Poll
Question: Should we adjust the RNG to represent THC probability more accurately?
Yes - 17 (42.5%)
No - 23 (57.5%)
Total Voters: 40

Pages: 1 2 [3] 4   Go Down
Print
Author Topic: Design Topic #2: RNG  (Read 5055 times)
Wrath of Dagon
Archmaster

Posts: 2368



View Profile
« Reply #30 on: June 01, 2018, 11:52:12 am »

No, you would just have to try different tactical approaches until you find one that works. Lead the attack with a different party member, use a different configuration of weapons and/or spells & abilities (as applicable). Or even go and do something else first and come back to the fight later.

I feel like figuring out solutions like these is more satisfying as a player than save-scumming until the RNG aligns adequately in your favor.
May be you already figured out the best approach, but it's nullified because the seed you happened to get causes you to be hit in the head 3 times in a row. So now not only do you have to figure out the best approach, you also have to figure out how to change the RNG sequence, which is something you shouldn't have to worry about at all. A fight you never have to reload and retry is a fight that's probably not hard enough.

On another note, everyone's talking like missing 3 times in a row with 90% THC is a huge issue. The chance of that happening at any particular time is 1 in a 1000! So if it happens that rarely, why worry about it at all? That's simply how randomness works.
Logged

but it’s just that right now it’s all about games as a service. We can make money out of it, but you can make more money elsewhere.
Vince
Developer

Posts: 8078



View Profile
« Reply #31 on: June 01, 2018, 12:20:35 pm »

Quote
1)   Should we rig the RNG to meet players’ expectations?

Can't believe you're actually asking this. It's like asking, Since the latest batch of children are having trouble learning to read, should we switch to a purely phonetic version of the English language?

Or more precisely, Since an incredibly vocal and relentless 10% of children are complaining, should we change it for everyone?
Well, to be fair, we aren't changing THC to accommodate the player but spreading the misses a bit to avoid long streaks. We won't make you hit more often or make the game easy for you.

As for %, I'd say that at least 50% (if I have to guess, I'd say 60-70%) of people who posted reviews or impressions mentioned RNG negatively, even in otherwise glowing reviews. We can ignore it, of course, but that would be unwise.
Logged
Scott
Developer

Posts: 2394



View Profile
« Reply #32 on: June 01, 2018, 01:50:53 pm »

How many games see many many threads complaining about RNG? By my count every single game using RNG. TNW isn't exactly going to stand out as a game where people complain about RNG. Make a sticky thread about RNG at the top of the forum and let the 50% who complain argue with the other half who explain how confirmation bias works. That's what every other game does. I can't see how bowing to ignorance is the answer.

Also have to say you draw a really hard line on a lot of the things that make RPGs great, which of course is why fans of Iron Tower's games love them so much. Can't see why this would be a point to compromise on.
« Last Edit: June 01, 2018, 01:55:19 pm by Scott » Logged

contributed to: Age of Decadence | Dead State | Dungeon Rats | Battle Brothers | Fell Seal:Arbiter's Mark
working on: Colony Ship RPG | Stygian:Reign of the Old Ones | Encased | ATOM RPG | Realms Beyond
AbounI
*
Posts: 772


View Profile
« Reply #33 on: June 01, 2018, 02:52:26 pm »

I would like to add that TNW will be a party based system, so even if PC will miss 3 times in a row, there are few chances for the whole party (and allies too) to miss 3 time in a row, so I guess this complaint will be lower than expected. Imagine every characters in the party missing all their attempts three time in a row, that would be a terrible lack of luck.

More than that, why not add a critical miss system (gun jammed, loss of balance,..)? After all, we got a critical hit, it would be fair to have the reverse side, no? Why not make this critical miss happening at the third miss in a row?  lol

Vince, how many of the same kind of complaints did you receive concerning Dungeon Rats? Is it comparable with AoD?
« Last Edit: June 01, 2018, 03:07:27 pm by AbounI » Logged
Vince
Developer

Posts: 8078



View Profile
« Reply #34 on: June 01, 2018, 03:06:59 pm »

Also have to say you draw a really hard line on a lot of the things that make RPGs great, which of course is why fans of Iron Tower's games love them so much. Can't see why this would be a point to compromise on.
It's not really about compromising or pandering. We review all complaints, dismiss nothing, and never assume that we know something better. While we'll never deviate from the core, we're always looking for ways to improve the core design.

I believe that raising your combat skill from 6 to 7, for example, thus raising your THC by 10 points, should make you better in a measurable way, give you a kind of certainty that your points weren't wasted. So I agree that 70% THC should feel like it because it's about your skill investment paying off not about probabilities. Just like I don't think that you shouldn't hit too often with a 10% THC, no matter what you roll, you shouldn't miss too often with 70-80% THC.
Logged
Vince
Developer

Posts: 8078



View Profile
« Reply #35 on: June 01, 2018, 03:10:31 pm »

Vince, how many of the same kind of complaints did you receive concerning Dungeon Rats? Is it comparable with AoD?
Being a much simpler game, DR gets less reviews and impressions, but the complaint IS there. Many people say that the RNG can make the same fight too hard or too easy and that's too random for my taste.
Logged
AbounI
*
Posts: 772


View Profile
« Reply #36 on: June 01, 2018, 03:23:28 pm »

Quote
Many people say that the RNG can make the same fight too hard or too easy
But are they right? I don't have the impression the RNG can make the same fight too easy or too hard, as we must also look at the tactical approach. Keep the same approach and tactic for the same fight, chances are low to get a radically different result.
Logged
Vince
Developer

Posts: 8078



View Profile
« Reply #37 on: June 01, 2018, 03:40:44 pm »

This claim has been fairly consistent. We can argue over it and say it isn't so or we can take a look at the RNG and ensure that you get results appropriate to your THC.

The way I see it, it's all about the size of the 'sample'. If you attack 1,000 times your results WILL reflect your THC but you can easily get a much higher or lower average in the course of one fight because the sample is too small. So we can tie our RNG to smaller fights and thus increase consistency, without helping you win or lying about your THC.
Logged
dogwaffler
Novice

Posts: 26


View Profile
« Reply #38 on: June 01, 2018, 08:35:24 pm »

For what it's worth - "accuracy" is really something that can be subject to a perception bias. Me personally, I enjoy the random wonky rounds that throw a wrench into your plans, but it's definitely not the same for everyone; I've had some friends who would get really angry during a bad string of luck in say, Xcom: UFO defense-- of course it's not a problem when your rookie troops make a bunch of impossible shots in a row. Of course it's a pain when a lucky shot takes your character down in spite of your best planning and execution, but those are the stakes.

There was a mod on page one of the AoD mods thread:
Here's an alternate balance mod. Basically, increases both AP and HP for you and enemies, making more combat actions possible while keeping the same pace, stabilizing the RNG output.
...

I obviously don't know the win/loss tolerances of combat in TNW, but the gist I get from the updates here is that it will be quick, dangerous, and somewhat unforgiving. That said, my thought would be to have a default setting with a rounded/corrected RNG curve, for more casual players who don't like to poke around in the game options, and just want to chill out and play for a few hours without getting shut down every time combat come into play. The raw/natural RNG could be places in a gameplay setting somewhere, perhaps in an articulated difficulty settings tab.
I did appreciate the difficult combat in AoD, which took some getting used to (even with the warnings); it made the "fight everything you can" playthrough that much more satisfying.

PS - A further thought for another thread - perhaps more difficulty settings for things like skill check values and shop item availability could work in a setting like this, it might add a little replay value.

As always, thanks for the great content.
« Last Edit: June 01, 2018, 08:37:22 pm by dogwaffler » Logged
aratuk
*
Posts: 11


View Profile
« Reply #39 on: June 01, 2018, 11:41:16 pm »

No, you would just have to try different tactical approaches until you find one that works. Lead the attack with a different party member, use a different configuration of weapons and/or spells & abilities (as applicable). Or even go and do something else first and come back to the fight later.

I feel like figuring out solutions like these is more satisfying as a player than save-scumming until the RNG aligns adequately in your favor.
May be you already figured out the best approach, but it's nullified because the seed you happened to get causes you to be hit in the head 3 times in a row. So now not only do you have to figure out the best approach, you also have to figure out how to change the RNG sequence, which is something you shouldn't have to worry about at all. A fight you never have to reload and retry is a fight that's probably not hard enough.

On another note, everyone's talking like missing 3 times in a row with 90% THC is a huge issue. The chance of that happening at any particular time is 1 in a 1000! So if it happens that rarely, why worry about it at all? That's simply how randomness works.


In your example, the "best" approach wouldn't be the one where you're shot three times in the head. You can't alter the situation to suit your theoretical "best," no matter how many times you reload, so you are forced to figure out a different "best" approach to suit the situation.

And, by the way, approaching the situation differently will probably cause the AI to behave differently, potentially not resulting in those three headshots. The only way you'd be guaranteed the three headshots would be if you did everything along identical routes, in identical order, but hoping for different results.

The predetermined seed simply forces the player to try different tactics on reload, instead of hoping to get lucky trying the same thing all over again. It is more fun to be forced to adapt my tactics and behavior, than just to hope for better luck, in my opinion.
Logged
Dewey_Master
Craftsman

Posts: 350


View Profile
« Reply #40 on: June 02, 2018, 01:46:03 am »

So we can tie our RNG to smaller fights and thus increase consistency, without helping you win or lying about your THC.

I absolutely agree, but hardcoding a maximum of three consecutive misses to certain high THC attempts doesn't actually address this problem.

As I understand your miss limit, it doesn't actually stop triple misses from happening at 70% THC, it just stops the fourth miss from happening. If you have a deck of ten cards, there aren't even four misses in the deck. The only way to get four consecutive misses is to stretch across the reshuffle. What is the probability there? Something like once in every 100 shuffles or once in every 1000 attacks (could be wrong; discrete math was ages ago)? A hard coded limit seems like a solution in search of a problem.

On the other hand, what your miss limit would do by fiat, cards would accomplish organically. Card draws ensure that each NFL game (30 attempts) is played by an NFL level quarterback. Decks not being drawn to exhaustion mean that there remain better games and worse games within a realistic range (about +/- 10%).
Logged
Wizard1200
Apprentice

Posts: 51


View Profile
« Reply #41 on: June 02, 2018, 04:47:50 am »

I think that the problem is the binary effect and not the RNG: If you have a THC of 70 % and you 'roll' a 64 you inflict full damage, but if you 'roll' a 71 your attack has no effect.

I guess that your grazing hits will partially solve this problem, but this solution is probably better:
- THC should be called Accuracy
- Your attack result is determined by 'rolling' a d100 and adding your Accuracy to the result
- If your attack result is 100 you inflict 100 % damage
- Every point of your attack result above 100 increases the damage by 0.5 % >>> Critical hit
- Every point of your attack result below 100 reduces the damage by 0.75 % >>> Grazing hit

Accuracy 30 and a 'roll' of 1 would result in a damage reduction of 52 %.
Accuracy 30 and a 'roll' of 100 would result in a damage increase of 15 %.
Accuracy 70 and a 'roll' of 1 would result in a damage reduction of 22 %.
Accuracy 70 and a 'roll' of 100 would result in a damage increase of 35 %.
Accuracy 100 and a 'roll' of 1 would result in a damage reduction of 0 %.
Accuracy 100 and a 'roll' of 100 would result in a damage increase of 50 %.
« Last Edit: June 02, 2018, 05:04:41 am by Wizard1200 » Logged
Vince
Developer

Posts: 8078



View Profile
« Reply #42 on: June 02, 2018, 08:40:20 am »

So we can tie our RNG to smaller fights and thus increase consistency, without helping you win or lying about your THC.

I absolutely agree, but hardcoding a maximum of three consecutive misses to certain high THC attempts doesn't actually address this problem.
Different problem. Basically:

Question #1: should your THC represent your hitting average during any given fight?
Question #2: should we limit consecutive misses to 3 regardless of your answer to question #1?

Plus not hardcoding but making it optional.

Quote
As I understand your miss limit, it doesn't actually stop triple misses from happening at 70% THC, it just stops the fourth miss from happening. If you have a deck of ten cards, there aren't even four misses in the deck. The only way to get four consecutive misses is to stretch across the reshuffle. What is the probability there?
Very low. I don't recall it happening to me ever and I have over 400 hours on Steam alone. 3 in a row - yes.
Logged
Wrath of Dagon
Archmaster

Posts: 2368



View Profile
« Reply #43 on: June 02, 2018, 01:58:04 pm »

Missing 3 in a row at 70% THC will happened 2.7 times out of a hundred. Missing 4 in a row will be .81 times out of a hundred (that's with true RNG).
Logged

but it’s just that right now it’s all about games as a service. We can make money out of it, but you can make more money elsewhere.
geezer
Neophyte

Posts: 8


View Profile
« Reply #44 on: June 03, 2018, 12:26:35 pm »

Please use a TRNG method like RDRAND/RDSEED for collecting your random numbers. Either use the random bits directly or whiten them to smooth and spread out the anomalies or give the user the option as to whether to whiten the raw random data or not. This will work for anyone with an Ivy Bridge or later Intel CPU or an AMD Ryzen. For the rest try autodownloading hotbits to a container and just refill it as needed with an internet connection. For anyone who doesn't have a newer CPU or an internet connection they could buy a TPM module for their motherboard if it is new enough or a USB device like OneRNG ($40).

Simulating dice rolls can really only properly be done with a TRNG and not a PRNG (unless seeded with a TRNG first). If you start with a good foundation of true randomness I think the combat will always seem more fair and balanced, particularly if you whiten the data as well so that it seems even more random. I'd suggest maybe even using more than one whitening algorithm and letting the user decide which one he likes best. Also giving people a choice of how much the combat relies on randomness might be interesting, although most people will probably converge on a particular amount of randomness vs skill/experience as being the most fun to play. Of course if someone doesn't have a newer computer or an internet connection and they haven't bought an external TRNG you'd have to default to a software PRNG like Mersenne Twister.
Logged
Pages: 1 2 [3] 4   Go Up
Print
Jump to: