My Random Numbers Test Application results

RNG is only part of the equation. There are other values, not necessarily the values given in the UI, that come into play. It's been all but confirmed that we aren't told the whole story when it comes to the modifiers.

 

get the RNG code from Ultima IV, problem solved.....




ps... I speak from experience on this.

 

I had my worst fail streak ever with Dash... 23 times fail in a row... yikes!! Chance of success is 40%
(Note the Satyr's response )

 

Hope you ran and bought a lottery ticket after that....

 

@Sycholic

You do not want a RNG based system that has memory to trim hot and cold sequences. That would make the system predictable, which means the one who has the time to farm the most always wins at production, and there is no point of even trying if you have less time than they.

When gambling runs hot and cold is when people see they have a chance to outperform the farmers. The gamble is the small guy gets on a lucky streak and overcomes the big guy on a losing streak, yes the risk is that it goes the other way, but the possibility of that lottery allowing it to happen is why people keep gambling.

The PRNG has nothing to do with the randomness of the system, for the simple fact that there is not a single bot consuming the numbers. The RNG could instead just be a sequential list that loops from 0-12 for each roll, and it is the fact that the next roll goes to a different player in the space time distribution of it being an MMO that makes the system appear random. The result of random sampling of a sequential list is not any different than random sampling of a PRNG is not any different than random sampling of a natural random process.

The solution is not in the number generation, it is what the numbers are used for. D&D 5e despite relying on thousand years of dice history, counsels the DM to find a way to weave a story that makes failure interesting, that it is not an outright failure but instead creates a complication, a change from it's decades old legacy of save or die.

Maybe in crafting instead of loosing your stuff, there is a secondary process you must do to rescue your thing at more cost by sacrifice a lesser thing to make a middling thing so that you can recoup some value. This way there is still a market for rares, but it also creates a market for the middle of things that people would buy so they can sacrifice them to make their greater things.

There are many ways to make a PRNG based system fun. SOTA crafting not being fun has nothing to do with it being random, making it predictable would actually make it even less fun because only the bot farmer that can monopolize the rolls would ever win.

 

Well that's the thing they are just as predictable as any normal RNG, but it doesn't create issues like this....



the left is quasi random the right is normal RNG. note the 'grouping' that happens to normal RNG. esp the d10 and d100 rolls (top 2 rows) and this is why I believe people are saying its broken. example 2 posts up.....

 

Again the patterning and clustering of a PRNG is only relevant if you are a bot that has taken all the rolls away from everyone else. No individual ever sees the PRNG sequence in an MMO because the rolls are distributed across players. So again even if the pattern was 1, 2, 3, 4....you would randomly sample it and it would appear random for the rolls that are distributed to you. Random by definition is not evenly distributed, it is randomly distributed, which means it has to have hot and cold streaks to be random.

The left is not random, it allows a bot that does take all the rolls to predict the results and game the system.

And even if one person was acting as a bot and was able to consume the next roll generated, the PRNG having hot and cold streaks, which by definition is what makes something random, prevents them from gaming the market. It allows random winners and losers so that rares stay rare because getting one is not predictable.

So again if you dislike the system, don't blame the PRNG. Blame the save or die result which is not very fun. Blame the poor display of the odds that is not showing hidden possibilities.

 

You people are priceless.

 

I believe the issue is on "what happens" when the so-called grouping occurs. That can be controlled and, once you do, it doesn't matter what the RNG does. A well designed game will make the RNG feel transparent and not feel like a slot machine.

If we're going to keep the conversation focused on the RNG generator, then the fix can be as simple as a check like: "on a critical failure, if previous roll was a critical failure, override the current critical failure and make it less severe".

D&D has a concept of "inspiration" that allows players to reroll certain dice.

So imagine this:
Whenever a player fails a skill check and the random number is near the cut-off point (for example: on a "random.range(0,100)+1 < 90" check, rolling a 95 would count as a failure but is near the cut-off point), add +1 to a (invisible) "inspiration" counter. The next time a player fails a skill check in a similar fashion (and that the inspiration counter is above zero), reroll the skill check and reduce the inspiration counter by 1. If the reroll happens to fail, do not reroll again.

That creates a buffer for the player. There's no high-concept science behind the buffer, it's just how the rules of the games are set. In a similar way that players in "Magic: The Gathering" can decide to draw a new hand at the start of the game; because it sucks to play with a bad hand. That is what good game design is all about.

In my original post, I mention "when is the next time that the player can reroll?" as part of an important aspect that causes the problem. In the example I used in "Magic: The Gathering", you only get to draw a hand once... so the rules allow some flexibility to prevent the players from feeling like they're being robbed of a victory... even though they built their own decks and that the randomness of it all is fair for every player. They have more control this way.

Again, this would be an adjustment regardless of the RNG generator used that would greatly improve RNG perception on the player but that's still not what I'd want the devs to focus their attention to when/if they decide to tackle the problem. I want them to re-evaluate at what happens on a failure and on a success like I mentioned on my original post:

 

Thanks Gix, at least SOMEONE understand what we did with the freeshards like 10 years ago, and that is where many games have found their inspiration (pun intended). However our dev team has choosen to ignore us... they seem adamant on making the same errors we found solutions for.

 

@Ahuaeynjgkxs I build, design and play Roguelikes so I like to believe that I understand a thing or two about RNG

I hope, despite your reported silence, that the devs are still looking into this.

 

I think that the problem lies rather with that everyone is reporting that there is something wrong with the PRNG rather than discussing great ways of improving the user experience of the features using the PRNG.
Where the bug reports on PRNG will meet silence or irritation, since that is working as designed.

However the problem lies in that discussing great ways of improving the user experience always ends up in the Feedback section of the forums which are usually very silent on dev responses. This since individual devs naturally focus on fixing bugs and less on feedback leaving that for the project leads like DarkStarr who has a limited amount of time to spread out anyway.

So it becomes a self-fulfilling prophecy in that there will be silence from the devs on this issue.
It is one of the things by not being small enough so that all the devs reads feedback, and not large enough to have dedicated people going through the feedback, adding surveys and looking at the metrics.


Which is why in other similar topics I have recommended people to list such questions/suggestions/discussions for hangouts or telethons or for the Violence tech talks.

 

Yeah that is what a few people were arguing back and forth, but not the entirety of the threads... many suggestions and solutions have been proposed to mitigate losses, but they all challenge RMT status quo in some way.

I agree that the devs are spread way too thin... they didn't expect the server bill to cost that much even if I warned them about unity because I had worked with it before we started development (hell I couldn't even build a proper business plan for the company I was working with, I needed 20000 active players to turn a profit). I have also freely gave away solutions to reduce server bill by as much as 50% and was ignored, I feel those solutions are worth millions of dollars, and we programmed them with 0 budget on freeshards. My prediction for the company that had me work with unity, cryengine and unreal engine to make sure we had the best options were including that 50% server load reduction.

Right now its the cloud server provider that is laughing and taking our hard earned money, AND precious devs.

I am red with anger...

 

@Spoon You're right however it is in their best interest to figure out where the problem lies, though.

I believe this issue is propagated throughout a good portion of the game overall. Some skills don't feel good enough to use and it's not necessarily a mathematical issue either. I'm hoping that this is the 20% that Richard Garriott keeps talking about.

 

I have read so many post about how we just do not understand nor do we know the formula they use for this system. I craft every day and no matter what @Chris says I break more items at a 95% chance than I make the roll. I use a crafting ring that gives me an added +3% on top of the 95%. This is not imagined and no I don't forget the good streaks because as of late I have not had any. With my all over Gm skills I loose way more than I make. This argument will go on for a very long time but I think we are all harping on the RNG but we need to just tell him the truth. This system just is not fun or fair.

 

All is well as far as I can see..

RNG Sample Testing 08/08/17
Damian Killingsworth
alchemy skill: 72
cooking skill: 83

Potion of Wolf Speed/Might
stated success chance: 95%
total successes: 296 total failures: 17
total attempts: 313
success rate: 0.945686900958466
failure streak high: 2
success streak high :69

Teleport Scroll
stated success chance: 94%
total successes: 11
total failures: 4
total attempts: 15
success rate: 0.733333333333333
failure streak high: 2
success streak high :6

Obsidian Bear/Venison Surprise
stated success chance: 63% total successes: 29
total failures: 18
total attempts: 47
success rate: 0.617021276595745
failure streak high: 6
success streak high :5

 

Again, I'm going to say that one of the issues is that the UI percentage IS NOT what is actually being tested behind the scenes when all modifiers are taken into account.

Players might have a decent idea if they are good or bad at something before they try, but GM's don't always have to tell the players what the actual pass/fail line is when they are rolling.

Now, you can argue that this design needs to be changed. It might actually be more productive than trying to prove the RNG is broken.

 

They do have to tell the players what the %. They allowed the selling of in game items and resources> to have the players bank on a % after say buying resources then the % better be right. They have said in past telethins that he plays with the numbers sometimes to slow down production but then they must stop the buying and selling of resources or tell everyone what the real %'s are. i don't sell them but I would get super mad if I spent a ton on some and broke everything I just spent my money on. I get pissed as it is because I spend a huge amount if my real life time gathering and loose it all.