Clamp-Width: The Blade Chatter Metric

[Iridian]

The Blackland A10... Blade goes Brrrrr

Haha, they for sure won't name a razor the "Warthog", but maybe after the Lockheed AC-130 gunships "Spectre" would also be an option. The handle could be modeled after the M102 howitzer. SCNR.

 

[spacemonkey42]

I love discussions about what we all have tried, what we like, what we don't, all of it. Add to this the common caveat of YMMV. All of this is so subjective, one's individual scores are really only useful to that person. I'm wondering if crowd-sourcing and averaging the subjective data would dilute the results.

A thought for possible consideration - what if instead of crowd-sourcing the result, you open-sourced your method? Here's a thought... You could further detail your method for everyone to use. We could help gather the first four columns (gap, exposure, clamp width, weight) as these are facts. Those with good tools could measure them accurately. The rest of the columns are entirely subjective. With your method and a nicely organized sheet, anyone could apply it to their razors and their results. That sheet could have the formulas and graphs ready to go for users who might not be so spreadsheet savvy.

I just wanted to share a thought in case it resonated with you.

 

[albsat]

Whenever I read a thread on blade chatter I have an uncontrollable urge to bust out a Wardonia.
View attachment 1969552
No chatter, nicks or irritation. Just glorious audio feedback, a super close shave and a big smile. What magic is this?

I got the same thought when I read this thread. I haven't used my wardonia in a long time.

I use it with a fatip grande cap and handle. If you do have a Fatip Grande razor, can you try it with my setup and find if there is any difference from the original wardonia in the shave experience?

 

[Teutonblade]

As razor performance is very blade dependent the biggest sampling challenge will be to have respondents use the same blade.

 

[JPO]

I love discussions about what we all have tried, what we like, what we don't, all of it. Add to this the common caveat of YMMV. All of this is so subjective, one's individual scores are really only useful to that person. I'm wondering if crowd-sourcing and averaging the subjective data would dilute the results.

A thought for possible consideration - what if instead of crowd-sourcing the result, you open-sourced your method? Here's a thought... You could further detail your method for everyone to use. We could help gather the first four columns (gap, exposure, clamp width, weight) as these are facts. Those with good tools could measure them accurately. The rest of the columns are entirely subjective. With your method and a nicely organized sheet, anyone could apply it to their razors and their results. That sheet could have the formulas and graphs ready to go for users who might not be so spreadsheet savvy.

I just wanted to share a thought in case it resonated with you.

That's a great suggestion.
You can also to some extent calculate the rigidity, given some simplification, if you have the measurements.
You can also look at what difference just the blade width and thickness can have. A small change in either of these variables can make a big difference.

 

[Lane101]

As razor performance is very blade dependent the biggest sampling challenge will be to have respondents use the same blade.

Yes and additionally even blade performance itself is highly subjective and variable per beard characteristics. My recommendation would be to get agreement on a few key parameters such as yours or the outcome based suggestions in my earlier post. Going beyond that would be a challenge to execute outside of a controlled lab environment.

Note that if we include classification blades for slice and dice analysis the number of respondents required for a meaningful sample size would be significantly higher.

 

[creamgravy]

I use it with a fatip grande cap and handle. If you do have a Fatip Grande razor, can you try it with my setup and find if there is any difference from the original wardonia in the shave experience?

Unfortunately I don't have a Fatip (yet) or a standard pin baseplate Wardonia (yet) but it sounds like a good combo.

I'm still testing blades (0.13mm carbon tigers on the way thanks to Helicopters mega blade thread!)

 

[BigAlVista]

I think I should rewatch Babylon 5.

Loved that show.

 

[Ouelljas]

I love reading these threads. These are the kind of threads where all the really experienced members, the cream of the B&B site so to speak bring a lot of valuable knowledge and information to the table, yet somehow it’s all done in the most gentlemanly and respectful way towards one another. I’ve learned so much stuff reading the first three pages of this thread! At the end of the day, I still judge the razors I’ve tried very subjectively based on feel, but I do love learning about all the technical aspects of razors.

 

[idjekyll]

Haha, they for sure won't name a razor the "Warthog", but maybe after the Lockheed AC-130 gunships "Spectre" would also be an option. The handle could be modeled after the M102 howitzer. SCNR.

Hot damn! Blackland Spectre does sound badass. Add black anodizing or cerakote finish and you might as well just take my money already.

 

[Guido75]

If anyone wants to double down on statistics and shave attributes, you might want to check out this database design by @Stikeyoda.

Creating a Shared Razor & Blade Research Database Shave Journal

Hi All, This started as a study of Chinese blades. As I worked on how to gather and analyze the data, I realized that with a little more work, it could be turned into something more valuable as a research tool and more useful to members of Badger & Blade. Attached is a PDF diagram of the second...

www.badgerandblade.com

 

[Stikeyoda]

If anyone wants to double down on statistics and shave attributes, you might want to check out this database design by @Stikeyoda.

Creating a Shared Razor & Blade Research Database Shave Journal

Hi All, This started as a study of Chinese blades. As I worked on how to gather and analyze the data, I realized that with a little more work, it could be turned into something more valuable as a research tool and more useful to members of Badger & Blade. Attached is a PDF diagram of the second...

www.badgerandblade.com

It's in the midst of another revision - I learned a lot with the prototype - both a need to make it finer grain (separating brushes into knots and handles - even knots into mixed bristle types, separating razors and handles - how to account for multiple plates and adjustable, etc.) but easier to use (more generic defaults for people who want to use the diary part.
Also turning into a lot more work to implements. The basic components haven't changed much - now just need time)

 

[SaguaroSmart]

Statistical inference has some value in engineering. However, as an engineer i would never board an airplane designed by a statistician.

That's funny. Probability and statistics are key to assessing any type of risk for any kind of operations (OR), military, financial, engineering or other types. I would not place them (statisticians) in an engineering design role, but in the case of a plane, it turns out Boeing or other companies could most likely use more statisticians to assess risks.

A big example that comes to mind is the O-rings failure on the space shuttle Challenger that led to its explosion. Turns out statistical analysis of the previous 23 launches revealed that the probability of having O-rings damage under a certain temperature was close to 1 ... This resulted in the loss of 7 astronauts and millions of dollars of taxpayer money.

That being said, I see a few issues here with the way statistics are used:

TLDR:

• Sampling technique: sample is very small and not random. For people interested look up (systematic sampling, stratified sampling or cluster sampling) it turns out this is an entire subfield of statistics called Sampling Theory.
• Linear regression is used whereas the data most likely violates the assumptions necessary to use such a model

More details:


To use linear regression, the data and model must not violate linearity, independence, normality or equal variance.
In the case of recording shaving scores daily or every 2 days, the big issue I see with it is the data observations are not independent from each other. After all, if you had a bad shave the previous day, it will impact the observation of the next shave, similarly if you had a good shave the previous day, you are having an easier time shaving the next and are more likely to have a satisfactory result. This means time independence of observations is violated. In turn this means linear regression is likely to yield poor results for modeling the phenomenon.


Ironically such a simple thing as wanting to model a good shave statistically becomes more complex than one would think. Since the observations are time dependent (meaning autocorrelated at lag 1 or 2) but the model intends to also rely on other predictors (Clamping-Width, razor category: edge clamper vs blade bender and other potential predictors). The only somewhat simple models I can think of are either using a time series model that allows for additional predictors (ARIMAX type model) or violating LINE assumptions with a multiple linear regression including lag 1 and 2 terms as predictors (can sometimes still yield decent results experimentally). Anything else would either involve more advanced modeling techniques (other machine learning algorithms or non-parametric methods).


Long story short, this is not impossible to do but might prove difficult for someone outside the field of statistics / data science.


PS: I sympathize with the intentions of the Author. The theory proposed is interesting. As a data scientist myself, I am intrigued by relationships in data and it would be interesting to find a way to model what features impact a good shave, even if just for the fun of it.

 

[Bhugo]

I am betting that Gillette has/had a few piles of data regarding many of the attributes discussed in this thread. It would be interesting to see how they made decisions on design.

 

[BigAlVista]

I am betting that Gillette has/had a few piles of data regarding many of the attributes discussed in this thread. It would be interesting to see how they made decisions on design.

We will see that research right around the time we see the formula for Coke.

Last year, I reached out to P&G, Gillette and AOS to see if the had any marketing materials for their many razor blades, so we might see how they position their products. I had hoped to proved this to @helicopter to incorporate into his research.

P&G was like. "Thank you so much for your interest in our products. We sincerely hope you take this opportunity to go pound sand".

 

[JPO]

That's funny. Probability and statistics are key to assessing any type of risk for any kind of operations (OR), military, financial, engineering or other types. I would not place them (statisticians) in an engineering design role

A big pile of bad date is just that. It doesn't get any better if it increases in size.
You can use statistical data to identify problems, but you need to do the engineering to solve the problem.

The tragic titanic oceangate incident is a good example. Combining bolted titanium end caps to a carbon shell would make any engineer with a basic understanding of material fatigue react, like the experts did.
The first time saw the design i said they are crazy.

You need to start with the fundamentals. Blindly combining allot of date that contradicts some of these basic principles is not leading anywhere.

 

[Bhugo]

We will see that research right around the time we see the formula for Coke.

Last year, I reached out to P&G, Gillette and AOS to see if the had any marketing materials for their many razor blades, so we might see how they position their products. I had hoped to proved this to @helicopter to incorporate into his research.

P&G was like. "Thank you so much for your interest in our products. We sincerely hope you take this opportunity to go pound sand".

Ha, predictable answer. Pretty good.

 

[SaguaroSmart]

A big pile of bad date is just that. It doesn't get any better if it increases in size.
You can use statistical data to identify problems, but you need to do the engineering to solve the problem.

The tragic titanic oceangate incident is a good example. Combining bolted titanium end caps to a carbon shell would make any engineer with a basic understanding of material fatigue react, like the experts did.
The first time saw the design i said they are crazy.

You need to start with the fundamentals. Blindly combining allot of date that contradicts some of these basic principles is not leading anywhere.

Of course. You're not going to perform engineering design solely with probability and statistics and certainly not with a statistician (after all everyone has their own expertise). What I meant is, in many cases, we do have useful data, we are just making very poor use of it. Often, failures lie in unidentified flaws and risk. The case of Challenger highlights that, I'm sure engineers did start from first principles, but the useful data was there and it was not used properly.

I can see you're skeptical about the use of data. It is true that bad data exist and can be misused. However, there are plenty of very useful data out in the world and when we leverage it, it can not only create significant value but even outperform humans at many things.

A great example of that is AlphaFold. Protein folding geometry is so complex that solving a single protein's 3D structure used to take years or even decades for some using experimental methods like X-ray crystallography, NMR spectroscopy or cryo-electron microscopy. It used to be a scientist lifetime endeavor, being worthy of award like nobel prizes (work on the ribosome or ion channels).

2020 comes around and AlphaFold gets released by Deepmind, and the model is so good at predicting 3D protein structures with accuracy that it has successfully predicted structures for hundreds of millions of proteins (essentially the entire known protein universe) and completely transcended a scientific field. This got Demis Hassabis and John Jumper the nobel prize in Chemistry.

So bad data is just that bad data. Good data paired with mathematics and statistics lead to entire scientific revolutions.

Anyway, I will quit writing about Math and Statistics as it's off topic at this point. For those who are still trying to use data to understand razors, feel free to contact me if you have questions.

 

[Bhugo]

The Gillette new improved razors had a wider “clamp width” when introduced. They had to go with a more narrow one after a lawsuit was filed because the wider one was breaking blades I believe. I could definitely be off on that that statement. It is from memory or imagination at this point…

Anyhow, I have 2 of the new improved razors with different clamp widths. I cannot tell a difference, but some say they can. Of course, some like the narrow end some like the wider…..

Interesting topic.