Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[IronyOwl]

Yea, which means each action is essentially a guessing game in the literal sense -- it's already been decided whether it'll hit or not and you just don't know, is all. I dunno, the practical result is the same of course, but it feels kind of cheesy to me.

Well, strictly speaking the same is true of real life as well (barring possible quantum shenanigans, I think). It's not literally random whether you hit something you were aiming at in real life, it's just based on factors you couldn't have meaningfully known about at the time.

That said, I do know what you mean. Probably best just not to think about it too much.

[Vel]

I'm not convinced the RNG is quite as R as it'd like us to believe, being that it basically decides what actions will work and what won't on a given turn right at the beginning of said turn.

Because (without specialized hardware) computers can't actually generate random numbers.  Read up on psuedo-random number generators (PRNG). The algorithms for how they do it are actually pretty interesting if you're into the mathy / CS sort of thing.

Huh, that's kind of interesting. Though, unfortunately I have only enough mathematical knowledge to know that I don't have enough mathematical knowledge to really grasp the wider effects and implications.

Personally, though, I've always felt that game developers who make games that rely on percentage chances often don't actually understand how percentage chances work.

The base accuracy for the crap guns, default squad size, and unskilled soldiers seems excessively low, for example.. you usually have like a 40% chance to hit if you have them in a good position, and it feels like they balanced it so that you had to take things slow and team up all four of your squad members on each alien, under the assumption that while hitting with one was unlikely, with all four you'd have a '160%' chance to hit.

Unfortunately, percentage chance doesn't work like that, and you definitely don't have a guaranteed hit just because you used careful team work. In fact, there's a very strong chance all four will miss entirely! 'That's X-Com Baby!', but, still, it seems excessively harsh and I'm not really sure in the way they were intending, based on experience in other videogames with percentage chance accuracy. It feels like they could do with a narrower band of accuracy, instead of having a serious improbability of hitting at the start and basically a guaranteed hit every single time at the end.

[Zangi]

Also, just tried some save scumming myself... moving to a different spot has a different outcome.  Like I said, do something different, doesn't necessarily have to mean not shoot at the xeno, just shoot from somewhere else.

[Vel]

Well, I don't actually like to savescum, it just sort of bums me out to think about it when I miss on a really high chance and extremely vital shot. After all, you can minimize risk, but there's no such thing as riskless.

It also feels kind of unsatisfying when I magically hit repeatedly with low percentage chance shots.. because it kind of feels like the seed involved is just making low chance shots arbitrarily hit, since multiple of them together is extremely improbable.

[IronyOwl]

Personally, though, I've always felt that game developers who make games that rely on percentage chances often don't actually understand how percentage chances work.

The base accuracy for the crap guns, default squad size, and unskilled soldiers seems excessively low, for example.. you usually have like a 40% chance to hit if you have them in a good position, and it feels like they balanced it so that you had to take things slow and team up all four of your squad members on each alien, under the assumption that while hitting with one was unlikely, with all four you'd have a '160%' chance to hit.

Unfortunately, percentage chance doesn't work like that, and you definitely don't have a guaranteed hit just because you used careful team work. In fact, there's a very strong chance all four will miss entirely! 'That's X-Com Baby!', but, still, it seems excessively harsh and I'm not really sure in the way they were intending, based on experience in other videogames with percentage chance accuracy. It feels like they could do with a narrower band of accuracy, instead of having a serious improbability of hitting at the start and basically a guaranteed hit every single time at the end.

I don't see why they wouldn't, or why you'd be able to tell if they didn't. I've certainly never been in a position where it really, really seemed like the developers thought I'd have a guaranteed kill but I didn't.

Also, if I'm not mistaken the odds of four soldiers firing at 40% accuracy and hitting 0 times is ~13%. That's hardly what I'd call a "very strong chance." Plus, default hit rate is 45% for partial cover, and for the most part anything you'd like to say about your hit chance applies at least as well to the enemy's chances of hitting you back.

Can't comment on the differences late game, however.

It also feels kind of unsatisfying when I magically hit repeatedly with low percentage chance shots.. because it kind of feels like the seed involved is just making low chance shots arbitrarily hit, since multiple of them together is extremely improbable.

Done properly, the seed/algorithm should generate multiple low-chance shots about as often as a "true" random method would.

[jocan2003]

Why dont they use system time as seed or part of their calculation? Since syste, time is always changing, it would be *truly* random dont it?

[JanusTwoface]

It also feels kind of unsatisfying when I magically hit repeatedly with low percentage chance shots.. because it kind of feels like the seed involved is just making low chance shots arbitrarily hit, since multiple of them together is extremely improbable.

If it bugs you enough, you should write down the percent chance of hitting and if you actually hit for an entire session. Then go back and look at it afterwords. In all likelihood, you're just remembering the patterns and forgetting about it when it doesn't as people are overly good at pattern recognition. If you do, I'd be interested in seeing it. I like numbery things like that.

(And if it actually is "fixing" the weights, that would be something interesting to see. It's actually really uncommon in games. It easier just to let people see what behaviors they see and think the developers planned it even when the didn't).

Also, "random" does tend to seem clumpy. It's one of the reasons that iTunes random playlist isn't random at all. Originally it was, but people complained about having clumps of the same artist or even the same song occasionally playing twice in a row (both of which an adequately random algorithm would do, given time) so they specifically made it less random--and people stopped complaining. Doesn't mean it's right or wrong, it's just interesting.

Why dont they use system time as seed or part of their calculation? Since syste, time is always changing, it would be *truly* random dont it?

Most PRNGs actually do exactly that. You can still duplicate the exact behavior if you get it to use the same timestamp (change your system time and get lucky basically), but it's good enough for most (non-cryptographic) purposes.

But what I think they're going for here is specifically to prevent players from easily save-scumming. You aren't supposed to be able to rewrite time just by saving and re-loading the game, even when you aren't on Ironman. As others have mentioned, the Civilization games do the exact same thing by default (although IIRC, there's always been an option to turn that off and specifically generate a new save on load).

It's actually more effort to save the PRNG's state so that you get the same behavior when you reload rather than just getting a new seed, so I'm pretty sure that the dev team probably went through this exact thought process at some point. The default would be to generate a new seed.

[Grakelin]

The base accuracy for the crap guns, default squad size, and unskilled soldiers seems excessively low, for example.. you usually have like a 40% chance to hit if you have them in a good position, and it feels like they balanced it so that you had to take things slow and team up all four of your squad members on each alien, under the assumption that while hitting with one was unlikely, with all four you'd have a '160%' chance to hit.

Why does everybody have so much trouble aiming? My 40-50% chances are always at funny angles against aliens in low cover at a distance. When I actually get up close, it always shoots up to 90%+. Granted, I usually stick with the 40-50% odds so I can keep my own cover, and end up tossing a lot of grenades/rockets.

It probably is balanced with the idea of having all four guys operate on a single alien at a time, but not for the reasons you think. You're not really supposed to spray them with everybody at once.

[Pnx]

Why dont they use system time as seed or part of their calculation? Since syste, time is always changing, it would be *truly* random dont it?

Most PRNGs actually do exactly that. You can still duplicate the exact behavior if you get it to use the same timestamp (change your system time and get lucky basically), but it's good enough for most (non-cryptographic) purposes.

Usually you'd just use some kind of clock function which counts the number of CPU cycles that have gone by, and take the last x number of digits of it, thereby getting a more or less random number. There are flaws with this method, such as if you wanted to get several random numbers in a row without player interaction, like what happens when the computer is taking it's turn.

... Oh, and if you can reliably press a button so that the last few numbers are something like, for example, "555" I would be very impressed.

But what I think they're going for here is specifically to prevent players from easily save-scumming. You aren't supposed to be able to rewrite time just by saving and re-loading the game, even when you aren't on Ironman. As others have mentioned, the Civilization games do the exact same thing by default (although IIRC, there's always been an option to turn that off and specifically generate a new save on load).

It's actually more effort to save the PRNG's state so that you get the same behavior when you reload rather than just getting a new seed, so I'm pretty sure that the dev team probably went through this exact thought process at some point. The default would be to generate a new seed.

I'm pretty sure XCOM and Civilization do predetermined randomness because of the multi player. You often want multi player games that are like this one to be predictable so that if an agent fires a shot at an alien, both player's computers know how the shot turned out without having to have one of them tell the other how the shot turned out, otherwise you're more prone to cheating where one of the computers can go "I rolled the dice and they were all hits".

It's all random in the end anyway, it's just choreographed randomness, and to be fair if you weren't cheating you wouldn't notice.

[Knirisk]

The base accuracy for the crap guns, default squad size, and unskilled soldiers seems excessively low, for example.. you usually have like a 40% chance to hit if you have them in a good position, and it feels like they balanced it so that you had to take things slow and team up all four of your squad members on each alien, under the assumption that while hitting with one was unlikely, with all four you'd have a '160%' chance to hit.

Why does everybody have so much trouble aiming? My 40-50% chances are always at funny angles against aliens in low cover at a distance. When I actually get up close, it always shoots up to 90%+. Granted, I usually stick with the 40-50% odds so I can keep my own cover, and end up tossing a lot of grenades/rockets.

It probably is balanced with the idea of having all four guys operate on a single alien at a time, but not for the reasons you think. You're not really supposed to spray them with everybody at once.

I don't really have trouble aiming, to be honest. If I miss a shot at a Muton in full cover, it's no big deal.

But when one of my best Assault soldiers misses a 99% chance shot on a Cyberdisk without cover and just out of blast radius, I just reload a previous save. I had him behind full cover with a Light Plasma Rifle and the Cyberdisk was just a few meters away. This literally just happened. Next turn, the Cyberdisk takes out one of my first veteran soldiers. If I still have the autosaves, I might just have to take a screenshot of the 99% shot that missed.

Because God damn, it's like winning the worst lottery ever.

[LoSboccacc]

Small maps are good for tactics and provide enough choice for it to be meaningful, but I like the broad sweeps of the original better

It is nice however because it makes less relevant the small squad size; in the first you needed twenty because of the necessity of cover large areas

[Glowcat]

Trying to get through what I assume to be the last mission.

1st time

Spoiler (click to show/hide)

My best soldiers are mind controlled fighting the Ethereals (For the Greater Evil?), my crucial attacks are reflected, and my VIP is shotgunned by the guy who can take down a Sectopod in a single round. Game over.

2nd time I crash halfway through.

3rd time's the charm maybe?

EDIT: Yes, yes it was.

[EnigmaticHat]

So I attack

Spoiler (click to show/hide)

the alien base

with my best troops, who have carapace armor and plasma weapons.  We go all out and beat the crap out of a bunch of mutons, but then we get a long stretch with no aliens.  I'm still being cautious, but it's gotten a little boring.  The a single move reveals two mutons, a berserker, and three heavy floaters all at close range, with a heavy out of position because he was guarding the sniper.  Despite all that, we tear the aliens to pieces, getting the last floater heavily weakened and under suppression.  I only have one move left, a heavy (who we call H20, its a long story).  I quickly take a fifty/fifty shot, miss, and then the floater kills Rascal, our best assault.  THE HEAVY HAD A GRENADE, it would have been a guaranteed kill.

Seriously, the golden rule of xcom is literally every move you do, take a few seconds and think "is this a good idea."  The especially sad thing is that the assault was part of a set, a trio of vets who had matching armor and invaded a UFO together as rookies.  Later, Bonus, a medic who was also part of the trio, got punched into critical condition by a berserker because H20 panicked mid-turn.  We had no medickits left, but we saved him anyways, one shot from our sniper

Spoiler (click to show/hide)

turned the boss battle into the biggest anticlimax ever.

[Anvilfolk]

Using system time is actually a fairly bad idea for RNG seeds. Basically only the last few bits of the time ever change, since we'd need to be quite a few years/decades ahead for the higher order bits to switch. Which actually means we are using a really small subset of seeds. What you CAN do though, is use the system time to generate a new random number, and use that as a seed. That might make things better, but I'm not entirely sure.


Also, you need to fire 70 shots at 99% chance for it to be more probable to have missed a shot than NOT to have missed a shot It's quite a bit, but not that much. Still unlucky as hell that it was in that particular situation!

[dogstile]

I have the best luck. I got the game a day late (Fair enough, postal service isn't brilliant) but then my CD key is invalid?

I want to kick things.