One single failure does not equal poor reliability! <simple maths lesson>

[Scottish_Skier]

There seem to be several broken touring binding threads on here.

A few comments on manufacturing reliability...

1) One single instance of a broken binding does not automatically equal manufacturing or design problems.

2) Reliability can only be measured by looking at the % of failures across a wide sample size. A sample size of one contains no information. So before commenting on design issues you need to look at reliability across a wide number of samples. (Reliability is usually measured using normal distribution and standard deviation)

3) In any kind manufacturing there is no such thing as 0.0% failures. Primarily because post production testing can never be 100% accurate with out actually inducing or stressing the failure your trying to detect.

4) Sometimes shit breaks... But generally the overall % of failures is very small. For example if marker sold 60 000 F12s and only (say) 30 get returned with broken toe pivot pieces then that is 0.05% failures. Which for most mass produced manufacturing industries is actually very good.

5) If people really want 0.000000000001% failures then the manufacturing cost would go up lots. (in terms of materials, production and test costs etc). As would the weight of your binding.

Finally, here is the proof that any touring binding can break...

broken f-12
[ame="http://www.tetongravity.com/forums/showthread.php?t=210851"]Marker Tour F12 Major Failure - Teton Gravity Research Forums[/ame]

broken baron
[ame="http://www.tetongravity.com/forums/showthread.php?t=214627"]Broken Marker Baron - Teton Gravity Research Forums[/ame]

broken dynafit
[ame="http://www.tetongravity.com/forums/showthread.php?t=213097"]DYNAFIT FAILURE - Teton Gravity Research Forums[/ame]

broken fritschi
[ame="http://dev.tetongravity.com/forums/showthread.php?t=144708"]broken fritschi - any suggestions? - Teton Gravity Research Forums[/ame]

[farmer]

Great post. I had a buddy who was doing QA/QC for NASA. They ordered 5000 of a certain battery, which they wanted the manufacturing process to be as identical as possible; same few people doing everything, materials all from the same run, etc, all so that 4998 of them could be tested to failure, giving them a very good idea of the failure point of the remaining two that they needed.

Most people aren't willing to pay for 5000 pairs of FT12's so that they know the exact failure point of their set.

[verbier61]

supercool post, thanks a lot!

[Kim Jong-un]

Good points.

Also, we all know TGR folks always ski harder, huck higher on their bindings than non-TGR folks. So as a sample size we should expect a higher failure rate here vs. the general public.

[XXX-er]

sure but when some one breaks a dynafit or a fr+ because they are hucking 20ft drops thats is just plain going outside of the desgn parameters

BUT If I read it correctly the F12 is not breaking on big hucks, its showing a pattern of breaking when people bobble & go down on one knee using up all the forward travel of the pivot and bottoming out during the skin up before they ever get to lockingdown & hucking anything

I think Marker has a BIG problem in a brand new product that was suposed to go head to head with and kill fritschi ,I don't know how they are gona fix that broken pivot short of a major redesign of the pivot,maybe more metal & more weight ??

maybe there is something to be said for the slow incremental (people here call it boring) changes that fritschi makes to their products?

The engineers simply can't test for everything ,sell any product and someone will come up with a use the engineer didnt think of, obviously Marker tested and this did not show up ,maybe the testers were TOO good never fell or bobbled in a dip and this didnt show up in testing

OR everything in america is a conspiracy so marker did this on purpose just to fuck your shit up eh?

[iscariot]

You points don't change the fact that I would not spend my money to build the test sample necessary to determine the failure rate of Marker bindings that have an obvious design flaw.

[Chowda]

sure but when some one breaks a dynafit or a fr+ because they are hucking 20ft drops thats is just plain going outside of the desgn parameters

BUT If I read it correctly the F12 is not breaking on big hucks, its showing a pattern of breaking when people bobble & go down on one knee using up all the forward travel of the pivot and bottoming out during the skin up before they ever get to lockingdown & hucking anything

I think Marker has a BIG problem in a brand new product that was suposed to go head to head with and kill fritschi ,I don't know how they are gona fix that broken pivot short of a major redesign of the pivot,maybe more metal & more weight ??

maybe there is something to be said for the slow incremental (people here call it boring) changes that fritschi makes to their products?

The engineers simply can't test for everything ,sell any product and someone will come up with a use the engineer didnt think of, obviously Marker tested and this did not show up ,maybe the testers were TOO good never fell or bobbled in a dip and this didnt show up in testing

OR everything in america is a conspiracy so marker did this on purpose just to fuck your shit up eh?

I've wondered for a while, how many of the toe piece failures with freerides were causes by cracks developed during knee falls but broke clean while skiing. If I remember correctly fr go barely past 45degrees before the toe bottoms out.

In any case the tour12 seems like a bad choice if you're still learning to skin.

[Scottish_Skier]

BUT If I read it correctly the F12 is not breaking on big hucks, its showing a pattern of breaking when people bobble & go down on one knee using up all the forward travel of the pivot and bottoming out during the skin up before they ever get to lockingdown & hucking anything
I think Marker has a BIG problem in a brand new product that was suposed to go head to head with and kill fritschi ,I don't know how they are gona fix that broken pivot short of a major redesign of the pivot,maybe more metal & more weight ??

Ummm - this is my whole point.

I read the F-12 thread at lunch time...

From my reading that entire thread is all about ONE broken binding. Statistically that is a tiny number of failures. (the original poster suggested that Marker may have seen 3 in total which is still nothing).

Just because a single failure is well reported on web forums / blogs / wild-snow doesn't mean there are design or manufacturing issues. You need to know the 'total number of failures' vs. 'bindings sold' to make any intelligent comment on reliability.

[cpj.slc]

Pffff. Take your bell curves and normal distributions elsewhere. This is TGR and we know all! If a binding breaks the whole company sucks and they owe every single person on the forum a heli trip to Alaska as penance for the sin of manufacturing consumer products. Yes, we may enjoy risky sports that could ultimately result in death, but we aren't willing to risk a nasty fall or injury because sometimes, even well engineered things break.

Great post by the way.

[skis_the_trees]

If I remember correctly fr go barely past 45degrees before the toe bottoms out.

The FFR+ will go quite a bit past that angle. It actually depends on the toe height and if you use the included red plastic shim under the front of the binding, which has a relief cut out of the center to allow for more pivoting before the toe piece will bottom out on the ski. Mine swing freely past 90 degrees. Depending on the toe height, the toe piece can "bottom out" on the ski, the sides of the front baseplate, or the center of the front base plate. Not sure how this applies to the FFR Pro, which has a different toe and pivot. Also, beware about mounting the FFR+ toe without the red shim, which you can see recommended some places online to increase the forward canting of the binding.

[iscariot]

Just because a single failure is well reported on web forums / blogs / wild-snow doesn't mean there are design or manufacturing issues. You need to know the 'total number of failures' vs. 'bindings sold' to make any intelligent comment on reliability.

I guess you're right...I mean ThinCover was just building the test sample size.

[jondrums]

From my reading that entire thread is all about ONE broken binding. Statistically that is a tiny number of failures. (the original poster suggested that Marker may have seen 3 in total which is still nothing).

wait a second - in that thread alone, there were three different posters reporting an identical failure. There was another post in ski forum, which I'm pretty sure was a different failure.

Ok, so 4 identical failures might be a small number, but to my ming we're seeing a trend.

And, although everything you say in the OP is of course true - you don't address design flaws, which is what this problem actually is. The engineers obviously overlooked a critical design feature, and its coming back to bite them. There is no way to refute that if the design was *slightly* different, the stress on the joint would be reduced by a factor of 3. That's just shoddy work on the part of Marker, and is inexcusable in my book.

[iscariot]

Yes, we may enjoy risky sports that could ultimately result in death, but we aren't willing to risk a nasty fall or injury because sometimes, even well engineered things break.

The problem is that products like this Marker binding ad the Salomon Quest weren't well engineered.

You are right in that there is a lot of risk in our sport.

Why increase the chance of injury, permanent disability, or death with even potentially, if not obviously, faulty equipment?

[Scottish_Skier]

Ok, so 4 identical failures might be a small number, but to my mind we're seeing a trend.

Maybe... But until you actually have the full data set those 4 failure's still tell you nothing.

Imagine 40 Marker F-12 bindings have broken at the toe in total. Maybe 10% of those googled "Marker F12 failure" and posted something on tgr. Those 4 reported failures would still be a small number if they had sold 20 000 of these bindings in total. (NOTE : I am making the numbers up!)

FWIW I don't care about specifically about Marker F-12's issue - there is another thread for that (and for all I know it could be a real problem).

My main point is that without the full data set then you simply cant make accurate or intelligent comments on reliability. Or make comments that one binding must be stronger than another. (say F12 vs Fritshchi).

[I've seen black diamonds!]

My main point is that without the full data set then you simply cant make accurate or intelligent comments on reliability.

http://www.wildsnow.com/2888/salomon...tings-failure/

Sometimes a simple test, or even a mere observation, tells you something isn't safe. If we're talking about one ski that delams (or that guy from La Grave whose Dynafit toe piece broke), I'm with you . But there are also cases in which conclusions may be safely reached without a large data set.

[ticketchecker]

Great thread and timely.

Trying to decide whether to return mine today, the retailer I bought them from has agreed to take them back based on my not trusting them. Hate to regain the Duke weight if the pair I have is not going to be one of the statistics. They ski fine, haven't gone up in them yet though.

[1000-oaks]

F10/F12 pivot failures aren't manufacturing defects, you could take every single one ever made and break it with a simple knee fall due to the stop being designed into the pivot. They don't need metal or more material at the pivot, they just need to remove the stop and give it more range.

[Hugh Conway]

My main point is that without the full data set then you simply cant make accurate or intelligent comments on reliability.

so you've said this forum's entirely useless "duh"

[old goat]

Of course one failure doesn't mean anything by itself. It does mean something if someone posts a thread about it and 6 other people report failure. And you don't need to use as the denominator the number of units sold--the proper denominator would be the number of people who regularly read and post on the forum times the percentage of those folks who have said binding--a hard to estimate but much smaller number.

What's an acceptable failure rate?--that's a judgement call. Obviously with a piece of safety equipment the failure rate has to be very low--a binding failure is potentially much more serious than a delam. To use an extreme example one airpliner crash is enough to mandate safety retrofits to airliners; for cars you usually have to kill a few dozen people (still a lot less than a full airliner) before anyone gets excited. (And the percentage of Ford Pintos that actually blew up was a lot lower than the percentage of binding X that have failed. I should know--I was the proud owner of a 73-- a genuine piece of crap. You had to have metric and English tools both to do a tuneup, although it did ok in the snow with snow tires and 360 pounds of sand in the trunk.)

[Beyond]

Ok, so 4 identical failures might be a small number, but to my ming we're seeing a trend.

And, although everything you say in the OP is of course true - you don't address design flaws, which is what this problem actually is.

OK, but I'm troubled by the backwards logic here. First, we wouldn't have known about a "design flaw" without the failures. And "to my mind" is key here: Humans tend to see patterns everywhere, even if the pattern isn't. Short of a reliability test, we don't know the real failure rate of the toe, or whether it's weirdly high. Nor do we know about user behavior; did every failure result from the same forces applied in the same way?

So let's say that any product, however well designed, will fail at some given rate. The failures will be clustered around pieces that experience particular loads, or are made of particular materials. Now we would like any failure rate to be essentially zero. But unless we want to PAY for a near zero rate, as we do with airline tickets, the rate will have to be a bit higher. So when we encounter examples of this rate in a very non-random sample like TGR users, we see a pattern (hey, plastic and kneefalls) and then we attribute this pattern to weak engineering.

But isn't the design always just aimed at putting the product below a particular failure rate? For instance, if you choose a certain design plate, you're not saying it will NEVER fail, you're saying it has a very low expected rate of failure that you're happy with. And odds are, when it does fail, it'll be a certain bit that fails more often. Not all design elements are equally likely to fail. There will always be a "pattern."

But if the failure rate is acceptably low, isn't the engineering design fine? (Or "fine enough"?) If the F12 has had this failure 4 times out of 10,000 units sold, is that negligent engineering? Should they have done more testing, spent more engineer hours, to anticipate this and eliminate the possibility? OK, what about the next possible failure upstream?

Put another way, how much testing, and how much designing are we willing to pay for? We could ski unobtainium bindings that were designed for years, exhaustively tested in all possible conditions, each single unit X-rayed for flaws, and say came with a failure rate of 1/10,000,000. But they'd cost $10,000. Who's first in line?

IMO this latest thing isn't about rapacious companies (re Salomon) racing weak products to market. It's about a biased sample of users wanting to pay comparatively little for a product that they expect to have nearly perfect design and a zero failure rate, cuz it's their ass on the line. Like good Americans, we want it all, all perfect, right now, for cheap...

[Matty_S]

I agree with what you are saying overall, but in regards to the FT12 it is still a binding i won't trust. We have seen 5 identical failures come through the shop in the past 5-6 months. That on top of the thread here is discomforting. It might not statistically be a high number of failures, but it is a design I won't trust until it is improved.

Other than that, yea the random/infrequent failure that happens here or there with most of touring bindings on the market is more acceptable.

[old goat]

In 45 years of skiing--admittedly not close to the level of many on this forum--I have never broken a binding--starting with cables binding and nonrelease Geze turntable, and including beat to hell rentals, Silvretta cable tele bindings, 3 pin bindings for tele on skinny skis, and all manner of ordinary bindings--Rossi, Solly, Marker, and Tyrolia--until I broke both of my z12's last year. And I never knew anyone else who had either. But as skiing pushes limits that weren't dreamed of in the past bindings are being asked to do more--absorb bigger forces, tour and lock down, and weigh less--than ever before, and the pace of development and release of new products is much faster--because skiers want it and the companies want profit. In such an environment failure is inevitable. That doesn't mean people shouldn't report the failures. That's the only way things get better. And remember--the Quest insert failure was just one incident and the product is off the market.

[farmer]

fuck it. dynafit or dukes. anything else is just half assed.

[jondrums]

ok, now we have between 3-4 reported failures here on TGR, and a fellow above reporting 5-6 failures in his shop. All of those failures identical in nature.

So, "one single failure" doesn't indicate poor reliability. Does 8-10 identical failures with no other types of failures reported? Methinks yes. If these were just general bell-curve type failures - one would expect a variation of types of failures following no general trend.

[Hugh Conway]

So, "one single failure" doesn't indicate poor reliability. Does 8-10 identical failures with no other types of failures reported? Methinks yes.

Especially with a new product.