long.

Overall, on the high level, I am impressed with what the team has achieved. This is clearly a game very much in the spirit of X-COM, but with enough things to set it apart. It's clear that a lot of effort went into the development, and kudos to all the devs for that! Now, on to more specific points.

The first immediately obvious change from X-COM, and it's an excellent one, is the limited supply of personnel. Getting them from nations, contingent on how much they like you, is great. In the original, you became self sufficient anyway, and then there was no reason to care about any of the governments. I also like the ability to create off-base installations, some radar/SAM sites sure are useful. I never had the aliens try and disable one, though, not sure if that's implemented - I was expecting one of the UFOs to just bomb a SAM site eventually.

I immediately enjoyed the variety of equipment and weapons available. The early equipment is very varied, with assault rifles, shotguns, SMGs and the heavy weapons selection. Heavies do especially seem well designed in how they perform different tasks. Machine guns rarely hit by demoralize aliens, flamethrowers are short-range but very lethal and the grenade launcher is just sweet. So, specificically about the...

Battlescape

The Battlescape is certainly fun but did, sadly, appear to have a number of significant weaknesses. Reaction fire in this version is just absolutely weird. RF doesn't happen that often, and when it does, it's often strange. Sometimes, I move a soldier that is, I'm pretty sure, hidden from alien sight, and that triggers alien RF. Sometimes (especially on the crashed UFO map), the spawn positions of my troops and the aliens seem to be such that attempts to move my squad during the first turn result in overwhelming alien RF. I've even encountered a situation where apparently an alien was spawned *behind* my trooper and reaction-fired when I moved. The system feels outright erratic.

Next, grenades. It's very nice to have the physics component, so that the grenades bounce, and it's different based on terrain, etc., but I also feel this takes away a lot from soldier accuracy. The key to getting good results with a grenade seems not to be throwing it accurately but rather predicting how it will bounce. Sometimes, I do it well, throw a grenade at a square some 5-6 squares away from the alien, and it bounces straight to the alien. Woo! That's good, but then what does the accuracy rating of my soldier matter if scoring a hit with the grenade is about how well I can predict the bouncing, not about how well the soldier throws? This also serves to make airburst mode from the grenade launcher very, very good since the grenades will actually explode where you need them to.

I really, really miss smoke nades from X-COM. I don't know the reason why they're not implemented, but that's an aspect of the Battlescape that is sorely missed, that is, the ability to create some concealement. Would be very valuable on some maps.

As I said, I enjoyed the variety of weapons. Trying them out in Battlescape, I came to some preliminary conclusions about their balance. Earth-tech weapons are well balanced in how they gradually lose usefulness. Early on, Assault Rifles and SMGs are good, but an armored alien may be able to eat an entire magazine from the SMG. So there's a real need to update your weapons. This is sweet. It did, though, appear to me that laser weapons are overall superior to plasma-based weaponry. When Laser Rifles first appear, they're excellent. Accurate and kill aliens quickly. As the aliens graduate to better defenses, Lasers do become less potent, but their accuracy advantage over plasma is staggering. I can give Plasma Rifles to my highly accurate assault troopers, but anyone else will miss too much - whereas with lasers, even when the damage may at times be insufficient, almost everyone seems to hit well. Of course later you get the Heavy Laser, to again beef the damage output of laser weaponry.

It's also strange, though fun, how the rocket launcher seems to be one of the most accurate weapons out there. My rocket guys were getting hit probability fairly close to that of a dedicated Sniper user. Intuitively, I'd expect rockets to be less accurate with perhaps a larger splash radius.

Medkits, guys! These are a big balance problem and should not remain like this. You can patch anyone up very quickly with a Medkit, it's actually more efficient than in-hospital treatment at the base. Seems obvious that as many of your soldiers as possible should carry medkits, and that anyone who gets hit should pull back for healing. You can also do a lot with a tactic where there are 2 medics on hand to heal someone. Medkits should certainly not instantly return someone who just took a plasma burst to full health. This is something I think the original X-COM nailed very well. Medkits help to prevent a soldier from just bleeding out on the spot, but actual healing requires a prolonged stay at the base.

The maps are mostly good, but a few suffer from being too predictable. The worst offender is the Big City (+city) map. There appear to be no random elements or spawn points in it, so I always know that my Firebird is to the east of the big building, that aliens will appear in specific squares on the roof, etc. On the map, I typically leave 3 long-ranged soldiers (MG, Sniper and Rocket, or 2 MGs + Sniper) by the craft, while the rest rush in through the entrance and go up the different staircases. It plays out the same every time.

Speaking of maps also, the base defense missions are poor - probably the poorest part in the actual Battlescape. Predictability is also a problem, of course - once you do a base defense once, for a specific base, you know where the aliens spawn. And the base maps are big, so these missions take too much time, especially given how the AI tends to be not all that bright and get stuck on the above-ground level. The first base defense mission I played was okay, though slow. The subsequent ones were just a chore. The main reason I invest into base defenses is just to avoid these boring missions.

Many of the issues in the battlescape are clearly related to the AI. So let's talk about that.

AI

A disclaimer first - I do know how hard it is to create an AI that acts even remotely intelligently. AI in general is a big interest of mine, and I have worked with game AIs. So I'm not going to be saying "the AI should do blah blah, it should be easy to implement".

The biggest AI problem I observed is that it seems to, overall, want to move towards the closest civlian or PHALANX soldier. It doesn't seem that some aliens prefer to keep their distance. On some maps (say, bungalow in the forest) this quite often results in a sight of the aliens running towards you in a big wave. I also experienced this very visibly in the final alien base missions - took a couple of turns to set up positions, and discovered that the aliens had just converged in a bunch around my forces. The immediate impression is that this approach-nearest-enemy behaviour takes precedence over safety considerations, as I often see the aliens end their turn in the open.

The problem is made more complicated by the fact that the aliens can all see through walls (a very understandable cheat from a programming perspective), but can not understand that these walls are unpassable. I've seen situations where a soldier is in front of a landed harvester and IR goggles show 3-4 aliens standing inside, just on the other side of the UFO wall. Ouch. And then they will pour out when you reach the harvester's rear entrance, but not before. Intuitively, I'd say that this is an AI change that could have the most impact on the game - making the aliens aware of impassable walls and so not standing next to one when there's a PHALANX soldier on the other side.

There's also something wrong, possibly, with the AI grenade logic. The AI can throw grenades - I've had that happen to me in the campaign, but only once. Plasma grenades are very powerful, and the aliens certainly would be more dangerous if they employed grenades more liberally. Especially if there's a soldier with 2 medics sitting around him. Ouch!

This may be wrong, but I got the impression that the AI will typically fire weapons at the end of its turn, not at the beginning, so it ends up exposed a lot after firing, as opposed to moving into cover after firing. Overall, I certainly need to look closer into the AI.

Game Pacing

Pacing is what I found to be arguably the weakest component of the game. In a nuthsell, there are too many missions, leading to the feeling that the strategic part of the game progresses too slowly. Despite me having a good amount of scientists and labs, it took too many missions until I was able to field some custom tech. On the other hand, the original X-COM probably had it happen too quickly, you'd only do a handful of mission with the standard rifles before getting lasers.

The pacing is a bit off to begin with, but it becomes completely off in the late game. It didn't help that I got stuck in the tech tree for a moment (more on the later), but by the time I encountered the Sheevar race or started seeing Supply UFOs, the game had become frustrating. UFOs would land on various missions several times each day, with a bunch more just flying around on my radars. I was just fast-forwarding the game, ignoring missions, while waiting for what I needed to wrap the campaign up. As a note here - UFO disassembly takes way too long even for a dedicated manufacturing base. The time is too long even compared to research of late-game technologies. I got stuck in the tech tree, as mentioned, when I didn't realize I needed to disassemble some UFOs to do more research to build alien-powered craft. Still, the time to disassemble a Harvester seemed absolutely disproportional compared to the time for research of the components, or even production of the Stingray.

The late game pacing is such that it would probably take several hours to play through one in-game day without ignoring missions. This is bad. If the designers intend that players mostly use auto-mission, I humbly submit this is also bad - if a player wants to use auto-mission too much, it's a sign of bad pacing. In a game such as this, success on the Battlescape should presumably depend on tactical prowess, not on a semi-random statistical calculation. The auto-mission calculation is also wonky, it appears. Just to verify, I tried it several times with base defense missions. Late-game, when auto-missioning a defense of my best base, where my soldiers were all experienced and equipped with the best tools, auto-mission would usually be a loss. Though this team of soldiers absolutely rolls over the aliens on actual missions. However, an auto-mission defense of a manufacturing base that only has 6 soldiers with pretty basic equipment and no experience other than from a previous base defense, would net me a win. Weird.

Unclarities

Some things in the game are not very clear. Examples:

- You need 10 live aliens to research a certain tech. It's not clear enough, when you only have a couple, that capturing more aliens will open up more research.
- Finding the alien base. I would have probably never found the alien base if I didn't know that in the original X-COM it can usually be done by following a supply ship. In UFO:AI, there is no hint to that extent.
- In-game texts about items and completed projects mention a lot of elements that are not found in the game - smoke, flashbangs, fighter escorts for some UFOs. Some of that is probably due to changes between versions or features not implemented yet. But the end result is that it's hard to tell which parts of the in-game text are basically flavour text, and which relate to the actual gameplay.
- When you can construct new buildings (Advanced Radar, Antimatter Storage), a more prominent message is needed. One would also be useful if disassembling a UFO in a base with no antimatter storage - a warning that the antimatter will be wasted.
- The game does a pretty poor job at explaining what some of the soldier attributes mean, like strength and speed (even though this can be inferred from knowing X-COM).

Miscellany

Just various thoughts:

- It's a shame about the destructable environments. I understand the technical reasons why there are none and why it's unlikely to happen, but I just wanted to mention this - that's one of the things making X-COM's battles so memorable.
- The civilians are dumb. They don't even seem to run away from the aliens in a concentrated fashion. And some are perfectly calm walking down a street with aliens on the one side and PHALANX on the other.
- Would be nice for the economy if item prices on the market did not always match the production costs.
- Speaking of, is there any use (past the initial research) for the huge amounts of alien armour and breathing apparatus you recover? Seems like it's just loot for sale
- Although they're cool, I never found much use for melee weapons like knives. Sidearms definitely have their use, but the variety of pistols seems far more useful for the purpose.

I'll add more if I forgot something, as I often do. I understand, in the meanwhile, that nightly builds are also being uploaded - probably time to check a nightly to see how the development progress has been! Last thing for now, please do not get the impression that I am being overly critical. I enjoyed playing UFO:AI very much, but when posting at a forum that the developers actually read, I concentrate on feedback that might hopefully help improve the game further.