Machinist Splitshot

[kikix12]

Computers don't do well with "random", seeing as there's nothing random in computers, unless they're broken.

Since there's nothing random in the "random number generators", developers always use some sort of calculation to decide. And most often that calculation is based on something basic like time.
That generates an issue in that in certain time windows, everyone can get really shitty throws or super lucky, for example. Of course, we're talking about fractions of a second here, but that is repeatable. Technically, it is possible to guarantee success or failure on every computer-generated random number...but whatever. I'm getting off topic.


Either way, most of the methods that are commonly used tend to have some repetition to them and cause strings that you wouldn't believe being common quite commonly. They make up for that in the grand scope of things, of course...since, again, they're not random.

A possible tool that developers sometime do make however is a very horrid mechanism that "corrects" inappropriate numbers. For example, lets say that for something with 10% chance of succeeding, the server at 9000 (or any other arbitrary, high number) notices that no one did succeed at it. So it will "fix" the issue by giving a string of successes that will drop the chance at 10%, so in this case, 1000 successes in a row.
Of course, it is never this blatant. But such mechanisms are used and may be part of the reason rare strings of successes/failures are more common than statistics could suggest.


In the end though...hey. Winning a lottery is a one in several (dozens) of millions, yet they are being won by SOMEONE regularly. It just rarely happens to be us...So if you got a big string of whatever...you just happened to be that someone somewhere that got the off odds in his "favor".

To simplify...There is no randomness, just a number of possible ways of calculated success or failure. Depending on how advanced the calculations are, results as described in original post may be far more common than statistics of REAL randomness claim, without it being any sort of bug.

There's nothing to do with that but change the way the random numbers are generated, but the good methods that provide high accuracy require more calculative power and are harder to make, so they aren't really used in games outside real-money hazard or research.

[Iscah]

Random chance is random.

It's entirely possible* to flip a coin 20 times and have it come up heads every time - and your 21st flip still only has a 50/50 chance of finally coming up tails, however "overdue" you feel it must be.

* Possible but unlikely, of course: specifically 1 in 2^20, which comes to 1 in 1,048,576. But the outcome of the next individual flip is still 1 in 2.

[RiyahArp]

It's entirely possible to flip a coin 20 times and have it come up heads every time - and your 21st flip still only has a 50/50 chance of finally coming up tails, however "overdue" you feel it must be.

The problem is the 50% chance is assumed by getting increasing amount of people flipping coins. Over time, you should see the total number of flips be 50%, but there seems to be so much weirdness in the individual that sometimes its useless as a guide in the short term.

Like I remember in older rpgs they would give you this massively damaging weapon that hits 25% of the time, but they were 100% useless since you could miss every time in any particular encounter. Youd eventually see an average of one in four hits, but that doesn't help when one encounter you game over. So its true, but it can end up being useless too.

[Iscah]

I realise it's no consolation when bad luck can make a RNG-based weapon useless, but it still does mean that there's nothing inherently "going wrong with the RNG" just because one person is having incredibly bad luck. Though of course it makes for a frustrating mechanic in the first place.

Back to the "20 coin flips" model, there's only one way to get 20:0 heads:tails, but 20 ways to get 19:1 if any one of those flips is tails (= 1 in 52,429) and... I'm not certain of the maths for getting 18:2 but if it's 20 x 19 ways then that's 1 in 2759, and 17:3 (still very bad luck) is only 1 in 153. And so on.

(EDIT: I'm doubting my maths logic more as I think on it, so don't take my word for this.)

Of course, the "good luck" versions of these ratios are equally likely, but I think you're also more likely to remember the times it went wrong than the times it went right!

Plus you have the 'special bullet' shots that guarantee a success on what could have been the times where you got lucky anyway.


Meanwhile of course, Murphy's Law ensures that all your lucky shots occur just as an encounter ends or the boss goes untargetable...

[NolLacnala]

I need to dig up my screenshot of four sets of four failed gathers (that's sixteen fails in a row) at 86% success. It's not that their number generator is broken, it's throwing fail strings.

[Reckion]

That is exactly why you have Quick Reload on a 15 sec. cooldown.