Litz and Cortina - follow up on barrel tuner discussion

[gunsandgunsmithing]

Does debating the meaning of the word work constitute this as a political thread?

 

[Chiquita]

During my short time in the sport, I have heard many many times "let your grouping on the target speak for itself"
Now, why is Litz questioning EC? Because he is the technical scientist guru and is looking for numbers or other technical way to prove something, in favor or against EC.

But EC proved his point on the target.

 

[ImBIllT]

What I’m saying is that the “distribution” is not symmetrical, Shooting for group is not an activity that lends itself to “normal distribution.” A bell curve is symmetrical on both sides of the peak.

Word choices aren’t the focus. If I could only reliably string together 2 free throw shots through the basket, instead of three, and I had to start over, and every time I start over I talley a mark, (on the side of whatever number applied), I’d have an even more skewed ratio of results on the poor side of 2 or more baskets.

The graphed results of a skill activity like this on an X Y axis is a tiny bump, then a steep downward slope, not a symmetrical bell curve. Now, do a lot of shooters being tested, fall into a bell curve, perhaps, but that’s a separate issue.

Actually your free throw shooting would follow a bell curve, it’s just that the left portion of the bell curve is cut off due to the boundaries of how you presented the data. You could rearrange the data from your string of free throws to move the boundary and both sides of the bell curve would appear.

 

[davidjoe]

Actually your free throw shooting would follow a bell curve, it’s just that the left portion of the bell curve is cut off due to the boundaries of how you presented the data. You could rearrange the data from your string of free throws to move the boundary and both sides of the bell curve would appear.

I have to disagree. Pick a competent ball shooter and desired success rate, say 50%. There will be some number of consecutive shots, four, six, what have you, after which they no longer meet that chosen success rate of 50% in many trials.

Although they have a 50% chance of shooting let’s just say 5 in a row, of the times they do not, less than five is far easier, and therefore much more likely to occur, than making six or more.

 

[davidjoe]

… That is why I would say a symmetrical bell curve of groups smaller than average, and larger than average, does not capture what is going on in our group shooting.

 

[Ccrider]

There are things going on between them and their targets that they cannot see.

I find this to be true even when using wind flags. Sometimes they are perfectly still, it seems perfectly calm and the unseen gremlins attack.

 

[drop_point]

Free throws are a binary and pass fail. If we convert it to a percentage, it is just a percentage of pass fail. That is a whole 'nother statistical phenomenon.

 

[davidjoe]

Free throws are a binary and pass fail. If we convert it to a percentage, it is just a percentage of pass fail. That is a whole 'nother statistical phenomenon.

I’m going to take that as a silent “agree” on the symmetrical point.

 

[drop_point]

Not quite, it is a binary phenomenon and is related to human performance.

 

[davidjoe]

Not quite, it is a binary phenomenon and is related to human performance.

Well at least agree with me on this…. It’s kind of hard to take that part, out of what we do.

 

[BoydAllen]

I find this to be true even when using wind flags. Sometimes they are perfectly still, it seems perfectly calm and the unseen gremlins attack.

One morning, at a Visalia match, twenty years or so back, during a delay before the first match, I was standing next to Eunice Berger (who won the event). There was not a sign of a wisp of wind anywhere on the range, and I said something like, " Looks like this one is going to be a trigger pulling contest". Her reply was that I should not kid myself, that even though the flags were absolutely still, that there were things out there that would move a bullet, that we had no way to see, and that there was no mirage, and that she shot mirage. Thinking about that, I asked her if she would prefer a light steady position to what we had, and she told me that she would. I mention this because it is exactly what you referred to, from a pretty good source. Eunice was an exceptional lady.

 

[drop_point]

Well at least agree with me on this…. It’s kind of hard to take that part, out of what we do.

I suppose outside of rail guns, ransom rests, and all that jazz, looking at the individual rifle/shooter combination's performance we have the unknown of variability initiated by shooter competency. Some days we are on, some days we are off. It is quite difficult to separate skill from the finite precision of the rifle. But when we do this, now hear me out, we are not having the same discussion that Litz and Hornady originally twisted knickers with. We are answering the question of "how precise (or accurate) am I at shooting this particular rifle/load/barrel combination?"

Forgive me if I'm a bit hard-edged about this sorta of stuff, it was pounded in my head as an economics student that had to take far too many statistics and econometrics courses as an undergrad. I hope I never have to perform and analyze another regression analysis as long as I may live!

 

[ImBIllT]

I have to disagree. Pick a competent ball shooter and desired success rate, say 50%. There will be some number of consecutive shots, four, six, what have you, after which they no longer meet that chosen success rate of 50% in many trials.

Although they have a 50% chance of shooting let’s just say 5 in a row, of the times they do not, less than five is far easier, and therefore much more likely to occur, than making six or more.

Your graph is based on your grouping. if you grouped consecutive misses and consecutive baskets then it would follow a bell curve. It’s the same string of shots, you’re just graphing it differently. A better player would still follow a bell curve, but the center of their bell curve would be toward more consecutive baskets than the center of your bell curve, and their peak would likely be higher, unless perhaps you had an obscene number of consecutive misses.

Even with the grouping you chose, you would have a bell curve, it’s just that you only see the far right side of it. The left side can’t be properly represented by the way you chose to examine your shot string.

 

[ImBIllT]

Free throws are a binary and pass fail. If we convert it to a percentage, it is just a percentage of pass fail. That is a whole 'nother statistical phenomenon.

Yes, but looking at an infinite string of free throws and examining consecutive pass outcomes vs consecutive fail outcomes you should still get a normal distribution. How tall, and how far left or right the center of the distribution was would indicate how skilled the shooter was. And of course, we can get a reasonable guess at what that bell curve would look like with a lot less than an infinite set of free throws.

 

[davidjoe]

Thinking is exercising the most important muscle. ^ Bill, we don’t have a competition where you just shoot all the shots and estimate hit density. Everything is about short trials. (Those short string group sizes are what is being said to now need to be numerically lengthened to be accurate.) I would argue that making a shooter/gun or a person with a basketball perform “more” times, is actually shifting deviation to the human, away from the object, just as the requirement for “consecutive” free throw shots tried to illustrate.

Our Fclass type of bullseye shooting is admittedly more similar to free throw shooting than BR, where a defined ring diameter, like a steel hoop, determines success - 20 liner tens beat 19 center X’s and a 20th shot 1 mm outside the 10 line.

I don’t shoot BR, but even in your group size shooting, there will be no more and no less than two shots that actually matter, right? Whether five shots, ten or twenty shots, only the two shots furthest away from each other constitute the group size. No points awarded for the coziness of all the others. Your game absolutely is to prevent a true flyer (consecutive free throw breaker) which starts over the free throw analogy, even more than it matters in BR that nearly all bullets went through the same hole.

Our working definition of group size or score in either discipline has pass fail traits and binary traits. Lastly, though, Consecutive free throw hits still will not ever follow a bell curve, there will be more singles than doubles, more doubles than triples, and so forth, it is a pure downward slope. ***Edit, the three point line, instead, is a better indicator of this last point, or a big sample of truly random people, at the free throw line.***

 

[wkdickinson]

I really think the key to this discusion is Erik's comment about "the gun doing something it has never done before" - that is real! I know enough about statstics to be seriously dangerous (5 years as a 6 Sigma Black Belt), but I think we make a mistake trying to quantify shooting results statistically. None of us mortals can shoot enough rounds to make our result statistically valid. So ultimately we need to use our deductive reasoning - "I made this change and this happened".

There is just too much anadotal evidence that tuners work to disregard them. I have great respect for Mr. Litz, but I think he really missed it here.

 

[jelenko]

5 years as a 6 Sigma Black Belt

Very cool!

we need to use our deductive reasoning

And, the expertise built up from all the experimenting and rounds down range.

 

[ImBIllT]

Thinking is exercising the most important muscle. ^ Bill, we don’t have a competition where you just shoot all the shots and estimate hit density. Everything is about short trials. (Those short string group sizes are what is being said to now need to be numerically lengthened to be accurate.) I would argue that making a shooter/gun or a person with a basketball perform “more” times, is actually shifting deviation to the human, away from the object, just as the requirement for “consecutive” free throw shots tried to illustrate.

Our Fclass type of bullseye shooting is admittedly more similar to free throw shooting than BR, where a defined ring diameter, like a steel hoop, determines success - 20 liner tens beat 19 center X’s and a 20th shot 1 mm outside the 10 line.

I don’t shoot BR, but even in your group size shooting, there will be no more and no less than two shots that actually matter, right? Whether five shots, ten or twenty shots, only the two shots furthest away from each other constitute the group size. No points awarded for the coziness of all the others. Your game absolutely is to prevent a true flyer (consecutive free throw breaker) which starts over the free throw analogy, even more than it matters in BR that nearly all bullets went through the same hole.

Our working definition of group size or score in either discipline has pass fail traits and binary traits. Lastly, though, Consecutive free throw hits still will not ever follow a bell curve, there will be more singles than doubles, more doubles than triples, and so forth, it is a pure downward slope.

When I shot BR, I shot primarily HBR, which is for score, and then I shifted to f-class, but then shifted to hunting and haven’t competed in over ten years. That said I grew up shooting at a benchrest club and have been a member(or shooting under a member) for almost thirty years. Although they have started shooting varmint for score at our club, they didn’t do that back when I was shooting BR.

Score shooting takes each shot into account, and this gives you a better statistical picture with fewer shots, sort of. Groups, and scores, and different ways of evaluating groups, will all fallow statistics. The main reason I didn’t gravitate to group shooting was the realization that in a lot of matches, only the worst couple of shooters were statistically different from the winners. That’s actually quite evident in the fact that the winner from match to match is not usually the same person, in fact it’s often someone who didn’t even place well in the previous match. Over the course of a season, the best does actually sort itself out. Jackie Schmidt, Tony Boyer, and a host of other top tier group shooters have a statistically significant number of wins, and that points to their factual superior ability. You cannot often prove that they were better in a single match, but you can prove they were better over a season.

Score shooting actually adds an element of difficulty, plus it measures all the shots. The additional difficulty separates shooters, and the additional statistical power of the sample separates shooters more rapidly.

Because our barrels wear out so fast, we cannot waste shots during tuning. Quick tuning is a necessary evil of the process. Acknowledge the lack of statistical power of what you’re doing is an equally valuable evil. Only by acknowledging that can you avoid going to down meaningless rabbit holes, or staking your flag on things that are essentially random. One reason every barrel seems different is because we often use insufficient data to declare that one barrel like a certain load and a different barrel likes a certain different load. Other trends can transfer at least partially from barrel to barrel, and thus we can generate a larger sample size and apply knowledge to a new barrel.

Consecutive free throws do follow a bell cure. By not mapping consecutive misses, you make the left portion of the bell curve invisible. You can easily shoot poorly enough that the middle of the bell curve is in the negative portion of the graph that you have not mapped. You don’t get to cut the bell curve in half, and then declare that because it only slopes downward on one side that it isn’t a bell curve. It is a bell curve, and your chosen to ignore one side of it. Graphing consecutive misses as negative hits makes the rest of the bell curve visible. There’s a peak in the middle, and it turns downward on both the positive and negative sides, and neither sides ever reaches zero, because if you performed the free throw an infinite number of times, you have a portion of the infinity sequence in which you missed an infinite number of times, and you would have a portion of the sequence in which you hit an infinite number of consecutive times. The Y axis of the graph represents the percentage of the sequence that you hit or missed X number consecutively. Yes you can consecutively hit infinitely, and consecutively miss infinitely, and still be within the bounds of the infinite sequence. Infinity fits into infinity an infinite number of times, but you can still derive percentages and distributions from that.

 

[BoydAllen]

I really think the key to this discusion is Erik's comment about "the gun doing something it has never done before" - that is real! I know enough about statstics to be seriously dangerous (5 years as a 6 Sigma Black Belt), but I think we make a mistake trying to quantify shooting results statistically. None of us mortals can shoot enough rounds to make our result statistically valid. So ultimately we need to use our deductive reasoning - "I made this change and this happened".

There is just too much anadotal evidence that tuners work to disregard them. I have great respect for Mr. Litz, but I think he really missed it here.

Nailed it!

 

[ImBIllT]

… That is why I would say a symmetrical bell curve of groups smaller than average, and larger than average, does not capture what is going on in our group shooting.

It’s symmetrical. It just isn’t necessarily centered on zero.