Two Shav's Drops in Four Days

Uniform "randomness" isn't really random at all.

Any truly random even WILL have streakiness to it at times- in order to attain uniformity you have to remove the possibility for streaks, which makes the event not really random to begin with(in the absence of streaks what's the difference between random drops and just saying "you get a good item every X hours of play"?)

"

crazytrain411 wrote:

"

hazydoc wrote:

"

exhortatory wrote:

For a very good write-up on the "streaky" nature of random systems, see this article.

In short, the left half of the below image has actual random placement of dots.



Hence, streaks. We have the internet, people can squee about them easily, they are perceived as more significant than they are.

Nice article that is exactly what our prof had told us and which was the point I was trying to make earlier. I will bookmark that for the future so I can actually give a source other than my prof from years ago, haha.

Draw all the fancy pictures you want, numbers tell you exactly what the likelihood of streaks are.

There is nothing complicated about it.. Probability of coin toss:
2 streaks 0.5
3 streak 0.25
4 streak 0.125
4 streak 0.0625

ect. ect. it goes toward zero probability to get infinite streak.

Rolling a die:
2 streaks 0.166
3 streak 0.027

Already less than 3% to roll any value on a die three times in a row.


That picture is hella misleading. A streak would mean landing on the SAME spot.

The thing is that is the chance of a single event. The fact of the matter is there are a ton of other equally unlikely events(some good some bad)- Individual the chance of any one happening is very low- but the chance of one of them happening is much higher.

For instace the possibility of any outcome for a nights farming is astronomically low(The probability of getting exactly the drops you got)- The chance of getting no valuable uniques becomes high only because it is a bundle of millions of different outcomes that hold that common property any one of which has a chance close to zero of happening.

Taking an isolated event and claiming ranger is broken because that event is unlikely is insane - Any random distribution has the possibility of outliers and while any given event(such as this one)- additionally even if the specific event is astronomically unlikely, the chance of "Something astronomically unlikely happening" is a lot higher.


At the end of the day though you can't surmise anything about the randomness of a variable from the outcome of a single trial after the fact.

"

tienbasse wrote:

You are not talking about the same thing.

The guy is talking about random generators not being really random (which is definitely the case) ...

While what you say is mostly true, the article linked wasn't about the quality of random number generators, read the text below the pictures. One is randomly generated (presumedly with a good pseudo-random number generator) and the other is supposed to imitate the way glowworms position themselves on the roof of a cave.

"

crazytrain411 wrote:

Draw all the fancy pictures you want, numbers tell you exactly what the likelihood of streaks are.

There is nothing complicated about it.. Probability of coin toss:
2 streaks 0.5
3 streak 0.25
4 streak 0.125
4 streak 0.0625

ect. ect. it goes toward zero probability to get infinite streak.

Rolling a die:
2 streaks 0.166
3 streak 0.027

Already less than 3% to roll any value on a die three times in a row.


That picture is hella misleading. A streak would mean landing on the SAME spot.

The article is pretty good, I think you missed the point. While the probability of HHH in 3 coin flips is around 3% as you mention, the probability of NOT seeing a streak of at least 3 heads or tails in a row given a sufficient number of trials is also very low. For instance in the article they mention that given 100 trials, the odds of not seeing a run of 4 heads or 4 tails in a row is only 0.1%.

The point is that truly random distributions are going to have "non-random looking" sequences and most people do not understand this and assume that the distribution is not random when they see these sequences.

"

Jack_Straw wrote:

"

tienbasse wrote:

You are not talking about the same thing.

The guy is talking about random generators not being really random (which is definitely the case) ...

While what you say is mostly true, the article linked wasn't about the quality of random number generators, read the text below the pictures. One is randomly generated (presumedly with a good pseudo-random number generator) and the other is supposed to imitate the way glowworms position themselves on the roof of a cave.

"

crazytrain411 wrote:

Draw all the fancy pictures you want, numbers tell you exactly what the likelihood of streaks are.

There is nothing complicated about it.. Probability of coin toss:
2 streaks 0.5
3 streak 0.25
4 streak 0.125
4 streak 0.0625

ect. ect. it goes toward zero probability to get infinite streak.

Rolling a die:
2 streaks 0.166
3 streak 0.027

Already less than 3% to roll any value on a die three times in a row.


That picture is hella misleading. A streak would mean landing on the SAME spot.

The article is pretty good, I think you missed the point. While the probability of HHH in 3 coin flips is around 3% as you mention, the probability of NOT seeing a streak of at least 3 heads or tails in a row given a sufficient number of trials is also very low. For instance in the article they mention that given 100 trials, the odds of not seeing a run of 4 heads or 4 tails in a row is only 0.1%.

The point is that truly random distributions are going to have "non-random looking" sequences and most people do not understand this and assume that the distribution is not random when they see these sequences.

That would make sense for high probability events like coin tosses.

The probability of shav dropping is immensely low (think play for a year and not see one). Therefore the probability of it repeating within a short time is abysmally lower.

If you now consider all players that have ever played POE, then you are increasing your space to the likes of which you're talking about in endless streams of coin tosses.

The chance that ANYBODY will get two Shav drops in a row is now higher. The chance that YOU will get two Shavs in a row is next to nil.

"

crazytrain411 wrote:

That would make sense for high probability events like coin tosses.

The probability of shav dropping is immensely low (think play for a year and not see one). Therefore the probability of it repeating within a short time is abysmally lower.

If you now consider all players that have ever played POE, then you are increasing your space to the likes of which you're talking about in endless streams of coin tosses.

The chance that ANYBODY will get two Shav drops in a row is now higher. The chance that YOU will get two Shavs in a row is next to nil.

It doesn't change the fact that streaks/clusters are inherent to random systems.
Obviously a streak of Shav's is less likely than a streak of Gulls or Peregrines (I got 4 of each of these helmets in ONE day last month), but it can happen and it is still random.

My previous point was also that non-perfect random generators can create some weird bugs, reinforcing streaks or nullifying them in different situations.

Interesting discussion tonight. ^^

"

Stormquake wrote:

One time I bought one soda from a machine, but it gave me 7.

Pretty sure I have you beat on luck.

Lucky SoB.

grats on your luck but this post is boring as hell. you could at least edit your videos into one highlight. doesn't make me want to click on your stream at all.

"

tienbasse wrote:

"

crazytrain411 wrote:

That would make sense for high probability events like coin tosses.

The probability of shav dropping is immensely low (think play for a year and not see one). Therefore the probability of it repeating within a short time is abysmally lower.

If you now consider all players that have ever played POE, then you are increasing your space to the likes of which you're talking about in endless streams of coin tosses.

The chance that ANYBODY will get two Shav drops in a row is now higher. The chance that YOU will get two Shavs in a row is next to nil.

It doesn't change the fact that streaks/clusters are inherent to random systems.
Obviously a streak of Shav's is less likely than a streak of Gulls or Peregrines (I got 4 of each of these helmets in ONE day last month), but it can happen and it is still random.

My previous point was also that non-perfect random generators can create some weird bugs, reinforcing streaks or nullifying them in different situations.

Interesting discussion tonight. ^^

Heh, you have to be very careful about defining the system under which something is likely.

Are we considering all players or just one?
Are we considering 1 minute, 2 years or a million years? With 100,000 players and a million years it is very likely that somebody will get a streak of 5 shavs. In 2 hours and 10 player.. no.

There is no streakiness inherent to any system.. Streaks just get more and more likely the longer you 'play' and the more 'trials' there are.

"

Grughal wrote:

It's okay, there, there.

Now

Give the other one to me. Shared fun is twice the fun. (So you'd have as much fun as if you found 4(!) Shavs.

makes sense, give that man one shav!