Science & Industry

you were just "unlucky" suck it up.

Honestly, it happens. It's probably extremely rare, but at 27% per attempt, that's a 73% chance of failure -- not exactly in your favor. I mean, what are you trying to invent? Ships? or do you just have low skills?

My skills are good, and yes, I'm inventing ships.


Cast two six sided dice 50 times. Only getting one instance of any number, like only one 6, or only one 5, is at roughly the same probability as what I'm describing. It's so very rare it's very nearly impossible. You'd have repeat your 50 streak of casting 2 dice like 40 000 times for it to enter the realm of plausible probability. That's not casting 2 dice 40000 times; that's casting 2 dice 50 times 40000 times; that's a total of 4 million times you'd have to cast a die for it to become even remotely likely that you'd hit a streak of 100 where you get only one instance of one particular number.

I'll get right on it, my good chum. It's a pretty neat idea to spend approximately 2,5 billion isk and approximately almost a year to prove a point to some random dude on a forum whose only remotely useful/constructive/entertaining comment was that I should've bought a lottery ticket instead.

See, now you're being somewhat constructive.

If those 22580 attempts had all been made at 27% chance, hitting a single 44/45 fail streak in all of 2011 would still be only be 1 (one) in 169(ish) probable. That's 0.6%. Ish. Per 22580 attempts. Which took a year.

As for skills, I'm at 5/4/4, going for 5/5/4 at the moment. I'm inventing -1 me marauders btw.

See, I wouldn't be overly annoyed about failing, say, 19 out of 20 attempts, because that's within the realm of plausible probability, at 1,4% chance for every batch of 20 @ 27% chance of success. If the rates followed a normal distribution, one such streak could be expected for every 73ish runs of 20. The thing is, probability ain't linear.

(total attempts)*(fail_chance^(# fails)*(success chance^(# success)) = chance of getting that particular distribution

20 attempts * (0,73^(19 fails))*(0.27^(1 success)) = 0.0137 -> 1.37% chance of getting 19 fails and 1 success, or 1 in 73 batches of 20

40 attempts * (0,73^(39 fails))*(0.27^(1 success)) = 0.00005 -> 0.005%, or approx 1 in 19824 batches of 40

45 attempts * (0,73^(44 fails))*(0.27^(1 success)) = 0.00000097 -> 0.000097% chance, or approx 1 in a million batches of 45

Getting 39 fails out of 40, is 274 times less likely than getting 19 fails out of 20 tries.
Getting 44 fails out of 45 tries is almost 52 times less likely than 39 fails out of 40, and 14124 times less likely than getting 19 fails out of 20.

Yeah, sure, if we talk really really huge numbers and tens of thousands of batches, the likelihood of getting one such swathe of fails increases with the number of tries, but, none the less, the chance of any single batch of 45 to produce 44 fails at 27% chance of success is still 1 in a million, in exactly the same way the eigth toss of a coin has a 1 to 1 chance of being tails; not 1 in 256.

On a different note; when you enter start and finish invention jobs, do you enter them with time in between them, or just one after another? When you finish them, do you finish them all at once or all at once?

I've tried all four variations of those methods, but I haven't noticed any difference in success rates.

I would like to introduce you to the little thing known as a random number generator

If you can not understand that then please go to a game that is much more baby friendly.

PS, I used to run well over 1000 invention jobs PER DAY, so I'm well versed in invention. Call it bad luck, ****** game design, your a moron! No matter what you call it, in the end and with thousands to tens of thousands of invention jobs, you will balance out to the "so-called" statistical chances of invention.

If I had an ISK for every one of these posts...

Seriously, you were just unlucky, suck it up, etc.

45 runs is NOTHING, it's such a small sample that using it to verify whether the numbers are correct is literally stupid. Even if you did get the right numbers, it would still be stupid to draw conclusions from such a small sample. You can flip a fair coin 5000 times and have 5000 heads. The likelihood of THAT is very very small, but it's not 0 (i.e. the colloquialism that should apply is "unlikely" rather than "impossible"). If that happens to you and you know the coin is fair, congrats, you're seriously unlucky. That doesn't change the fact that the probability of tails or heads is 50%. The probability of you getting shitass results is different from the probability of each individual event occurring, don't confuse the two.

Run the numbers for two years like I've done, and you'll see that they really do work out pretty much right on. In fact, just do it for two months running a lot of jobs to have a sample of several hundred data points, and you'll probably see that you're approaching the expected %

Also, I just ran 50 drone invention jobs and got 45 BPCs (true story, not just a "for example"). If that happened to you, would you still come here angry that the system is broken? Chances are, no because you ended up on the good side of the distribution; likely you'd chalk it up to good luck rather than "baw broken". I've also run a few dozen jobs in a row before with 0 successful results. Very frustrating on an initial emotional response level, but I know that the numbers will work out in the end over a large enough sample (and they have), which tends to make me shrug it off and just go buy more datacores. This is why you need to have capital in excess of the estimated amount to do invention safely, as you may hit a rough patch like that and you should be able to bankroll your business.

PS: this might be of interest if you want some theoretical basis to steel yourself in case of terriawful invention luck. Pretty basic stats but helpful and applicable.
http://en.wikipedia.org/wiki/Random_variable
http://en.wikipedia.org/wiki/Independent_and_identically_distributed_random_variables

Agreed.

but if you consider all players doing invention, it's very likely that such thing happens to one of them once in a while.

so yeah, you're unlucky to be that one, this one time.

I'm sorry, I tried invention the other day, 10 runs, 50.4% chance and got 60% success, so my extra 9.6% must have come from yours