Thoughts on Ankerson's cut testing results

[tvenuto]

Plenty of people are posing your same concerns more eloquently and with far more clout. Maybe it's time to cheer them on from the sidelines. You know...so people listen to the message instead of trying to figure out why a crazy person is stalking Cliff Stamp.

Indeed. I'm fairly sure all of that was posted in the bushes at Cliff's house. I hope Cliff has a dog! 14 times he typed Cliff's name, on the second account, made for the purpose of typing Cliff's name. If he's trying to DECREASE Cliff's ego this may not be the best strategy...






Cliff

[Cliff Stamp]

And do you expect me to believe you have recruits to do your testing for you on all the marathon cutting trials you have generated?

No, I don't expect you to do anything.

Do you think the erudite and justly esteemed scientist Richard Dawkins speaks as a scientist when he says there is no God?

I don't know what "speaks as a scientist" means, are you asking if the statement he makes is justified by science? I would need to know what the justification he is using. However the question of what is/isn't science itself isn't trivial and quite frankly I don't think it is meaningful in the normal sense. I would argue Science isn't a thing which can be rigorously defined in the way that things belong to the set of things which are science when they have specific characteristics. I would argue it is defined in the Wittgenstein sense of family resemblance and is more of a fuzzy logic determination than binary one. However without getting bogged down in those details, Feynman makes an argument where he says he doesn't believe there is a God in the Christian sense and his argument is based in science, however Craig provides a strong refutation of the argument Feynman uses which would have been curious to how Feynman would have addressed.

[Cliff Stamp]

If he's trying to DECREASE Cliff's ego this may not be the best strategy...

Dude, seriously, do you even think that is possible, come on you are more likely to roll a double critical.

[Cliff Stamp]

....besides all the wildcat variables you have now introduced into your trials

Every time you use variables in that sense an angel loses it wings. Those are not variables, they are sources of error.

Aside from the terminology-geeking, this is a valid point, there is an increased random error when you the kinds of things I described in the above. However I am not overly concerned about random error, that can easily be reduced through sampling size and I calculate it and note it in the results. My main concern is to prevent systematic error because that is an unknown unknown, can't be calculated or even estimated well. In short, I don't mind having more scatter centered on a true result than being more precise about false one.

[Cliff Stamp]

Just for clarity, feeling, smelling, seeing, or tasting a piece of steel won't tell you anything about them except maybe stainless vs non-stainless.

Maybe you can't ...

Nonsense aside, the more you know and the more justification you have the more a kind of inertia builds up where you can become really resistant to changing your perspective. This is one of the reasons why I appreciate the work you do and others who are willing to participate in critical discussion because that is an invaluable resource I can use to try to prevent myself from being too heavily swayed by experimental bias.

[paladin]

In short, I don't mind having more scatter centered on a true result than being more precise about false one.

Please define "true result" in the context of your message, if you so choose.

[Cliff Stamp]

Please define "true result" in the context of your message, if you so choose.

That was sloppy wording, I meant to write I would prefer a result which had less precision but more accuracy, that is to say the sample is providing more of unbiased estimator of the population even if it the spread of the results was significantly increased.

Now of course if I had a magic wand and time/money were not constraints I would do things differently. I will be getting a knife soon which is using an accelerated quench on 52100 based on ultra-fine aus-grain. Ideally I would like that knife, an exact copy of it with a regular HT and another copy with a ideal version of same (Landes style) and another copy with the Fowler/triple quench style, and really ideally at least three samples of each. But that isn't going to happen so I have to try to figure out if there is a difference in that steel based on the argued "superior" hardening when I don't have identical cross samples. The main difficulty is trying to ensure that the differences I see are not going to be due to differences in geometry between that knife and what I compare it to which is not trivial.

[paladin]

Please define "true result" in the context of your message, if you so choose.

That was sloppy wording...

Perhaps...but I think your Freudian slip was showing...

As for having an identical sample of your test blade...cut it in half

[Laethageal]

Well I think the first post was meant to understand how steel could have lower rank in coarse edge rating then they were with the high grit rating when Jim is saying coarse perform better then polished for all the steel.

I think he explain how and why it could happen.

I was one of which who didn't thought about that possibility and clearly the answer he and some other gave showed me what I had missed.

Can't say if anyone didn't get it yet, but I guess if I was able to understand, most other surely did.

[Cliff Stamp]

[...]

As for having an identical sample of your test blade...cut it in half

That doesn't help, I would want multiple samples to see if the difference from one to another was larger than the difference in the samples. Lets assume for example I check the ultra-fine grain/accelerated quench vs the Fowler/triple-quench (I have one of those) vs a normal 52100 (have one of those). Now how do I know that the difference I see is nothing more than some kind of production variance and that I if got three more blades the relative results would not actually invert. Hence, ideally multiple samples.

I generally avoid things like truth as my epistemology isn't correspondence based, it is more pragmatic coherentism so it is based on convergence not absolutes.

[paladin]

...point taken on the sample blade...

I generally avoid things like truth as my epistemology isn't correspondence based, it is more pragmatic coherentism so it is based on convergence not absolutes.

please, cut the doublespeak...

I see that you are very pragmatic...

Coherent-- at times, but that's debatable...

As for your other medical conditions, I know of a good urologist in Winnipeg to help you with your epistemologitis

[Cliff Stamp]

[...]
please, cut the doublespeak...

Now you are the one who opened that door by trying to infer something from the use of the word truth. It was however a nice catch.

[bearfacedkiller]

Someone once said to me, "Sometimes you have to give up your right to be right because sometimes it just isn't worth the fight." It didn't make much sense to me at the time because I was young and proud but over time I have seen that sometimes it is true.

Seriously. :confused:

[chuck_roxas45]

...point taken on the sample blade...

...

[paladin]

Thanks Chuck for paying attention...made my day! :p :spyder: :)

[Fancier]

[...]

As for having an identical sample of your test blade...cut it in half

That doesn't help, I would want multiple samples to see if the difference from one to another was larger than the difference in the samples. Lets assume for example I check the ultra-fine grain/accelerated quench vs the Fowler/triple-quench (I have one of those) vs a normal 52100 (have one of those). Now how do I know that the difference I see is nothing more than some kind of production variance and that I if got three more blades the relative results would not actually invert. Hence, ideally multiple samples.

I generally avoid things like truth as my epistemology isn't correspondence based, it is more pragmatic coherentism so it is based on convergence not absolutes.

Fascinating! A lot of stuff begins to make sense now. Thank you! "Truth" is inherently based on the means of observation, and in this context that means the knife test method. The lack of a "true value" does make accuracy and bias rather fuzzy concepts. Thank you for your willingness to continue this discussion.
(Note: Full quote included to provide any context implied in the original post.)

[chuck_roxas45]

Thanks Chuck for paying attention...made my day! :p :spyder: :)

You're welcome dude. :D

[Cliff Stamp]

[...]

The lack of a "true value" does make accuracy and bias rather fuzzy concepts.

There is a very wonderful interview where Feynman is being asked some really demanding questions (Does God exist? Is there an ultimate theory?). He makes the argument that he doesn't really think like that, that all he does is observe and see where that takes him. Maybe there is an ultimate theory of everything, maybe it is like an onion but the layers are infinite and all you do is keep peeling back layers. The tricky part is his infamous quote - "The first principle is that you must not fool yourself and you are the easiest person to fool." . He doesn't use the terms, but what he describes is pragmatic coherentism and you can clearly see he is talking about convergence of ideas vs absolute positions.

If you want to see this in action then this thread is an interesting example : http://www.cliffstamp.com/knives/forum/read.php?3,37143" target="_blank . I start off with a really simple goal :

-show the scale and direction of the influence of grit on edge retention on a slice

However alone the way then a lot of interesting things start to come up and lots of ideas start being explored :

-what sharpness actually means in regards to slicing and how two force measurements are needed to get the full pictures

-how the cutting mechanic changes at high polishes and it makes the performance rebound up from a local minimum

-details on the actual nature of the teeth which form and how they change at ultra low grits and why this makes the performance fall off after a local maximum

-that the edge retention can actually be predicted from two sharpness measurements and using them to calculate the edge retention factor (initial slicing sharpness / initial push cutting sharpness)

It also asks a bunch of questions such as :

-If a 3Cr13 knife at < 55 HRC with a really less than ideal geometry and grit finish can still slice ~1000 pieces of 1/2" hemp and readily slice newsprint then how important really are tests on edge retention on a slice for steels

-how can you combine edge retention on a slice and push together to get some kind of total edge retention statistic and what does this mean in terms of what would be a kind of ideal finish

Of course there is always time/money constraints. I would like to actually see what would happen if I did the same work with AEB-L, ATS-34, S30V, M4, 121REX but that isn't practical.

[demoncase]

So.......Did we get down to what is the best edge and best steel then? :)

I'm reminded of some of the debates that go on regarding processer speed and Android Vs Apple....Am I going to notice that such-and-such device is 0.0000001s slower to load that webpage?

Am I really going to notice that my otherwise decently sharpened and robust mid-range Spyderco is not at the absolute outer limit of ultimate scientifically proven sharpness?

Either way, I'm one of them weirdo heretics that like serrated edges. :D

[jabba359]

So.......Did we get down to what is the best edge and best steel then? :)

I believe the conclusion was Chinese 420J2 with the faux Damascus screenprint at 46 HRC, 12° inclusive (no microbevel) finished at 60 grit.