Anyone else notice this?

When I have a low chance in vats (1-20% ish) I will still take the shot and usually hit the target. Why is it that vats is still so accurate even with such low chances?

Keyword: Chances.

You just happen to be making the shot is all. :shrug:

Because computer's can't produce true random patterns. They can, however, produce very good pseudo-random number generating algorithms that generate decent statistic values over time. But in games, often you'll find short term/small sample size patterns that feel like they don't make any sense. Or at least, that's my theory & I'm sticking to it.

There is no such thing as a false percentage chance. May 100% or 0%, but the key word is "chance." A lot of things are involved in chance in fo

That's incorrect, humans are incapable of random patterns, machines can produce as random as possible without quantum processors. They typically use a timestamp to determine the results as a nanosecond level; slot machines are the same way.

I don't see where I mentioned anything about human capacity for producing random patterns.


The key there is 'as possible.' That's not, imo, truly random.

I'm most definitely not a mathematician or whatever. Only go by what I read. If you'd care to link me to something, that'd be cool. But I have yet to read anything that indicates that computer program formulas consist of anything but pseudo-random number generators - which includes the ones that use the time/date for the unpredictability seed. IMO seeming random because most minds aren't familiar with the program formula that generates the numbers, isn't true random.

My point is... if I have a 4% chance I should only hit that target like 4/100 times (roughly speaking)

But the way it is... I hit a 4% chance probably around 1/5 times. I've played 160+hrs of new vegas and this is the way its been.

See what I mean?

I disagree, machines cannot produce random patterns because they have to be programmed into them, it cannot be random if it is preplanned. Humans, however, can produce random patterns, ever watch a child doodle?

For sake of example, let's say the chance is 1/100. That means for every single bullet fired, you have that 1/100 chance. It's not cumulative. Therefore, it's theoretically possible for you actually hit your target 20 times in row.

Of course, in real life the odds of that actually occurring are likely astronomically low. It's like hoping to roll a 7 many times in a row or hoping to see a ballplayer steal a certain high number of bases in one season more than once in your lifetime, if ever...or whatever. But since random in the game = a computer program, well...it happens a lot. Pseudo random generation at work. Especially in games where I assume they don't spend tons of money or time trying to create/buy the absolute best random generators possible into the equation.

It's also possible in terms of the game, that despite 160 hours of playing, that it isn't a large enough sample size of actual shots fired (with the same exact numbers) to generate the average you're expecting. Sample sizes vs. accurate probabilities aren't something I know much about so not going to speculate.

It's not truly random, no, but it produces the exact same end results. As I said, truly random results require quantum computing, something that isn't widely used.

That's not correct, but I'm not gonna get into the finer details of statistics. Just think casino odds, the law of large numbers always prevails over time.

Instead of just telling me I'm incorrect, would you mind being more specific what I'm incorrect about? My theory for why the game isn't producing results like the OP is expecting, or just the general concept about sample size vs. accurate probabilities, where I wasn't claiming truth/big knowledge in the first place? Without knowing how many shots are actually fired, hard to say whether the OP's results are an accurate representation, is it not?

First off, you are using the word "theory" in the wrong context; you have a hypothesis, not a theory. I only have to say one thing, the law of large numbers. It really is too much to get into, a college level course of statistics worth. It's like trying to explain the theory of evolution to someone with no knowledge of science...it's just not something you can truly understand in a few paragraphs.

Here's a few primers though if you're really interested in it.
http://en.wikipedia.org/wiki/Law_of_large_numbers
http://en.wikipedia.org/wiki/Random_number_generation

They are not easy reads, even summarized on Wikipedia.

Anecdotal evidence is not statistically relevant.

However, if you'd really like to test the accuracy of VATS percentages, heres a good start.

1) Build a character with NO vats effecting perks. Commando, sniper, concentrated fire, etc ... None of them.

2) Find an enemy that is a good distance away, and check VATS. Make sure you are standing (not crouching) move closer or farther away and recheck until you have exactly 10% chance to hit. Do not fire.

3) save.

4) recheck vats, make sure you are still at 10%. FIRE.

5) record the result

6) reload the saved game and repeat 4 & 5 at least 99 more times. Honestly you should get into the thousands for statistical clarity, but that might get tedious, especially with the next few steps.

7) reload that save and move closer, until you have a 20% in VATS. Take 100 samples at 20% and record the results.

8) repeat for 30%, 40%, 50%, etc.

9) now go all the way back to the beginning and retest everything while crouching.

10) now repeat all of the above with each of the VATS perks.

Once you have all that data collected, we'll actually have a decent statistical sample.

Edit: should probably do a complete set of tests for pistols and rifles, energy and conventional. Explosives and burst-fire weapons would be a bit tougher, but possibly worth testing.

I agree with LadyCrimson here. Computer randomisation is never completely random, since it is a computer program applying its logic and code to produce results. Therefore its just the appearance of randomness, or pseudo random.



This statement is contradictory. If its not truly random then it cannot produce exactly the same results as a real random event. It is one or the other, it cannot be both. You are right that computers often use a timestamp as a means of producing a value that appears to be random, but isn't really.

@hawkeyenick - lol, yes, I wasn't trying to get a lesson in statistics etc. as a lot of that would definitely likely go over my head. I just don't like it when people say "you're wrong" without saying what they find to be wrong, if that makes sense. (edit)Outside of not liking my terminology, I'm still not clear exactly but oh well, sometimes that's just how it is (I do know of the law of large numbers, btw, but don't claim to completely understand it).

@Low-key - "Anecdotal evidence is not statistically relevant." -- agree, and (to this layperson) that sounds like a decent plan for trying to get more relevant information.

early in the game vats is more correct. As your man becomes a killing machine vats may not know that you are now 100 in guns.

It may not know what perks you picked or how high you luck is.

I don t think vats understands how good you made your player is my point.

The end result is ESSENTIALLY the same.
http://en.wikipedia.org/wiki/Pseudorandom_number_generator

So the number generator pulls up 34763824637284326784243678 one second and then 34763824637284326784249637 the next. There's no way for you to land on any value you want as the counter goes by the nano second, but it still produces "random" results.

Think of it like this, you can roll a 10-sided di, or you can use a quantum calculator that only goes to the number 10. The results are still 1/10, they are just obtained through different means.

A better example is to think of a synthesized chemical. It's never going to be the "real thing", but it can get the exact same end result.

Definitely agree here, other factors probably aren't taken into consideration. Luck for instance can make you a millionaire in the game with 7 or better, but you'll go broke with any less.

Noses!