rng solved?

[Nairn Niteraven]

Streaks = Random. When you look at the long haul, you should see a pattern, and I've never read anything that made me think that anybody in the long term is out of "sync" with anybody else.

Ride the wave if you're on a roll

[Lumenku]

rng

I read somewhere here that the RNG used the clock to get the base number for the RNG.

It could therefore be possible that if you do 3 quick combines the same number is used.

[Seedling]

It's not using minutes as a base, Lumenku More likely to use milliseconds.

The simplest RNG would return (min+random(time)%(max-min+1)). Considering the amount of randoms a game like EQ needs, it would have to be simple and fast.

But in the end, the only true random is the one that's never rolled.

[Lumenku]

Logically I would have to agree.

But illogically it doesn't seem very random to me.

[Dvinn]

Verant's RNG has been failing since it was introduced. A good random number generator should be such that you don't notice any trends in the results. With Verant's RNG, trends are all you get. Grade - F.

[scarbrowed]

If you've noticed a consistent trend, please share with the rest of us. If the RNG had a trend, we could all abuse it, and skillup faster. Until I see this happen, I'll continue to see the RNG as working properly in a mathematical sense.

[Finavar]

With even a trivial psuedo RNG, it's REALLY unlikely you will get enough of a sample to notice a flaw or a 'system'.

Based on comments from developers at the various Fan Fairs over the years, the RNG is believed to be a very sophisticated one. Think about how many times RNG is called in any *one* person's play. Multiply that out some. It has to be fast, efficient, and as random as possible.

Remember that 'random' in this context means it's hard to predict the *next* number in a sequence based on all previous numbers. Not that the next number is necessarily *different* from the current one. Given some large number of runs, all numbers in the range should be shown to occur the same number of times. The better the RNG, the fewer runs it will take to show this behavior.

Given a fixed hardware environment (like Sony's machine room), it's fairly trivial to write a PRNG that's essentially completely random. For instance, sample the data on an unconnected microphone port, or use data from an antenna. Also, there are hardware cards that produce cryptographically secure random numbers. They probably don't go to such lengths, but that is an example of things you could do.

Now, that isn't to say that there isn't some code in there that rewards non-macro TSers. Which would have nothing to do with the RNG per se. I just find it unlikely, but SOE has done some whacky things... Grats on your skillups

[Renjah]

A random number generator can and will have streaks. Some people who play the lottery think the numbers cna NEVER turn up the same 2 nights in a row. In fact the odds the numbers will be the same as the last draw are the same odds as ANY other combination.

Same with a coin flip. Flip a coin 20 times. The odds of all 20 heads or all 20 tails is the same as a heads tails pattern. Its random folks streaks dont break the randomness. Each "choose" (or flip or combine) yields INDEPENDANT results. A success or fail on one combine is the SAME EXACT chance of a success or fail on the previous combine. Just because you failed 10 times in a row doesnt mean the 11th will suceed or fail.

[qwkhyena]

I knew it was rigged...

Originally posted by scarbrowed
..... it was rigged.

It IS rigged and the RNG hates me!

[Dvinn]

Lets put it this way. With sense heading maxed at 200, what do you think the odds of logging in and failing the first three sense headings in a row is. Do the math if you like statistics. I'd like to know the chance, because it happened to me.

[scarbrowed]

Well, I have no data on sense heading failures, but let's say 1 in 20 for failures at 200. To fail three times in a row would be 1 in 20 * 20 * 20, which is about 1 in 8000. Considering that 500,000 people play EQ, if the average person logs in 3 times a week, that works out to 187 times a week that it should occur.

(And honestly, I made up 3 times a week to make the number come out small. I have days I log in 3 times a day.)

Of course this is now moot, since sense heading is no longer used. But it's still a useful exercise to demonstrate things.

BTW, since we are no longer discussing tailoring, I'll say more in OOC if anyone cares. (If not, I don't blame you.)

[Naedea]

Seemingly unlikely streaks would be more common if the RNG followed this model