Whats The Forums Opinion Of Mad Dog?

Bruce,
I agree with you about no middle ground on Maddog.It is somewhat similar to the Sebenza controversies.I also agree with Steve Harvey about the truth being somewhere in the middle.The information is out there for anyone to read.If someone feels a product is worth a $1000 and has the money to pay for it, more power to them.Other people may think it's worth $100 but they shouldn't call the $1000 people naive.If Maddog can charge the prices he does and deal with all of the controversy surrounding his knives, more power to him also.He seems to have more business than he can handle, so for now, he must be doing something right.There always seem to be 4 types of people in Maddog threads:

1.People with legitimate gripes

2.People who are satisfied customers

3.People who are jealous

4.People who want to stir the pot

The people who haven't owned Maddogs usually fall into the last two.(Which one am I? I'm jealous of people who can spend that much money on a knife.) As I have already said, I have my reasons why I won't be buying a Maddog product.If someone asked me in person or in a private e-mail why, I would be more than happy to tell them.But since I have no personal experience with him or his products, I will do my best not to state my opinions publicly.

Jim

[This message has been edited by HUNTER3897 (edited 04-21-2000).]
 
Man oh man- six pages of this stuff!

I sure won't miss this kind of discourse when I'm not in front of my PC all the time during the coming warm weather. For myself, fall and winter are for knives, spring and summer are for motorcycles. Enjoy your discussions, I'll see you all in the fall.

I'm going outside to play.

clip
 
Bruce - two tests is enough if you're a Bayesian
wink.gif
Ha, all those statistics courses finally paid off!

I work down here in good ol' St. Mary's County, Maryland at what used to be called the Naval Air Test Center. When we test the more expensive items to structural failure we usually only test one. You back off from that abuse level by some amount and say, "we're won't abuse the rest of these any more than this." Testing to failure gives you a good idea of how far you can safely go without a catastrophe. Are Cliff's MD tests interesting? Darned right. Are they useful to some people? I assume so. Are they useful to me personally? Not really, since I don't plan to use my knives the way Cliff tests them. Around the house I pry with my pry bar and chop with my hatchet.

No, I don't own a Mad Dog.

Yes, I am jealous of the people who do.

Yes, I do own expensive knives and hand made non-knife objects, and yes, they are worth the money even though similar, less expensive products might be "functionally" superior.

I guess my only point is that one or two tests to structural failure are valuable, and if you like something it's worth the money. Okay, that's two points. You gotta cut me some slack, after all I'm sitting here flicking knives and drinking Guiness. mmmmmm

David
 
COL Woodbury,

I deal with statistics and the dynamics of population sampling on a daily basis, but I know you have had much more experience in testing and evaluation than I ever will (you da Tabasco man
smile.gif
). While I agree with what you've said regarding the statistical validity of the tests, I don't recall Cliff or anyone else making the statement that their tests showed conclusively that all MD knives were bad. Those conclusions have been inferred by some readers. Their reports are closer to a case report used in the medical and other fields. Patient X presented the following symptoms..... Any conclusions that are drawn carry the caveat that they are made on one or two cases, not the general population. Ideally, any unusual findings should be followed by a more rigorous test design. What I gathered from Cliff's test was 1) He bought and subsequently tested (as only Cliff has the cahones to do with a $$$ knife)the TUSK based on the claims of the manufacturer. 2) It failed the test, and thinking it might have been a lemon, he returned it. 3) MD replaced the knife and all should have been well if knife #2 was what it was advertised to be. 3) It failed what seemed like a perfectly reasonable test. 4) Will MD replace it and let Cliff have another go? Apparently not, and I can't say that I blame him. If a third one fails, doo doo. If it passes, good, but MD is still out of the $ and he's got pups to feed.

Conclusions from 2 cases: The TUSK is advertised as a durable utility knife that can and has done it all (my paraphrase), but it failed tests that other knives passed. Based on these 2 cases, the truth doesn't appear to match the manufacturer's claims. Is this an epidemic in the MD knife line? Hard to tell unless more cases present themselves, but as a consumer or manufacturer, I'd think about it.

Jeff
 
bruce- I am going to post to separate thread about the size of sample.

There is a middle ground on the MadDogs, and I am on it.

They are what they are.

------------------
Thank you,
Marion David Poff aka Eye, Cd'A ID, USA mdpoff@hotmail.com

Talonite Fire

"Many are blinded by name and reputation, few see the truth" Lao Tzu
 
Bruce :

A meaningful sample would probably have to be around 30 knives, picked at random from across the many times that knife was produced.
Click to expand...

You are not trying to quantify the detailed nature of the parent population, all you are trying to do is estimate the mean behavior to be expected in any given blade. If you needed that large of a sample size to estimate the average performance of a knife it would indicate that the maker was incompetent as the variance in performance from blade to blade would have to be huge. And of course as other people have posted, you are ignoring another critical point, if McClung didn't think the performance I saw was to be expected - why didn't he just give me another TUSK so I could see the real performance. Would not that be the logical thing to do? Would not be be bound to do so - as I what I bought and paid for turned out to be a substandard blade.

The idea that you need such a large sample size is not sensible. Every day people collecting data (or doing anything in fact) have to make such decisions about methods, changes they can make and so on which may or may not improve the quality of the data they are getting. I know of no one who would run thirty sets of experiments to see if a method change was significant. Nor as elementary statistics would show is it necessary. Do a sample confidence interval and see what the odds are at drawing one and then two points at random and falling outside the 3 error range.

Dan, thanks for being specific about the wire I think I will start using a set stock from now on. Note concerning the Bucks, edge durability was a big factor in the steel choice, this was discussed on the Buck forum awhile ago.

Cobalt :

Some people say that impacting the spine led up to the damage on one of the knives.
Click to expand...

This is actually a serious slam on MD knives, if McClung is heat treating his O1 so that at 52 RC is is not as tough as D2 and 1095 at 56-58 (the other blades I was hitting), he is doing something seriously odd to the blades. That is just a red herring and another example of the "fake MD" excuse that was given to Spark.

David :

I'm sitting here flicking knives and drinking Guiness
Click to expand...

Which is an excellent way to start any statistical analysis. I would never be without at least a dozen on hand in case a serious problem turned up.

-Cliff


[This message has been edited by Cliff Stamp (edited 04-23-2000).]
 
Grif, if you're there, could you please a review of the Saxon?
There's too much talk of looks these days.

TIA,
Smoke
 
From a gut feeling I'd have to say that 1 or 2 knives is not really a scientifically adequate sample size by any stretch of the imagination. And I doubt that Mr. Student (of the famed student's t test would think so either-btw, his name wasn't student either but he sure knew how to brew Guiness--but I digress).

What really USUALLY determines sample size is power, i.e., how powerful your analysis is. And there are all kinds of fancy ways of determining power but there is also a subjective side to power, i.e., most researchers, through years of experience with a particular type of test and material involved and by examining the literature for similar experiments, will know pretty well how big a sample size they need to detect treatment effects at the p<0.05 level. Obviously, the more variance you have in something, the more power and larger sample size you need to detect that something is different from "average" or from something else.

{Edited remark: I forgot to mention that if your power becomes TOO great, you make a different type of error. You detect differences that are irrelevant or insignificant. I don't think that's the case here.
smile.gif
}

I don't think either of these things are cleary being tested here in a statistically rigorous manner. But the tests are rigorous enough, and Cliff has the kind of experience that I feel that these tests are qualitatively useful. Additionally, if 2 knives in a row exhibit similar properties, this at least suggests that variance is LOW, i.e., repeatability is HIGH. But of course, I could be wrong and the next ten could all be different--but it seems unlikely. Especially when there is some anecdotal evidence as well as other tests that have been performed to support his tests. You might not be able to publish this in the Journal of Knife Testing but at the very least, there has been enough useful data presented to give one pause.
smile.gif



------------------
Hoodoo

The low, hoarse purr of the whirling stone—the light-press’d blade,
Diffusing, dropping, sideways-darting, in tiny showers of gold,
Sparkles from the wheel.

Walt Whitman

[This message has been edited by Hoodoo (edited 04-24-2000).]
 
Giving it a little thought, I am in agreement with Cliff. Two TUSKs is enough to learn something important about them. My problem is, not having been present at the failures, I will never be quite sure how meaningful the events were to me personally. Cliff has done his best to describe everything in detail, but doubt still remains in my mind whether I would ever push a knife to the extent that he did, in the same way that he did.

A couple of different reactions to the TUSK failures seem reasonable to me. If you were the owner of the failed $800 knife, and it was pretty much the only Mad Dog you had experience with up to that point, I think it not surprising that you might have a poor opinion of the knives. But, if you were someone who had owned half a dozen Mad Dogs over a period of several years, and never had any problems with them, you might be merely puzzled by the failures, as I am.

To really get my own grasp on the issue, I would have to buy an ATAK, which I consider to be the real baseline for Mad Dog knives, and do some heavy levering with it. One of the reasons Mad Dog has given for the $800 price of the TUSK is the difficulty in getting a satisfactory heat treat on the bigger blades. Maybe the 10" blade allows such leverage that the extreme heat treat that he uses is not the most durable solution. There is more to a knife than simple strength, but maybe the TUSK is not supreme in that particular category. I consider the ATAK the better format for Mad Dog's formula, but that is only an amatuer's opinion.

So I think Cliff's experience was meaningful, I am just not sure in what way, at least for me. I like knives with hard edges. Nothing I have read has changed my mind about the fact that Mad Dog knives work really well for me.

It seems to be the perception of a number of people that Mad Dog has claimed that his knives are the best in every way, and indestructible in the bargain. I admit to having assumed that was true myself at one time. I never really contemplated putting enough lateral force on a knife blade to break or bend a knife ground out of .25" steel. Now, in the light of everything that has happened, it seems perfectly logical to me that another blade, perhaps saber ground out of uniformly heat treated tool steel, with a greater cross section, would be stronger. But thinking further, I always arrive at the same conclusion. The thicker they are, the stronger they are, but no knife is indestructible, and at some point, the blade ceases to be a very good knife. Mad Dog knives may not be the strongest knives on the planet, they are simpley the best expression of Mad Dog's personal vision of the perfect knife that he has been able to produce. I know that is not going to satisfy those who feel Mad Dog should be punished for strongly believing his knives are the best knives, but oh well.

 
Regarding the TUSKs, I think it is significant to note that the blades were far from sequential; demonstrating an habitual heat-treat pattern. While it might work for a shorter blade, the longer ones of approximately the same cross-sectional area may be a problem.

This may also be one of the reasons that the Bladesmiths of Nepal (who zone harden all of their blades) use significantly thicker spines than we do in the west for any given length of blade, ie. as the length increases so does the cross-sectional area. And as Cliff and others will attest, the blades sold by Himalayan Imports just wax the other contenders generally.
Dan
 
Hoodoo, 2 knives tested to failure is more than a representative sample. Failure testing on any product is not done in high quantities. This is true in all mfg industries. How many S-class Mercedes do you think are crashed into walls? Very low percentage. When you are talking normal load testing to 100% or to the additional safety factor included in the design, yes, there will be a good sample size, but that is not normally to failure.

DanK, I don't really know if it matters that the HI khukuri spines are soft or not, when they are almost 1/2 inch thick.

Steve, I agree with you that the ATAK is probably much tougher than the tusk, much like the CS Recon Scout is tougher than the Trailmaster. The shorter blade is always tougher. However, the edge chipping and/or fracturing on the blades is an issue of heat treat. I would like to see some pics of the ATAK's that were chipped or had edge damage and how they got it. I think both Will and Mike had that problem.

Cliff, I find it hard to believe that Rc52 is not going to be tough. MD himself states that heat treating the tusk is not easy, hence, the additional cost of the knife. Differential heat treating is not easy as you probably know, and the imposed stresses within the metal may be a big part of the problem if not relieved. This is all guess work of course, but it almost sounds to me like that first tusk broke along a temper line, could that be(this is question, not sarcasm, I just want to know were it broke)?

Also, Cliff did you get the second knife back since, MD did not replace it or pay you back for it? I assumed he did, since he has no right to the blade if he doesn't replace it.
 
Originally posted by Cobalt:
Hoodoo, 2 knives tested to failure is more than a representative sample. Failure testing on any product is not done in high quantities.
Click to expand...

Cobalt, all I can say is: prove it.

------------------
Hoodoo

The low, hoarse purr of the whirling stone—the light-press’d blade,
Diffusing, dropping, sideways-darting, in tiny showers of gold,
Sparkles from the wheel.

Walt Whitman
 
Cliff, you are right that variance (standard deviation) is important here, very important in determining the importance of the mean. But you are putting the cart before the horse. How do you get a mean and standard deviation? By testing. You must first test a significant number of items (not one or two) to develop the analysis of variance -- then you can use this as a predictor or future results.

What was the mean and standard deviation you are comparing the two Tusks to? As far as I can see, you were comparing the two knives to Mad Dog's statements of quality, only.

Spark, as I've said before, I agree with you that testing a large number of expensive items is not the easiest thing to do. All I said is that if we aren't going to test a significant number of items, let's stop drawing faulty conclusions from the incomplete testing.

Cliff, I think it's important that you state whether your test design was intended to test to destruction or not.

Bruce Woodbury
 
Bruce,

We may not have enough data to draw a conclusion, but we certainly have enough information to raise a concern. If this maker (or any other) wants to address the concern he can either run a controlled test on a statistically significant sampling, or put his money where his mouth is and up the warranty. I don't think its appropriate for a third party to prove anyone's marketing prositioning.

Nor, should we diminish the value of the information we have. Its inconclusive, but indicative of the problems that some users may encounter. While we cannot say that there is a problem with every knife we can certainly say that there are problems with some of these knives.
 
I wouldn't call it faulty conclusions, I'd call it pretty certain conclusions - for a knife that is advertised as being capable of doing it all, and having done it all, the TUSK didn't live up to it's claims.

How much of the population will follow the same behavior? That's an uncertainty, but not one I'm willing to bet against given the $900 price tag of the knife. Maybe that's why we aren't seeing any more 9" utility models coming out, but plenty of "fighters"... but who knows for sure.

Spark

------------------
Kevin Jon Schlossberg
SysOp and Administrator for BladeForums.com

Insert witty quip here
 
It is my pleasure to announce Mad Dog's newest small fighter: the Bayou Hunter, a 9" version of the ATAK. I have one on order, and it should arrive in a few weeks. Can't wait.

Bayou_Hunter.jpg


[This message has been edited by Steve Harvey (edited 04-24-2000).]
 
Hoodoo, the proof is in your ballpark not mine, bud. The only proof needed now is for those who think a Tusk won't break to prove to themselves. Right now the record is 2 and 0 against it. As far as I'm concerned the tusk is no different than an Ontario Bagwell Bowie, with less reach, less corrosion resistance, less penetration in soft tissue, six times the cost and just as apt to break in survival mode.

 
Steve, I don't have, in general, a low opinion of McClung's knives. As I have illustrated many times, they have several very high performance features. I had a Lab Rat on order at the time and indended to keep it even if the second TUSK failed as the stresses were not relevant to the use the Lab Rat would see. For its design I think the steel and heat treat is a good combination. I cancelled my order because of McClung's reaction because there is no way I would support someone like that, which is the same reason I gave away my Operator.

Bruce I have rather specific performance expectations when I buy any knife, however this is always discussed with the maker/dealer, on the rare exception it is I will take responsibility for it. McClung knew exactly what I was going to do, and as I stated before, I was told of far greater abilities on every aspect of which I discussed the performance. If the intent of the work was to just break the knife I would have just done the 4x4" pry first.

As for the stats, you are also ignoring the consequences of your argument. If the variance in the population which represents the blades performance is so high that a large sample is needed to deduce the mean behavior then the maker is incompetent. Read some makers comments in the thread Marion started.

As for the standard of 30 test items, one of the first things you learn in statistical analysis is how to adjust from nonnormal behavior in small sampling because large samples are not practical for day to day work. Awhile back we ordered a small cylinder of very expensive gas, when it was cooled it ruptured at the nozzle. We did not buy an extra 29 bottles and see if they all failed. This one failure was reported which resulted in a change of the design to handle low temp. cooling. They also replaced the gas we lost because of the rupture with a new bottle.

Cobalt, no he still has it. I have emailed him about it but just got excuses. It is well over a year now. My only concern in this regard is how it is returned if it ever is. If it is labeled incorrectly I could easily get hit with hundreds in duty/customs charges. I have of course detailed how it has to be shipped. But there was a mistake on how the first replacement was sent out that set me back as few dollars.

-Cliff

[This message has been edited by Cliff Stamp (edited 04-24-2000).]
 
Originally posted by bruce:
Cliff, you are right that variance (standard deviation) is important here, very important in determining the importance of the mean.
Bruce Woodbury
Click to expand...

Bruce, I doubt this is ANOVA stuff. More like
Chi Square. It's counting data, i.e., fail, no fail, or +, -, good, bad. There might be some aspect of it that is continuous data and therefore could be analyzed by a t test or perhaps a nonparametric variant of it such as the Mann Whitney but the parameter of interest usually should be determined in advance of the testing to be statistically valid (unless you use nonparametrics) but there are exceptions. Anyway you look at it, n=2 will not make it in any test in any statistical software that I own (SAS, JMP, SPSS, and Systat). But that still doesn't rule out the qualitative value of the data. Here's a cold hard fact: not all data lends itself to statistical analysis nor does it have to in order to be useful. Contrary to what most people think, statistics are very rigorous and their use is LIMITED to specific kinds of data. Certain assumptions HAVE to be met. If they are not met, then it is not appropriate to use statistics. That would be an abuse of statistics, not a use of statistics. And by statistics, I don't mean numbers, I mean the statistical analysis of data.


------------------
Hoodoo

The low, hoarse purr of the whirling stone—the light-press’d blade,
Diffusing, dropping, sideways-darting, in tiny showers of gold,
Sparkles from the wheel.

Walt Whitman
 
Originally posted by Cobalt:
Hoodoo, the proof is in your ballpark not mine, bud. The only proof needed now is for those who think a Tusk won't break to prove to themselves. Right now the record is 2 and 0 against it. As far as I'm concerned the tusk is no different than an Ontario Bagwell Bowie, with less reach, less corrosion resistance, less penetration in soft tissue, six times the cost and just as apt to break in survival mode.
Click to expand...

Cobalt,
I can easily prove it statistically. n=2 is invalid in any statistical test I know including the student's t distribution which was designed for small sample analysis.

I'm not disputing the qualitative value of Cliff's testing, but the reality is that mathematically, nothing can be said about the variance in a population with a sample size of 2 that would be statistically valid. Now if he had 5 samples, maybe even 4, we might be able to make a go of it. But 2? It just ain't gonna fly. Can we make inferences? Yes. Statistically, this is preliminary data or at best, incomplete data. Can we make statements with what passes as statistical validity? No.

You name the test. You provide the null hypothesis, and I'll be glad to run the statistical analysis. I know lots of people with incomplete data sets that would love to test their data with a valid statistical test so if you know of a stat test I haven't heard of that can be used on sample sizes of 2, I'd love to know about it because I could sure use it.


------------------
Hoodoo

The low, hoarse purr of the whirling stone—the light-press’d blade,
Diffusing, dropping, sideways-darting, in tiny showers of gold,
Sparkles from the wheel.

Walt Whitman
 
You must log in or register to reply here.
Back
Top