Reliability=Glock. Problem is I don't have a degree just an answer!
This is a discussion on Question for the Engineers & Scientists ? within the Defensive Carry Guns forums, part of the Defensive Carry Discussions category; I am NOT a engineer, but I did stay in a Holiday Inn Express last night and I can conclude that you people have to ...
I am NOT a engineer, but I did stay in a Holiday Inn Express last night and I can conclude that you people have to much time on your hands. Buy a Glock and carry a small BUG and dont worry about it.
Glock 22, 27 Gen 4
Ruger SP101 .357mag
S&W 637 Airweight
Ruger Single Six
Ruger Blackhawk Bisley 45 Colt
Mossberg 835 Grand Slam
Reliability=Glock. Problem is I don't have a degree just an answer!
Member:USCCA, NRA, GOA, WVCDL
U.S. Navy vet 1955-1959, USS Dashiell DD 659. Glock 19, Ruger LCP, Ruger .357 Mag.
When you are at the end of your rope, tie a knot and hang on.
This really is not a good question for engineers or scientists.
Engineers and scientists will solve the problem by adding thousands of variables to the equation and just end up complicating the answer.
It is a good question for applied mathematicians (stay away from the theoretical math people though because they typically don't have real world experience and will also complicate the answer, too).
Kilowatt3 provided the best answer to the original question using simple linear regression method statistics.
On a side note, if your firearm fired successfully 1000 times without any malfunctions right out of the box, I want to purchase it. (Just kidding)
there are two separate issues here, one is the machines typical function in working condition, the other is it's Mean Time to Failure or how long on average it takes to break.. Mean time to failure should be published by the manufacture somewhere (if it's not publish it probably means that they're not very reliable) The other issue is how common a malfunction is the only way to get to this number is to shoot the weapon long enough to get at least about five failures and builds some plots of failure, vs number of rounds fired between failures. To see if there are infact trends of increasing failure rate through time. This may be very hard to do with only 5 failures but if there is a strong trend you should see it. for a better overall failure rate I would fire it till i had 20 failures. This would get kind of expensive though..
G20 / G29
I didn't say that the 15/15 outcome means his gun is only 95% reliable. My point was that even if it were only 95% reliable, there would still be an almost 50/50 chance that he could shoot the 15/15 string.
You concluded that the 15/15 string indicates that the probability of the next 15 rounds working right was one (100%). That's wrong. I'm saying that based on a statistical analysis, it can only be concluded that that probability is something better than 35%. Again, you're confusing an outcome with a probability.
If you toss a coin once, and it comes up 'heads', that's an outcome. You've gotten 100% heads. Does it mean there's a 100% probability it will come up 'heads' every time? Of course not. If you toss the coin twice, and it comes up 'heads' both times, you've gotten 100% heads. Still doesn't mean you've got 100% chance of a head on the next throw, and it never will. Your results are an outcome, not a probability.
I agree that the exact time a malfunction will occur cannot be predicted, but the point is, the probability of a malfunction CAN be predicted, and the more 'historical' data you have, the better it can be. A single 15/15 string does not provide enough information to predict it very accurately, but you can narrow it down a bit. It's pretty simple statistics.
The experiment, in this case firing 15 shots, is utterly incapable of giving the information needed to answer the question being asked. Therefore, all of the mathematical stuff becomes irrelevant. Now, if you did an experiment a little differently, lets say 20 sets of 15 shots, and scored +/- for each set where there was a failure, you might start to get somewhere with non-parametric analysis. But, that approach supposes that the gun doesn't change with each successive use, which of course, it does. Recoil springs don't last for ever. Magazines get dirt in them. etc.
It would be great if the manufacturers provided reliability information, or if there were some "Consumer Reports" for gun owners, but there isn't. All we can go on is brand name and brand reputation, and an element of luck.
I don't think the OP's question is even remotely one of statistics and probability as the "test" or "experiment" is naive in its premise.
Actually, in the world of manufacturing they also use some vodoo magic called SPC, which stands for Statistical Process Control. They think that they can predict a point at when a failure occurs by measuring the trend toward a failure.
Example, if I fire 1000 rounds out of my glock, and measure the dimple made by the firing pin on the primer, I can determine the rate at which the pin is wearing out, and if I know at what dimension it will fail to cause a primer ignite, I can use SPC to predict how many more rounds before this happens. The same could be done fro gunk buildup on the slide. At some point it will build up to the point that the gun does not cycle properly and fails to feed. All I need to know is how much gunk each round deposits on the slide and at what poiint the slide will not move freely enough.
To some extent every shot you fire move you one shot closer to a failure because nothing last forever, (except maybe twinkies).
I AM an engineer, but don't want to touch the statistics with so many variables outside of my area of expertise (such as...if there WERE a failure, tracking down the cause is a big factor...was it the ammunition, for example?)...
BUT...many times the purpose of a break-in period for many products is not to establish reliability, but to seat seals, mesh parts which have some minor variances a little better than machine finishing, etc.
I actually spent a few years working in the fields of probability, failure and reliability. Statistics can only be applied to large numbers of items or events, not individuals. Reliability of mechanical devices (such as handguns) follows a Weibull curve, that kind of looks like a bathtub. You have a larger number of failures at the beginning of life of a product ("infant mortality") when all the individual items that have inherent flaws (be they material, assembly or design) fail. Then you have a flat section of the curve that represents the normal failure rate of the item. Finally you have the end of life section of the curve, when the collective wear and tear starts causing frequent failures. The curve can be reset to the flat section by restoring worn or broken parts to new condition. You can establish Mean Time To Failure of an individual handgun by firing a lot of rounds and tracking how often the gun malfs per x number of rounds, and how that number is trending. When you have the malf, you need to clean, lube, replace springs, etc., that return the gun to normal operating condition. Then start over again. You'll need to do multiple runs. A good rule of thumb for the minimum sample size is the square root of the total number of rounds that is the actual normal lifespan of the handgun (as opposed to the designed lifespan) Now all this will only work if you use the same ammo throughout, and don't have any significant change in conditions. See how simple this is?
The good thing is a kind of meta-experiment is going on all the time with a number of gun owners shooting a number of handguns of the same manufacturer and design with a variety of ammo, reporting it over the internet, that gives you a general idea of how reliable a gun of that model by that manufacturer may be.
I will not buy purchase a weapon that has a manufacturer break in period... If I purchase a weapon that has ftf or fte issues then it gets sent back to be repaired and SOLD. I only jeep weapons that give me trouble free service. It doesnt matter if it is a glock or a keltec..... repeated FTF or FTE issues that are not magazine or ammo attributed go bye bye. Also if it is an ammo problem but not an ammo problem, a problem with the weapon... it is gone. I sell it. I had an issue with a few manufacturers... some which I will never purchase from again, some I will. if the weapon is a lemon or needs a break in like companies like Kahr and Kimber have stated, I won't waste my time with it. Too many other choices.
That being said I usually run minimum of 400 practice rounds or 200 rounds through each mag ( whicever is greater) through my carry guns and about 100 of the carry ammo of choice through it or 50 rounds through each mag, before I can safely carry the weapon and feel confident in it.
I have put thousands of rounds through my XDMs with 100% reliability, that is "100 percent". No breaking period needed. I for the life of my can't understand why anyone would buy a gun that needs a breaking period. I have bought speakers that have a breaking period but my life might not depend on my speakers not malfunctioning. No formula needed for common sense.