Rare part system - WHY are there 'rare' rare parts?

[LiL T]

yay lil t is back.

I think your first idea will leave too much of a gaping hole in the loot table, there will be no reason at all to hunt once capped, basically.

Then they need to make it more random, or there isn't really anything wrong with it in the first place, and people need to consider there is a pseudo-random number generator at work and to try and hunt mobs at abnormal server hours.

If they wanted it to be truly random, they could use something like a running server side value as a seed, so that it won't be exploitable, then run loops of the function 10 times, then compare output values on a scatter graph or something to check the randomness. They could use the server time in seconds as the seed for each call or something like that.

They probably should make it more random, I don't think they ever intended the loot drops to be random... They should remove the distinction between L, T and E parts and just have the mob literally spit out a random part, that would make it more fun than having to grind thousands of mobs for the parts.

[Cursed Shadow]

The in-game time has minimal impact on this, it is mostly due to the reseeding being too frequent.

They ARE using a server side variable as a seed. KernelTicks()
The reseed time is to frequent, which is actually breaking the randomness.

[LiL T]

The in-game time has minimal impact on this, it is mostly due to the reseeding being too frequent.

They ARE using a server side variable as a seed. KernelTicks()
The reseed time is to frequent, which is actually breaking the randomness.

I've just had a mess about in codeblocks and witnessed this heh, they could as you said be seeding too often, it should be seeded once like srand(time(null)) or srand(GetTickCount()) just after main, in other words its only reseeded at server reboot. But as someone else in the other room explained to me, even if they seeded when the server starts, a random number generator such as rand() can just really be a crap random number generator due to the algorithm used in that function and it can end up just being in a cycle which produces not very random results.

[Cursed Shadow]

We did suggest that Alduin even just applies a +1 to his current seeded value, at least it would temporarily provide a different correlation/distribution.

However, as he rightly said, that could just be the last block on the tower, and it all falls down.

I'm looking forward to the solution that comes out of it, many people are wondering if the same algorithm is being applied to Weapon Slot chances, Weapon Stat chances. Which could be the reason certain weapons are perceived as harder to slot. (Although this may be an indirect perception, due to the weapon already being harder to get parts for. E.g. Devourer)

This is going to need a lot of testing once Alduin has something in place.

[LiL T]

We did suggest that Alduin even just applies a +1 to his current seeded value, at least it would temporarily provide a different correlation/distribution.

After playing around some more I can see that just adding a +1 like you suggested still doesn't produce very random results, it ought to be using things like total ammo left in weapon clip + how many hitpoints of damage the player received from the mob + the distance the player moved in x amount of time. Then feed those values into their own algorithm which produces a seemingly random value between 1 and total number of parts that exists, even seeding rand() with the previous number generated by rand() does not provide random enough results.

Then there is the human perception of random not being random because we look for patterns, there is also a study which suggests we are able to effect random numbers by observation... -.-

computer why u no random GRRRRRRR >.<

[LiL T]

What if...

The player kills the mob, once the player enters the hack screen to open the mob, the number of bullets remaining in weapon + damage player received + distance walked are put into 3 variables which are then put into some sum, the result of which is placed into a fourth variable, which is then looped between 1 and the total number of parts for the duration of the hack screen. Once the hack screen is finished, what ever the fourth variable is will equal the part number, it would not be random but it should be convincing enough...

[Divide]

there is also a study which suggests we are able to effect random numbers by observation... -.-

It is a commonly understood truth that observation itself impacts outcome in any experiment. Observer y u no impartial >_<


Also, zomg it's Lil Terror.

[LiL T]

It is a commonly understood truth that observation itself impacts outcome in any experiment. Observer y u no impartial >_<


Also, zomg it's Lil Terror.

Aye, and I'd assume as the more complex the random number generator is, the more it would appear to be influenced by observation but That is more likely down to how we perceive things with our own minds, because humans only see patterns. Our brains filter out anything we don't recognize as being something we have not put into a class and we then only see what we want to see or recognize as something we have been programmed with. Yes our brains are really bio computers trying to create artificial versions of it self and we get annoyed when they can't function like we do when we demand random.

for example a range of 0 - 50000

147
we would assume the next number would be much higher
3344
5000
we assume now that the next number should be either 5 digits or 3 digits
3577
3421
134

We then look at this as having a pattern because the second and fifth numbers were 4 digits long, and 3 of them started with the number 3. If we have a problem were there are parts rarer than others, we could in fact be seeing a perceived pattern in a set of truly random data, until we can perfectly emulate the human brain on a quantum computer, we won't be able to really generate numbers we perceive as random.

Could we code how the human mind picks random numbers? Possibly, maybe

I actually think we have our mathematics wrong, and that is why computers are not random, if we built computers on a math system, which was a more accurate one, based on the quantum mechanics of our universe. Then we would have computers in a permanently random state which when a problem is presented the solution is produced out of the chaotic randomness but then the universe is not even random, so we are fucked then.

Oh dear I think I just opened a can of worms on this...

[-2XS-]

All parts came from warbots/titans @ J-03. You guys already confirmed its flawed but I leave this anyway.

http://rares.techhaven.org/listings/2xs/titan

[LiL T]

All parts came from warbots/titans @ J-03. You guys already confirmed its flawed but I leave this anyway.

http://rares.techhaven.org/listings/2xs/titan

Humm, could it not just be you are looking at it as a pattern emerging in random data? Has anyone actually bothered to feed the data into graphs to see if it is generating predictable patterns?

[-2XS-]

Well, Alduin confirmed there is an issue with the generator in general. It felt odd to find the summer leech frame & comp that often but like you said the human mind loves its pattern and it's technically possible to win the lottery x times in a row.

[LiL T]

Well, Alduin confirmed there is an issue with the generator in general. It felt odd to find the summer leech frame & comp that often but like you said the human mind loves its pattern and it's technically possible to win the lottery x times in a row.

Or lightning striking the same place many times... Reality isn't even what we perceive to be random in some cases, which kind of birthed the whole philosophy about the concious mind being a variable which helps to shape the reality it perceives, so nothing is truly random, there will always be patterns in the concious mind that creates it.

simply looking at the numbers is also bad, which is why people should put this data into some visual representation, perhaps the devs could probe the random numbers for about 100 calls then pull them into a spreadsheet and generate a scatter graph/chart for each call being set as a different colour. What we perceive when we look at numbers and words is completely different from what 1000 people would look like at an out door party, its different when we see it. Its like being a part of a company for some months and perceiving there to be about 2500 employees, only to be told 6 months later the company has 1500 staff, 1200 on site at any one time. Our visual interpretation to things is so different from the math we use when assigning values to things, the values are arbitrary.

[Divide]

There is no reason to continue evaluating the numbers of the existing system. We have confirmation it is flawed and will be improved with a new RNG.

[Doc Holliday]

There is no reason to continue evaluating the numbers of the existing system. We have confirmation it is flawed and will be improved with a new RNG.

Im glad it has finally been confirmed. november last year during we (myself and some friends) argued that this was the case and despite best efforts (with none of us being math geniuses or able to push the discussion hard enough with adequate proof other than personal results) we were smacked down by the dev team and some of the forum fanbois who said there was no conclusive proof that there was an issue.

To quote ren and stimpy they couldnt see the trees because the forest was in the way.

[Alduin]

Im glad it has finally been confirmed. november last year during we (myself and some friends) argued that this was the case and despite best efforts (with none of us being math geniuses or able to push the discussion hard enough with adequate proof other than personal results) we were smacked down by the dev team and some of the forum fanbois who said there was no conclusive proof that there was an issue.

To get a few things straight here: up 'till recently the code looked fine. There is no inherent problem in the random number generation itself and I have to admit, I did not consider the possibility of there being a reseeding somewhere in the code, therefore I did not search for it. So all in all everything looked alright and with the statistics at the beginning of Titan were non conclusive.
However, while fixing several beam weapon related issues I noticed that there is some reseeding going on which should not be the case, because it introduces correlations. Changing the reseeding to reseed differently will not solve the problem, it will only change the correlations to be differently. The reseeding has to be removed. Different measures of networking synchronization have to be taken. Which are now being implemented step by step, but that takes time, changing the random number generation needs a lot of testing and can easily break everything, but it is in progress.
Some of the changes which will be introduced with R#184, reduce the number of reseeds and will therefore change the correlations and therefore should affect the distribution heavily.

Nevertheless, neither the 'forum fanbois' nor anyone from the dev/volunteer team is to blame for the informations supplied throught the last months, except maybe me.