Beta Feedback on RNG

Fear leads to anger, anger leads to hate, hate leads to suffering.

 

Quite possible my math is incorrect I was just using the formula the previous poster used. Either way it's an exceedingly small probability to do what's happening to me on a regular basis.

 

Ah, now that's not what you said originally:

Even if you have a 1% chance of something happening, it might. Just because something happens once in a million years, does not mean you have to wait a million years for it to happen. Also, "statistically improbable" has a much, much, lower threshold than 8%, I think.

Here's a little tale for digression:

School Fair decides they need to have a big draw to get the punters in. The decide to have a new car as a prize in a game where you roll 7d6 and if they all come up sixes, the car's yours. Each ticket is $1 and the organisers think that due to "probability" a) no-one will win, b) they will sell a lot of tickets and c) they can return the car to the dealership they borrowed it from. Flawed thinking. First person up ponies up their dollar and rolls the dice, guess what? Yup, seven sixes.

You're now saying that one turn in a game you don't draw an attack card for one character? That does happen, to everyone. I would, however, like to see the games where you ALWAYS get a move card when you need an attack card and vice versa as per your original statement.

 

You claim every game you play which lasts 5 rounds that you have a least one draw where no attack cards show up in a deck that has 24 attack cards out of 34. So tell me, of all the games you play, what percentage of games (that lasts 4 drawing rounds) do you expect to see at least one round of non-attack cards?

You claim this has happened to you EVERY GAME. How many games is that exactly? Is it enough games for you to claim that this is "not statistically probably or even possible"? In other words, how far apart is your observed data from the expected results?

 

Quite right.

Also need to keep in mind that there's hundreds of repetitions even for single player. And hundreds or thousands of players. So basicly anything that is possible will eventually happen to someone. Lottery is good example. Probability of winning is very small, and yet usually someone wins every week.

And hence constant outcry in most game forums to "fix" the RNG. Which is funny in a way, since that is a sign that RNG is probably working just fine. People just want rigged RNG with clear patterns and removed extreme cases. So in other words something that isn't random.

 

Judging from results you can never be 100% sure that a RNG is working well or not.

It becomes meaningful to check it out when you're above two standard deviations or below two standard deviations from an average result.

Then for a sample to be meaningful you'd actually need about a 1000 samples minimum, but we're going to go with your 4 turns instead of 1000 turns.

The chance of having one draw that has no attack cards for the first hand:

First hand
(10/34)*(9/33)=0.08
second hand
(10/32) *(9/31)=0.09
third hand
(10/30)*(9/29)=0.10
fourth hand
(10/29)*(9/28)=0.11

(ps I'm being slightly lazy here, I should account for the chance of move cards depleting as well. I'm estimating it'd be closer to: 0.08 //0.087 //0.094 //0.10 )

The chance of one of these occuring is:
0.08 + (0.09 * 0.92) + (0.10*(1-(0.09 *0.92))) + (0.11*(1- (0.10*(1-(0.09 *0.92))))

Which is roughly 25%.

Which means that the chance of this happening three games in a row is about 1% and indeed two standard deviations below average result.

But counting 'two move cards in one turn' is a bit backward if we're trying to find out if the RNG is not working right. Why not start 10 games against the computer with that deck (or make the same balance of attack cards for all three characters for testing three times as fast) and report the drawn number of attack / non-attack cards?

 

Not sure the exact number of games I have as a sample size actually. I'm level 12 now and I have a fail rate of about 50% overall. I'm not sure if you get exp from failures or not. I haven't checked that. However I would guess with 24 sites of battle unlocked and taking into account that I have repeated all at least twice including my losses with 3 failures each and an average of 3 battles per site, I would estimate close to 300 battles. In EVERY game I have had draws with 2 non attack cards typically occurring when I most need an attack, I think I have data that justifies my hatred of the card dealing system. But I can see from the overwhelming doubt of everyone else here that I'm going to have to post some replays to prove my point.

Anyone know if replays are allowed to be posted?

 

Nobody is insulting you or calling you any of those things. Confirmation bias is human nature and we all do it. Every game that involves a random element inevitably spawns a few forum threads like this. People tend to notice and remember streaks of misfortune far more than when things are going fine.

Everyone here has been incredibly patient and courteous with you given how nasty you've been throughout the discussion. I mean, seriously, you opened the discussion by invoking "the fires of Auschwitz," and now you are telling people (who are trying to help you) to "get bent." Classy, dude!

 

Well, considering the chance to draw no attacks for a single hand is 8% and you have three characters, on average you will be drawing a hand with no attacks once every four rounds.
And the battles are usually action filled enough that 'when I most need an attack' tends to cover every round past the first one. And on some maps especially the first one

 

I'm not talking about on all 3 heroes, just one really. If you talk about the group as a whole I basically get each hero with 2 non attacks at least once usually more EVERY BATTLE.

(I don't really have my priest stacked for attack so my odds are much much higher for that hero to draw two non attack)

 

Are you saying that the game is some how programmed to realise the most critical moment that you need cards, and then at that point deliberately chooses not to give you those cards? Sounds like a lot of hard work compared to just using a random number generator.

Personally i think you need to go back and look at your decks. Check that you have as many 'attack' cards as you think you do. For example 3 weapon slots does not equal 18 attack slots, you may have blocks or negative effects in there. Even if you are right, I can't believe the hyperbole of 'every battle all 3 heroes get no attacks for at least one turn'.

Also remember that you only draw 2 cards per turn, the 3rd is always a move card. also just because you have an attack card, doesn't mean you have to play it. Save it for a better opportunity

 

Ok.
Well, like the others said, the easiest way to check if it's a real problem with the RNG or if just feels like it is to keep a simple log.
For the next ten battles keep notepad open and mark on it how many attack cards you drew each round and how long the battle lasted. Then it's easy to verify it it's truly happening as often as you've said.

 

Lomi, I spent a project of about 80 hours on random number generators in high school. There's a wide range in difference in quality of computer RNG's.

Rather than philosophizing about it, I decided to put it to the test.

Thanks to Mutak for assisting me!

I made a team of 3 characters that each had:

25 attacks
2 trait
9 non-attacks
I did not record the traits, I simply counted the next card instead.

Here's the number of attack cards drawn:

1. 2, 2, 1, 1
2. 2, 1, 2, 2
3. 2, 1, 2, 1
4. 3, 1, 1, 2
5. 3, 2, 1, 2
6. 2, 2, 2, 1
7. 2, 0, 1, 2
8. 1, 2, 2, 2
9. 2, 1, 2, 1
10. 3, 2, 2, 1
11. 3, 2, 2, 2
12. 2, 1, 2, 1
13. 2, 1, 2, 2
14. 2, 1, 1, 2
15. 1, 1, 2, 1
16. 2, 2, 2, 2
17. 3, 1, 0, 2
18. 1, 2, 1, 0
19. 3, 1, 1, 1
20. 3, 0, 2, 2
21. 1, 2, 2, 2
22. 3, 1, 1, 1
23. 2, 2, 2, 2
24. 2, 2, 2, 2
25. 2, 2, 2, 1
26. 3, 2, 2, 0
27. 2, 2, 2, 2
28. 2, 2, 2, 1
29. 1, 1, 2, 1
30. 2, 2, 1, 1

This also shows why a big sample is more effective. If I'd just started counting at 17 and stopped at 20, I would have been inclined to agree.

I'm too lazy to also do the exact math on this now, but there doesn't seem to be anything wrong with the RNG on its average values produced. (Keep in mind that a cosine also produces an averagely okay RNG, yet would be a horrible RNG. It's only one facet of the RNG that's tested here)

 

What would be wrong with changing the system so it's not so random? After all, randomness only leads to luck, and luck isn't skill/strategy based at all. I'm not great at math so I'm not even going to bother trying to figure out numbers and probabilities and all that, but why couldn't we be guaranteed 1 attack card per round like with movement cards? And why couldn't we be guaranteed 1 'other' card(armor, block, buff, heal, spell, etc.) each round also? So 1 movement + 1 attack + 1 'other' card each round. What would be wrong with that? It would still have a random aspect to it because you still wouldn't always get the exact card you need at the time. And if you didn't have any of the guaranteed cards in your deck at the time, THEN it could pick a random card. And if you drew more than 3 cards one round, THEN THOSE cards could be random.

What would be wrong with a system like that?

 

Well, for one it would complicate deckbuilding, not simplify it. It would be a more obscure and less intuitive way of doing it.

It would become easier to play the system. For example, build a deck with just very few attacks, but the ones that are in are all 14+ damage. So that you are assured a 14 damage attack draw every round.

Or have a mage that only has a single firestorm as an attack card, so that you can firestorm every single turn. Your other two characters have only reinforced hide as 'other' card so that they are always immune.

That's what's wrong with a system like that.

 

This would be possible? I don't think I've seen any weapon with only 14 damage cards on it. The cards seem pretty balanced. Same thing for a wizard. Would that even be possible to only have 1 attack type in your whole deck? Same question for the Reinforced Hide you brought up... How would one make it so that piece of armor would be their only 'other card' option?

 

You can get only one attack type in your deck, sure. Not the most powerful ones, but yes. Hammer Of Bashing for instance. But wizards could run two Staff Of Winter and warriors something like Sword Of The Lion just as examples.

Besides the card building aspect, it is a very different system of calculations for balancing, story arcs etc. Not saying better or worse, just a major overhaul to a game that's pretty complete already.

 

Randomness doesn't only lead to luck. It also provides unique situations and challenges for the player even when repeating the same scenarios. Strategy games (especially turn-based games) without a significant random element tend to become very puzzle-like or chess-like. I understand lots of people like puzzles and chess, but I don't think that's what Card Hunter is meant to be.

In Card Hunter (and other games like it) part of the skill is knowing how to best use the cards you've been dealt even if they're not the cards you wanted. Sure, sometimes you will be unlucky and simply lose a battle because you had a bad draw at the worst possible time. That certainly stinks, but fortunately Card Hunter battles don't take very long so it's very easy to shrug off an unlucky loss and try again.

In any case I'm guessing it's a bit late for Blue Manchu to completely redesign the core gameplay so suggesting that they do so is probably not the most constructive feedback.

 

I totally agree with this. I think the problem is that Card Hunter has just a bit too much randomness/luck involved. We have the RNG for dice, which is fine....dice being random makes sense. And then we have all the randomness with card draws. The system I mentioned would leave the dice aspect alone and only slightly alter the card drawing. But, if you guys don't think that system would work well, maybe someone could come up with a better one? Unless you guys feel that Card Hunter's random aspect is fine as it is now? Me personally, I feel like the random factor could be toned down just a bit. I like some luck in my games, it adds an unknown element to things. But too much of it can be a bad thing.

 

How about, if it ain't broke, don't fix it? Skill can have a pretty big effect on randomness. I like to pass the first turn. By shifting through the cards, deciding which to discard and which to keep gives me a lot more options rather than charging in and hoping for good cards. I think that a better skilled player beats a lesser skilled player 9 times out of 10, maybe 8 out of 10. Despite the large random factor, there isn't a large luck factor. And I'm not even counting cards yet.