Proven: RNG of starting position is broken.

RCubed1967

At the start of this month's run, I had ten starts in a row with the second pick. The odds of that occurring are 0.09744% (or 1 in 1026). I didn't mind that, thinking that I would not see that occur again for a long time (given that I play ~250 ranked games, this should occur about once every FOUR YEARS).

I have now had a second string of second picks this month. This time with 11 in a row...and still going. The chance of getting second pick 11 times in a row is 0.04883% (or once in 2048).

TWICE IN ONE MONTH, 10 or more turns with second pick.

Yup...definitely a problem in the programming.

PinkApple

I have had similar experiences, although this does not necessarily indicate that it is broken but also leaves the possibility there is some weighting and we just don't know the formulas for how it is decided.

If the devs do intend for it to be 50/50 I would encourage them to look into things as it almost certainly is not functioning this way consistently.

RCubed1967

I agree that the probability of two occurrences occurring in the same month is not zero...but it is over 1 in a million chance (closer to 1 in two million, so I would only need to play consistently for 10,000 years to have it happen again, lol).

I definitely agree that there must be something inside the programming formulas. To resolve this, the programming answer should be simple. Either increase the probability of getting "head" every time you get "tails," then reset at 50/50 OR force 50/50 over short durations.

The additional concern on this RNG is that the same programming issue may lie in the board and in natural draws.

Thank you @PinkApple

FrancisCrawford

Some fallacies here, but yes.

If you play 512-3 games, there should on average be only 1 streak of 10 or more starts from the same position (first or second as the case may be).

The odds of having 2 such streaks, and both for the same position, would be more than 1000:1 against for any given player who plays that many games. (Indeed, I'd guess it's more like 2000:1 against, but I could be making a logical error.)

If you've played fewer games than that, the probability would be yet lower accordingly.

I'm not aware of any proven bugs in ESO RNG, but given how often loot tables get messed up, accurate analysis of randomness does not seem to be a ZoS priority.

RCubed1967

This issue extends to the competitive balance between players, which also must be considered.

A prime example is that in 3.5 months, I have played @Meji six times. @Meji has gotten the first pick each of those times.

The necessary changes in the RNG coin flip and the requirement for the ranking system to consider series instead of one-off games must also consider balancing the starting positions. Each player needs to have relatively equal times with the start.

If the starting position is skewed to one player, and the starting position has a statistical win rate advantage, the results become meaningless. Even using the writ of coin versus an extra gold at the start would most likely cause a skewed result without balancing the start position.

Come on @ZOS_Kevin, give us some feedback on these issues. The ranking system must change, and the coin flip formula must be reviewed.

Meji

Its funny you say that but I've had the majority of my starts being second turn, whether or not there is some hidden factor, it's definitely noticeable.

Personofsecrets

There is absolutely a seeding issue between first and second player choice in Tales of Tribute.

I've had very very long 1st player streaks. There are also seem to be certain players were I tend to go second against them maybe 4/5 games.

It's a big problem since it calls into question other game RNG which could be improperly seeded. Not only could people figure out how to abuse the seeding, but it's also the case that TOT, being not the best balanced game between 1st 2nd player will end up with leaderboard outcomes that have just as much to do with seeding as they do player skill.

spartaxoxo

I experience this as well. I've pretty much given up on expecting to play 1st a fair amount of times.

IncultaWolf

First turn has too much of an advantage honestly, the 1 coin the player that goes second gets is worthless a majority of the time. Very rare exception where you can get armory, but that's only happened a couple times in the 100+ matches I've played

DragonRacer

IncultaWolf wrote: »

First turn has too much of an advantage honestly, the 1 coin the player that goes second gets is worthless a majority of the time. Very rare exception where you can get armory, but that's only happened a couple times in the 100+ matches I've played

And in my experience at going first, I have seen that Armory card situation happen for the second place opponent so often I honestly am wondering if it's rigged against me at this point. LOL

Seraphayel

IncultaWolf wrote: »

First turn has too much of an advantage honestly, the 1 coin the player that goes second gets is worthless a majority of the time. Very rare exception where you can get armory, but that's only happened a couple times in the 100+ matches I've played

First or second turn basically doesn’t matter, the tavern matters. Often enough is the tavern for the first player in the first round pretty trashy as well and the second player gets to buy the 6 coin card which often is Arsenal. Often enough you can buy Luxury Goods and maybe another card in the first round as player one. It‘s all random. There is no real advantage to starting first or being second as it all depends on the tavern RNG.

RCubed1967

Incorrect. I have shown over several months that the win rate for the first pick is much higher than for the second pick.

That does not mean that the second pick never wins, but there is a significant statistical advantage if you pick first.

This has been verified by other people tracking games.

I encourage you to track your games for wins and losses relative to starting position. You need at least 30 games with first pick and 30 games with the second pick to be statistically valid.

Also, please look at my post RNG, Luck and Strategy (https://forums.elderscrollsonline.com/en/discussion/637813/rng-luck-and-strategy#latest)

IncultaWolf

RCubed1967 wrote: »

Incorrect. I have shown over several months that the win rate for the first pick is much higher than for the second pick.

That does not mean that the second pick never wins, but there is a significant statistical advantage if you pick first.

This has been verified by other people tracking games.

I encourage you to track your games for wins and losses relative to starting position. You need at least 30 games with first pick and 30 games with the second pick to be statistically valid.

Also, please look at my post RNG, Luck and Strategy (https://forums.elderscrollsonline.com/en/discussion/637813/rng-luck-and-strategy#latest)

Yeah I have about a 80% win rate against player matches when I have first pick, and yes I have even won against people constantly in the top 10 ranked, so it's not a skill issue. First turn has an unfair advantage

Seraphayel

RCubed1967 wrote: »

Incorrect. I have shown over several months that the win rate for the first pick is much higher than for the second pick.

That does not mean that the second pick never wins, but there is a significant statistical advantage if you pick first.

This has been verified by other people tracking games.

I encourage you to track your games for wins and losses relative to starting position. You need at least 30 games with first pick and 30 games with the second pick to be statistically valid.

Also, please look at my post RNG, Luck and Strategy (https://forums.elderscrollsonline.com/en/discussion/637813/rng-luck-and-strategy#latest)

I‘m sorry, but what you’ve shown is not statistically proven by facts (as the sample size is way too small) and we don’t know the mechanics behind the game. I don’t agree with your assessment that the starting position has such a deciding factor.

The tavern is in every scenario way more decisive than going first or second is. I appreciate the work you’ve done here, but in this case it‘s an agree to disagree situation. I will track my games as you’ve said though and give the result when I’ve reach your given threshold.

What I agree on is the fact that the RNG of the starting position is broken and that there are unnatural strings of either being first or being second in a row.

RCubed1967

@Seraphayel: Fair enough, but your last comment would also support my findings. It does not matter what the source of the issue is; it matters that there is a win rate discrepancy between the first and second pick. The question becomes how to fix it. If it is a programming issue, then ZOS can change that (maybe). If not, the Writ of Coin change may be enough to tweak things closer.

Note that I have recorded 770 games which support the difference between first and second picks. These results include different scenarios in an attempt to discount patron combinations. There are others who are also tabulating their results and have found that this discrepancy holds.

You need to support your argument that an advantage does not exist. Adding to the database by recording and reporting your results is one way. Another way is to run a properly constructed Monty Carlo simulation. Either is acceptable.

@ZOS_Kevin likely has access to everything we could ever ask for to prove this. ZOS just needs to share its database...but they never seem to do that.

FrancisCrawford

Seraphayel wrote: »

RCubed1967 wrote: »

Incorrect. I have shown over several months that the win rate for the first pick is much higher than for the second pick.

That does not mean that the second pick never wins, but there is a significant statistical advantage if you pick first.

This has been verified by other people tracking games.

I encourage you to track your games for wins and losses relative to starting position. You need at least 30 games with first pick and 30 games with the second pick to be statistically valid.

Also, please look at my post RNG, Luck and Strategy (https://forums.elderscrollsonline.com/en/discussion/637813/rng-luck-and-strategy#latest)

I‘m sorry, but what you’ve shown is not statistically proven by facts (as the sample size is way too small) and we don’t know the mechanics behind the game. I don’t agree with your assessment that the starting position has such a deciding factor.

The tavern is in every scenario way more decisive than going first or second is. I appreciate the work you’ve done here, but in this case it‘s an agree to disagree situation. I will track my games as you’ve said though and give the result when I’ve reach your given threshold.

What I agree on is the fact that the RNG of the starting position is broken and that there are unnatural strings of either being first or being second in a row.

You're offering a false dichotomy.

Some initial Taverns favor the first player. Some favor the second. People think their stats show that the first kind are more common.

GCJ_

Where the secret lies is in what the subroutine for the draw pile curation actually is. Is it a single event following a shuffle and the order of the cards in the draw pile are set and not altered at any point between shuffles? Or is it really that only when the card draw function is triggered that card selection for the next card(s) order is generated and subsequently dealt, allowing for the game to pick and choose when and what cards it deals from both populations of each of player's draw piles in attempt to match pre-determined parameters unknown to the players. The tavern replacement card draw function could have the same subroutine in it's programming for the intent on attempting to satisfy the same pre-determined parameter set. This would be more obvious to see happening as it happens. When the smallest odds of probability seem to occur repeatedly. Like when you use the cheap 1 coin treasury card to remove another card from the tavern and both cards are replaced with the exact same two cards, way late in the game but before a tavern shuffle. When improbable is actually mathematically impossible because there's only so many of each card in each deck, the treasury's tavern cards being the smallest of all collections. I cant even count how many games I've played where the tithe card appeared more times than there are in total before a tavern shuffle occurred its happened so often and still happens. That cant be true, and the draw piles being set in a scheduled order after they are shuffled without any unscheduled order changes in between shuffles also be true at the same time. Without human dealers or any transparency in the automation of digital entities that support a fair system that can have its fairness actually be measured against, it cant be said for certain that RNG seeding is used part or in whole in the programming. It is therefore not reliable to track and compare any history of your experiences because we don't have all the information about the given set/control schematic to measure against. The card game itself being a part of a larger total propriety is most likely why that specific piece, or any, hasn't been revealed or confirmed at all, rather than of the programming itself. And probably never will be. Frustration and anger at the powers that control this pastime is what is meant to be experienced, otherwise such info would already be a well shared known thing just as with everything else within this community. Especially by now.

Seraphayel

Okay, I played 22 matches in the last days and one thing I can say: out of all of them I never won as second player. That’s a 100% loss streak for me starting as second player.

I know the sample size makes this irrelevant, yet it‘s a disturbing find. I have no idea how the RNG is rigged, but it definitely is. It doesn’t even need to be deliberately rigged, it just seems to be outright broken.

Your proposed change to give the second player a Writ of Coin directly instead of an additional coin for one round is something I support.

My suggestion is, as I still believe the tavern RNG is the biggest problem overall, is to make the first round of tavern cards an equal distribution of cards ranging from 2-6 coins (no agents) so that every player is guaranteed to at least get one card in the first round. A player might even get two cards if they‘re lucky (2+2 or 2+3 vs 5 or 6 for example if the second player gets 6 coins).

There’s nothing worse than rounds where you waste coins because you cannot buy cards from the tavern because a) the other player bought everything or b) everything is too expensive. It‘s no fun for either player when the tavern is stacked with crap and both players need to do 3-4 rounds of Writ spamming to even get the game going.

RCubed1967

@FrancisCrawford - this feels baited...but I will respond slightly. Please read the entire post as well as my other posts related to the overall RNG (the basis of this one was originally for the coin flip of the starting position.) You could argue a false dichotomy if I used the same patron choices and played the same people, but I am not. The strategies, patron and people are all altered. Unquestionably, some patrons work better in first than second (better in second than first), but that is accounted for. If you still feel that the data is suspect, please offer an alternative hypothesis and supporting data.

@GCJ_ - there may well be a very broken programming issue (I highly suspect there is). The origins of this post were based on a coin toss to see who plays first, which is undoubtedly broken. I feel that programming issues may push into other issues, such as the cards on the table and how our decks are shuffled. The entire program related to RNG should be reviewed.

@Seraphayel - thank you for testing! The more you track, the better your data becomes (and it also helps improve your in-game strategies). The entirety of the game's balance needs a review. I feel like two simple tweaks may make the RNG more palatable in the short term: convert a coin to a writ of coin and review the RNG programming (ZOS could simply check the program with a Monty Carlo simulation). Something more significant might be needed as well, but that takes time.

I am sure that @ZOS_Kevin has access to all the data to review for RNG busts. ZOS just needs to share it.

GCJ_

RCubed1967 wrote: »

@GCJ_ - there may well be a very broken programming issue (I highly suspect there is). The origins of this post were based on a coin toss to see who plays first, which is undoubtedly broken. I feel that programming issues may push into other issues, such as the cards on the table and how our decks are shuffled. The entire program related to RNG should be reviewed.

I totally didn't intend on high jacking your op more than it already was with my feature length response (lol). I was responding to everything that I had read up to that point. I can understand any frustrations I may have perpetuated in getting responses not relevant the op, and for that I apologize. To that effect, a solution to making the 1st and 2nd player starts equal in opportunity is to have the match be a 2-game mini series, alternating who goes first in each of the two games. Patron choice remains across both games to ensure that starting order is the only isolated issue being addressed by this modification, and that any and all other issues are not affected by it nor make any new ones. It removes any RNG (whether or not it's in the programming) from the coin toss because there wouldn't need to be one, or from resulting in perpetual matchmakings where starting 1st or 2nd would be equalized across those multiple matches making fairness be the redundancy that players are subjected to in place of a random event generation no one knows much about. The only random mech left affecting player starting order is if you are assigned as the host or the invited guest when queuing for a pvp matchup using the finder, provided your would be initial opponent doesn't decline the ready check putting you at the front of the line. But since 1st player would alternate across 2 games its a self defeating mechanic, not requiring any reprogramming to implement the 'mod', or any other type and amount of extra labor. It could be a queue option listed within the Tales of Tribute section of the activity finder, equalizing the opportunity afforded by being the 1st to act when the player commits to alternating between acting 1st and 2nd in a 2-game series against the same player, played in succession, no substitutions.

Seraphayel

So, after having played almost 70 matches I can say one thing:

I’ve won as many matches going 1st as I’ve lost going 2nd. In my sample size that means the win:lose ratio is 66:33 in favor of going first.

Honestly, I have no clue where the error is, is it in the programming of the tavern or some kind of RNG, but it‘s incredible how much going 2nd puts you in a losing position by simply making you go second. There needs to be done something about this ASAP. I did not want to believe it first, but something is rigged towards the first player winning. This is just not possible under normal circumstances.

Personofsecrets

What may also be interesting is to see when player 2 does win. For the most part, it may be obvious, but there are some number of games where player 1 get's gifted things like double Luxury Exports turn 1, Hlallu Councilor turn 1, or Grand Larceny turn 1 and still manages to lose. For player 2 to win such situations, they must have done something very broken. But that isn't so obvious in the player 2 turn by turn.

RCubed1967

Well, I had another round of continual rolls. This time 10 in a row of first picks.

Programming a dice roll is relatively easy, or at least can be made fair.

Please address the issue ZOS.

Seraphayel

Six games in a row I went second. Six games in a row I’ve lost.

I [snip] hate this. I don’t want to blame everything on the game and not on myself, but it‘s hilarious how biased the game is towards the player that goes second. I have lost twice as much going second than I did as going first.

[edited for profanity bypass]

RCubed1967

You know @ZOS_Kevin and @ZOS_GinaBruno, the easier way to deal with the problem is to simply state that you recognize that there is a problem, that you are working on it, and that you will provide an update (not necessarily a solution) by x date.

Weird how much better that would work.