Come 1, Come all, to our first forums Whackoff

not2sharp

not2sharp

Platinum Member
Joined
Jun 29, 1999
Messages
20,499
How do you feel?

There are many good knife reviews out there, and I usually like to read every review I can get a copy of, yet most of them tend to leave me somewhat unsatisfied.

Almost all of our reviews are weak in two respects:
1) The reviews are not scientific. We seldom, if ever, see someone test a knife, even in the most basic sense, and I have never seen a knife test involving statistacally valid sampling quantity.
2) Most reviews fail to provide a sufficiently clear perspective. Yes, the knife performed well, or not so well, but how did it compare with other similar knives available in the marketplace? Should we buy this knife over another?

Understandably, cost and time constraints, make it virtually impossible for even the best intentioned and most knowlegeable of us to overcome these two limitations. However, we as a community of knife enthusiats can and do have the ability, and resources, to collect, assemble and correlate the required information. We can overcome these limitations and conduct scientifically conclusive testing, collect the data, and draw meaningful conclusions.

Far starters:

We can start by testing for something simple. Perhaps we can look at edge retention (i.e. examine blade design, steel type performance, and quality of heat treatment).

We can start with a common and basic reference knife, perhaps a Buck110, and use the results to establish a reference benchmark for performance criteria.

Once our data begins to build we can develop additiona reference tests and continue to run new knives through our testing protocols.

The test:

1) 100 members volunteer to conduct independat testing of a specific knife and model.

2) The testing is simple, cheap, and performed under controlled conditions:

a) Startiung with an unsharpened factory edge on a Buck model 110 plain edge folder.
b) Given a certain type of rope (agree on a Home Depot sku for the rope), and a certain cutting surface - like a pine wood board,
c) Count how many times you can cut through the rope, without sharpening the blade, and using only hand pressure.
d) Each participants send their results via E-mail to a moderator who volunteers to consolidate our results. In order to reduce bias, participants do not share test results with each other.
e) The highest and lowest 10% of the results are eliminated to reduce distortion.
f) The results are published and we again ask for volunteers to test another model.

If we do this for some of our popular knives we can at least develop a criteria for an informed discussion.

Who wants in?





 
That's a great idea nonbox,

1. I appreciate the tests done by everyone regarding spine whacks, toilet-paper cutting, etc. However, they often lack validity problems.

2. Worse yet are subjective claims plagued by personal biases. For example, many people swear by Sebenzas. However, it would be interesting to see just by how much a $300 Seb is better than, say, a $100 Benchmade Pinnacle 750 when it comes to measureable, quantitative specs. like lock strength, blade strength, scale strength.


[This message has been edited by Full Tang Clan (edited 10-14-2000).]
 
Catchy title
biggrin.gif


Sounds like a good idea, but it could be limited by our wallets. It would be hard to find 100 members who have, or are willing to buy the aforementioned Sebenza, NIB.
Sounds like it would work for some of the more popular models like the LCC, BM Axis locks, Spydercos-stuff we can afford, and want anyway.
I would make an effort to participate in this. Sounds interesting.
 
Actually, we should be able to defray the knife expense by volunteering knives that we already own. All we have to do is to volunteer to remove our favorite knife from our pocket and cut something with it. Then we share the results with others, and they likewise tell us about their knives.

Between the 8000 of us we probably own something like 300,000 knives in a 40,000 different patterns (inc. many antiques,military, and custom items). We simply need to look at some of the more popular models (concentration greater than 1,000/300,000) and pursuade 10% of our members with those knives to actually use them and share information with us.

Again, the most informative test in this format would be simple to conduct, inexpensive, safe, and non-distructive. we can start with something simple, for instance starting with a factory edge (it probably wouldn't make a differnce if we start with our own edge) count how many times you can cut through a specific (need to agree on a part number) 1/4" cord. For simplicity we can assume that the entire blade edge is used to effect the cut. I would suggest we start with very common knives such as the Buck 110 and Spyderco Endura. Once we have mastered the technique it will be easier to test more expensive knives reliably using smaller, hence easier to obtain, sample sizes.

I would be willing to participate with either of these knives and would in addition volunteer to help tabulate the data.

I am not a statician, however, so if anyone wants to crunch the standard deviation and significance levels you are welcomed.

To me this sounds like an interesting way to leverage and involve our community in a worthwhile endevor. It should be a good deal of fun ....and to keep it that way we should leave the Rabid Dog(I'm sure most of you get it)products for last.
smile.gif
 
As stated above I for one volunteer to participate.

We need 99 more are there any takers?
 

I like the idea and I believe it is a good one. If all the test will not be real harmful to the knife(like seeing which knife breaks first under same pressure), it sounds great!

I will help when I can!
mason(FLY GUY)
 
Thanks Fly Guy that only leaves 98 to go.
smile.gif


BTW welcome to the forums.
 
Are We testing the Buck 110 or whatever we have? Also, none of my stuff is factory edge anymore, does this count?? I have an EDI Genesis 2, 2 Tramontina kitchen knives, a few kit knives from Jantz, a bowie and assorted blades. Can this help???
 
First up is the Buck 110, but feel free to suggest other knives. just keep in mind that the objective is to find ~100 examples (100-20(extremes)=80 which is about twice as many as we theoretically need for an accurate distribution). We can decide which knife to test after the Buck 110.

If we are testing edge holding (steel + blade geometry + temper) then it probably wouldn't matter if the knife was lightly sharpened. You should get a very similar result, unless your knife has very significant blade wear to the extent that the profile and length have been altered.
 
I think this is a good idea, but I still don't understand how the sharpening issue can be neglected. It seems to me the only real not destructive tests that can be performed at home are edge retention and edge cutting performance test (Not many people are going to agree to strength tests whether it be lock strength, tip strength or handle strength). The cutting performance tests would be drastically affected if I were a terrible sharpener of knives--which I am. I always can't get rid of burrs, so if I were to compare a factory sharp cold steel knife to my buck ranger, the CS would win even if it wasn't actually superior in terms of cutting performance. And for the edge retention test, there is a bit of subjectivity in judging when a knife has lost its edge. For example, I have a friend who used to never sharpen his knives, and he just sort of forced his way through whatever he would cut. To him that was a normal clean cut, but to me it was more a rip than a cut. So he would probably continue to cut through hemp rope long after I deemed the edge unacceptable.
 
Can I respectfully make some suggestions?

1. If I'm not mistaken, the law of high numbers suggests that the benefits that come with exceeding thirty samples are small, as long as the samples are random. If all you want to ask is "does this maker's steel hold an edge better than that maker's steel" all you may need is 30 participants. Someone who has a better grasp of stats than I do may want to comment.

2. A way around the factory edge vs. resharpened edge may be to use difference scores. Test the knives in pairs. For instance, if you want to compare a BM AFCK to a Spyderco Endura, everyone who has a AFCK, an endura and a Buck 110 is asked to participate. Everyone puts their own "home" edge on all three knives. Do the cutting tests suggested above then subtract the number of cuts the 110 performed from the number the AFCK and endura each performed. Presumably all three knives started with the same edge so essentially the 110 acts as your baseline or zeropoint. Send your difference scores to someone and have them do a dependent t-test comparing AFCK-Buck 110 vs. Endura-Buck 110. If I'm correct, the results will tell you which knife, AFCK or endura, holds a better edge.

The other option would be to send the test knives to a central source and have that person do all the sharpening but that probably defeats the whole purpose of the experiment.

Didn't an old Maddog thread (I think it was "the" Maddog thread)get into using stats to validate tests?

Switchback

[This message has been edited by switchback (edited 10-15-2000).]
 
I'm in, Got a SPyderco Endura, CRK M16, MOD CQD4 and a Tac-11. Just let me know what to do
smile.gif


Have fun! www.gogetemgear.com

------------------
Keep your powder dry.

Dan

[This message has been edited by Javahed (edited 10-15-2000).]
 
Generallobster,

In principle I am in complete agrement with you. Under ideal conditions we would effect complete experimental controls and take our sample test through to full distruction testing. This would probably yield the most complete set of performance data. For most of us though, I suspect that we would be mostly interested in observing and comparing performance within a more practical operating range. Distruction testing would yield information of limited practical use and would of course be expensive and most of all dangerous.

A simple cutting test yield a limited amount of information. This information though can be used as a very practical litmus test on knife performance. Anyone can break any knife, what I want to know is whether a knife will perform well when used, not when abused.

I have intentionally selected an oversized sample set because we intend to eliminate some of the extreme performance results. There are two key elements to this test:
1) The participants will not have access to the results scored by other participants until the results are tabulated.
2) The upper and lower 10% of the results are eliminated to yield a more representative range of scores. The intent is to address distortion due to differences in sharpenning skill and other related factors. Steve Schwarzer had an interesting article in the January 2000 issue of Blade "Knife Tests: Cutting Through the Hype", in which he talks about some of the distortions that can come into play.

I tough part in any statistical sampling is to obtain a honest representative set of data. By using a slightly oversized set and eliminating some of the more distorted result we should be able to get a pretty reasonable set of points.

Keep in mind that the test will yield a performance range rather than an absolute result.

I have recommended the Buck110 for the simple and expedient reason that I believe this knife to have the largest representation within our combined collections. If so it would be a good starting point for this exercise. If it works we could them move on to other popular knives ASAP.

If we can find the volunteers we could probably do the testing next weekend (the weekend of October 21-22. I will open up a seperate thread in the Blade Discussion Forum for a quick rollcall on the 110. Feel free to open up similar threads on other knives if you belive we can obtain ~100 participants with that knife.

N2S
 
I will play up to the point of destructive testing then I back out. I WILL NOT attempt to break any of my knives
rolleyes.gif

I have a very old Buck 110 with my modified edge if this will qualify.

Ron

------------------
Ron,
Bremerton, Washington
 
Let be clear on this. Whoever participates in this will need to resharpen their knife. That is the full extent of the damage we intend to cause. Distruction testing, or anything like distruction testing is not required, and will not be measured.
 
Well
wink.gif
Sounds interesting
biggrin.gif

I have around 70 knives currently of 60 patterns probably, will be interesting
smile.gif
))


------------------
zvis.com
Have Fun,
Alligator
 
Regarding the test...

I did a little shopping today and this is what I have come up with.

Walmart currently carries 1/4" Sisal cord which should work out well for our test. The cord comes in 50 and 100 foot rolls, is manufactured by Lehigh under part number 8010L (bar code 0-71514-06505-2); price is under $4.00 for the 100' roll. If everyone who wants to participates in this obtains some of this cord and uses a telephone book as a base we would share a simple and inexpensive test medium.

The objective is to see how many times your knife will cut through the cord before it fails to cut. The entire length of the blade can be used to effect the cut, and a knife will fail to cut if it does not completely sever the cord in a single pass. When this happens twice in a row [note: start with a fresh piece of cord on each pass] we stop and tally up the count.

This is a simple test, but, I expect that it will be interesting and informative non-the-less.

 
Lets us know if you want to participate. We will try to get this together over the upcomming weekend.
 
You must log in or register to reply here.
Back
Top