Massive Moon Heh's Greatest of All Time (GOAT) Ratio...

[MassiveMoonHeh]

So like everyone I have always struggled to figure out who the greatest smiths are of all time.

 

To be fair I don't think there is actually an answer but I thought I would dig a little deeper and have a little fun with it.

 

Starting with the Toko Taikan, Hawley & Fujishiro all had their shot at ranking the smiths - how could I improve on this?

 

As inexperienced as I am I thought start at the awards and accolades, to me to be a great smith you need to be in the realm of Kokuho, Nihon Sansaku & Tenka Goken smiths. I also figure there are some Shinto smiths that come up consistently so I threw them in to (although age bias is there). 

 

I then came up with a rudimentary formula that could allow for changes in taste and appreciation of new qualities. 

 

The formula is very simple: Divide the smiths total number of higher status blades (TokuJu, JuBi, JuBu, Gyubutsu & Kokuho) by the number of Juyo blades to give you the smith's GOAT ratio.

 

Higher Status Blades / Juyo Blades = GOAT Ratio.

 

And then see where we land. 

 

This of course is updated bi-yearly to account for the two additional Juyo Shinsa and the TokuJu Shinsa that happens once every two years.

 

My thinking is that if you have more than 100% on the GOAT ratio you can be considered a GOAT smith. Some appear to be more GOAT than others with quite a few having zero Juyo blades. (For the sake of the exercise I simply divided the number by one rather than leaving them at an error.)

 

This exercise was a bit of fun, but it has helped me look at some smiths a little differently and made me realise how small the number of existing blades for some smiths were. The top smith really surprised me. And I learnt from this exercise that if you had a lot of money and decided to build the ultimate collection by collecting a Juyo or above blade from every Kokuho smith then you may have to actually buy and own some of the Kokuho blades. We can all dream!
 

So there you have it, if your favourite smith does not appear here and you want to know what their GOAT ratio score is let me know.

 

NOTE: These are my own numbers taken from the Shinsa, books and notices - I have tried to be as correct as possible but it was not an academic exercise so I suspect there are plenty of mistakes - let me know if you see any. It is just a bit of fun and should not be taken too seriously. Cheers.

 

[Zoglet]

Let's go Norimune the GOAT!

But in all seriousness, this is an interesting exercise. Would love to see how it evolves over time.

[YourBabyBjornBorg]

Interesting how neither Norimune nor Munechika had any Juyo, it's like not only are signed swords extremely rare, NBTHK also tries to avoid giving out their Komei Kiwame.

 

I guess there might be some Honami Kinzoganmei one that got bent to "Kinzoganmei Norimune To Aru, Fukuoka Ichimoji" by NBTHK orz 

 

...Wait, I found there is maybe one Signed Norimune Juyo (in the 8th Juyo)?

 

 

Yes, I think this is the one.

 

 

 

Wow, this is the first time I looked inside my Juyo 1st to 10th official Reprint, and the printing quality is poor!

 

Eh, I only paid 3000 yen for them 5 years ago, so there is not much to complain about orz

[Lewis B]

Tametsugu way above Norishige. Ouch

[MassiveMoonHeh]

Just now, Lewis B said:

Tametsugu way above Norishige. Ouch

To be fair this Kokuho Tametsugu from Ko-Aoe. There is only one blade.

[Lewis B]

Just now, MassiveMoonHeh said:

To be fair this Kokuho Tametsugu from Ko-Aoe. There is only one blade.

Doh. I blame the heatwave. My brain is discombobulated at the moment. 

[MassiveMoonHeh]

Just now, Lewis B said:

Doh. I blame the heatwave. My brain is discombobulated at the moment. 

Mate - keep cool there. I don't know how Europeans live without aircon. We could not survive down under without it.

[eternal_newbie]

Nice work, always good to see different ways of considering rankings other than the usual Fujishiro/Taikan/Hawley trinity. Reminded me of this article, which attempts to do something similar but also for schools rather than just individual smiths:

 

https://web.archive.org/web/20210123052432/https://blog.yuhindo.com/pass-factor/

[MassiveMoonHeh]

25 minutes ago, eternal_newbie said:

Nice work, always good to see different ways of considering rankings other than the usual Fujishiro/Taikan/Hawley trinity. Reminded me of this article, which attempts to do something similar but also for schools rather than just individual smiths:

 

https://web.archive.org/web/20210123052432/https://blog.yuhindo.com/pass-factor/

I have never seen this article, thanks for sharing. Always glad to stand on the shoulders of giants.

[Hoshi]

Hi Brett, 

 

This is a fascinating topic. 

 

This analysis is called "Pass factor" and was first proposed by Darcy Brockbank. You can find the original blog post here: https://web.archive.org/web/20210123052432/https://blog.yuhindo.com/pass-factor/

 

There are a couple of issues you will run into as you refine your analysis: 

 

1.  Small sample sizes: prevents you from drawing valid inference due to the statistical noise. 

2. Unequal value of the higher designation: A Kokuho ought to be worth more than a Tokubetsu Juyo. 

3. Contaminated categories: there are arguments in favor of excluding Jubi and Gyobutsu post-Nambokucho.

4. The condition confound: signed works can achieve high designation on a condition basis rather than workmanship. 

5. Some elite smiths such as Tomonari or Masatsune are actually sub-schools within Ko-Bizen, but they are treated as a single individual. 

 

To solve for (1) you may try setting the different levels of thresholds on the minimum n of blades. Test 5, 10, and 20, and see how the ranking evolves. Alternatively, a more formal statistical way is to use a Bayesian Shrinkage estimate which is in my opinion the elegant path. Essentially, you have a prior that everyone is average, and the model must 'see data' in order to change its mind. To solve for (2), you can use a frequency-weighted approach (more defensible), or assign some other weights that fit your mental model of the gap between the average TJ and the average Kokuho. For (3), you can run the analysis excluding Jubi and excluding post-nambokucho Gyobutsu and see how the rankings evolve. Regarding (4), you can rerun the analysis while controlling for the ratio of zaimei to mumei works, but you enter rarified air territory. The best you can do is just accept the state of things and hope for statistical independence between the observations. Regarding (5), one just has to live with it. 

 

You can find a draft of an article on the topic here: https://nihontowatch.com/eliteranking 

 

This is what NihontoWatch uses under the hood for rankings. 

 

Here is Awataguchi Yoshimitsu: 

 

 

In the end, whatever method you choose, once you control for the unicorns by setting a threshold or a bayesian prior, you will converge on the traditional literature such as Toko Taikan and the list will begin to look very familiar. Personally, what I find most interesting is not where these analysis converge, but where they disagree: there are historically underrated smiths by Fujishiro and Dr. Tokuno that will "pop out" and this is what one should look for as it is the most informative. 

 

Correlation with Toko Taikan:

 

Period-adjusted correlation with Fujishiro: 

 

 

Hoshi 

 

[John C]

2 hours ago, Hoshi said:

There are a couple of issues you will run into as you refine your analysis: 

Hoshi:

I would also add as a control the particular judges for any given shinsa, as their evaluation is subjective, this could influence the number of K or TJ issued. It is possible one set of judges are more forgiving than another set.

 

 

John C.

[Rivkin]

16 hours ago, MassiveMoonHeh said:

To be fair this Kokuho Tametsugu from Ko-Aoe. There is only one blade.

 

The funny thing is that while I am a very minor collector, I do own two ko Aoe Tametsugu, though one still needs to be papered. Signed.

They are not the most beautiful swords one can imagine. Coarse masame to the point I took one for Naminohira. A friend whose kantei I respect a lot, when looking at it spent a few seconds before putting it back saying "some Yamato".

Much of it is condition issues, but generally speaking ko Aoe Tametsugu tends to be a historically important blade more so than artistic masterpiece.

 

That's one negative of rankings - it piles everything without trying to understand where its the signature that makes it important, or history, or aesthetics or maybe even something else.

Another aspect is that rankings tend to reflect dealer's culture.

I've seen it taking over every collecting community in the past 30 years. It values certainty, confirmation, branding. For a true collector the process is based on partnership with objects, an honest relationship with the tradition. Uncovering new, unrecorded swordsmith of significant skill is an honor. For a dealer, its mostly nuisance. "The work is good, but no one will spend money on this name". They much rather take out the signature and get another Shizu Kaneuji... and a flood of mumei.

 

There is such thing as collector's duty.

To honor and appreciate.

A duty to a "minor maker" to recognize his masterpiece, even if its one and only. You are then holding in hands what this person's entire life was about, no less.

Trader class should behave as traders, collectors as collectors.

Edited June 25 by Rivkin

[MassiveMoonHeh]

58 minutes ago, Rivkin said:

The funny thing is that while I am a very minor collector, I do own two ko Aoe Tametsugu, though one still needs to be papered. Signed.

Wow bud you literally own a major part of this Kokuho Smith's life work! That is an incredible honour and responsibility. What an opportunity. Did you actively seek this smith or happenstance into this position? Curious to here the stories behind both blades.

[Rivkin]

Just now, MassiveMoonHeh said:

Wow bud you literally own a major part of this Kokuho Smith's life work! That is an incredible honour and responsibility. What an opportunity. Did you actively seek this smith or happenstance into this position? Curious to here the stories behind both blades.

 

Except there are at least a few in Japan at TH level. One came from gunto mounts, another was a chance buy at a show.

[Rivkin]

Another great example why traders love rankings and why collectors should have a different take on this practice.

Tomonari.

One of the greatest names. Every dealer selling Tomonari will cite how he was the founder, and how many works are rated this and that.

There is one problem. The name was used till mid Kamakura for certain. There were people adopting it even later, who might or might not be of the lineage.

If one believes genealogies the really first generation worked no earlier than 1050 no later than probably 1140. The earliest papered Tomonari could be late Heian. A lot are Kamakura. Without exception - sold and advertised as the first generation, Heian work. Even if its not said directly "this is first generation", you read description on dealer's site and its all about the first Tomonari.

One more issue.

Goban Kaji did not invite any of those "generations" of Tomonari. It was not very keen on "ko Bizen" branch altogether, instead focusing on Nobufusa's Fukuoka lineage. Much of ko-Bizen traded today is not something that preceded other Bizen schools, but rather was produced in parallel using more antiquated style. At the time when such style was not really valued anymore. And if you look at the work - there are very impressive signed Tomonari and those that are not. 

Frankly we don't even know how many of those Tomonari are simply early Kamakura period fakes.

 

The other ko Bizen founder. Masatsune. Slightly less time-elongated story, but a variety of signatures and work. Those with katana mei go to ko Aoe, those with tachi mei to ko Bizen. Except for exceptions. And without exceptions, admitting there are ko Aoe blades that look very Bizen and Bizen blades that tend towards Aoe jigane. Same shop, two related shops? How long did the production last in either.

Again, replacing all of that with a number measuring how great Masatsune was - achieves nothing. No viable research purpose, no advancement of understanding what having the signature Masatsune means. The only thing it does is allowing to convey importance of this attribution to people who have exceptionally meager knowledge of the field. But might consider buying.

[Tsuku]

A couple of humble observations.

 

First, I think there are some taxonomic problems with elite ratios.

 

One is that the various classifications measure different things. For example, a blade might be jūyō bunkazai on the strength of its historical importance, which has nothing to do with whether it is a good example of that smith's work. How should that blade be weighted against a "mere" tokubetsu jūyō that was promoted almost entirely on the strengths of its artistic representations and health?

 

Another issue is that the Bunka-cho and NBTHK run parallel schemes. I have a tokubetsu jūyō blade that was part of a pair; the other one is a jūyō bunkazai in the TNM. What makes one of those "better" than the other one? If the daimyō family that had owned them had swapped which one was submitted to which track, we would still end up with one jūyō bunkazai and one tokubetsu jūyō. So why should we credit one smith as "more elite" than the other?

 

Second, as humans and collectors, we tend to want to reduce everything to positions on a list. So-and-so is #4, thus-and-so is #5, thus so-and-so is better. Part of the brilliance of both the Fujishiro and Tōkō Taikan rating schemes is to instead operate on equivalence classes. Fujishiro's are coarser, the Tōkō Taikan's are finer, but in either case we can say that so-and-so and thus-and-so are both saijō-saku or 1800 man — and so who are we to say which is "better?" I think this is a more structurally appropriate view than trying to argue about whether Tomonari or Hisakuni is better.

 

...well, unless we are drinking after a long day of kantei. Then we can argue!

[jdawg221]

3 hours ago, Tsuku said:

A couple of humble observations.

 

First, I think there are some taxonomic problems with elite ratios.

 

One is that the various classifications measure different things. For example, a blade might be jūyō bunkazai on the strength of its historical importance, which has nothing to do with whether it is a good example of that smith's work. How should that blade be weighted against a "mere" tokubetsu jūyō that was promoted almost entirely on the strengths of its artistic representations and health?

 

Another issue is that the Bunka-cho and NBTHK run parallel schemes. I have a tokubetsu jūyō blade that was part of a pair; the other one is a jūyō bunkazai in the TNM. What makes one of those "better" than the other one? If the daimyō family that had owned them had swapped which one was submitted to which track, we would still end up with one jūyō bunkazai and one tokubetsu jūyō. So why should we credit one smith as "more elite" than the other?

 

Second, as humans and collectors, we tend to want to reduce everything to positions on a list. So-and-so is #4, thus-and-so is #5, thus so-and-so is better. Part of the brilliance of both the Fujishiro and Tōkō Taikan rating schemes is to instead operate on equivalence classes. Fujishiro's are coarser, the Tōkō Taikan's are finer, but in either case we can say that so-and-so and thus-and-so are both saijō-saku or 1800 man — and so who are we to say which is "better?" I think this is a more structurally appropriate view than trying to argue about whether Tomonari or Hisakuni is better.

 

...well, unless we are drinking after a long day of kantei. Then we can argue!

Exactly what I was thinking when looking at this list. For one the nbthk rankings are based off a few things, some of which are just plainly subjective. Such as who owned it, hype surrounding the smith, or, and this one might be controversial; whether it is shortened or not. Suriage has little affect on the artistic or functional qualities of the blade(unless it is done very poorly of course), yet it is a big mark against how a blade does with papers once you get past a certain point in history. Specifically with the hype surrounding the smiths: I have handled a juyo kotetsu exactly once, so take this with a grain of salt, but I was not that impressed. I find other shinto smiths to be of equal if not greater quality, specifically Tadayoshi 3, shodai tadayoshi, hell even yamashiro kunikiyo. But because kotetsu blades are rare and there is a level of hype surrounding the smith, the nbthk shows favoritism towards ranking them. In fact you can find th mutsu no kami blades that are just as cleanly forged if not more cleanly forged than juyo kotetsu examples, yet theyre th because that is what is expected of the smith, and the smith's work is not quite as rare as kotetsu's is. 

[jdawg221]

Also another thing with this ranking is that its purely from an artistic perspective. What the people that used the swords considered the best is very different from what modern collectors would consider the best. A good example of this is how sue bizen swords were seen during the edo period. Frequently they were considered the swords to use when you wanted to "get the job done", and famously after the igagoe revenge, during which araki mataemon's rai kinmichi broke during the fight, Tonami told him that he should have used a sue bizen sword instead. Hizen swords were also highly praised, and while mainline hizen still rightfully gets lots of respect and praise in the community, you have stories from ww2 in china where cheap swords from hizen were preferable over swords from higher rated smiths like inoue shinkai. 

[Hoshi]

 

Hi, 

 

Quote

One is that the various classifications measure different things. For example, a blade might be jūyō bunkazai on the strength of its historical importance, which has nothing to do with whether it is a good example of that smith's work. How should that blade be weighted against a "mere" tokubetsu jūyō that was promoted almost entirely on the strengths of its artistic representations and health?

 

Regarding the provenance confound you mention, I think it is not so pronounced that on average, it prevents inference. When comparing Jubun to TJ, there is an average difference on many dimensions, especially condition, signature status, motohaba. A fair minority of TJ (10-20%) I would estimate as "Jubun class". This is not to say provenance plays no role, only that it is not strong enough to invalidate the approach on its basis alone. That said, it is a good addition to the limitation of the Pass Factor analysis. 

 

6. Provenance confound: provenance can steer a blade to higher designation. 

 

It does raise the broader point of what it means to be a 'the most historically important' smiths. Is history of ownership not an intrisinsic part of this measure? We certainly cannot make the claim that we are looking purely at artistic merit in these rankings. This is part of the reason I systematically extract provenance from the data (which has another big limitation of its own: it's incredibly incomplete). I would like to understand how to both control for it, and how to make it legible to the market. Work in progress. 

 

Quote

Second, as humans and collectors, we tend to want to reduce everything to positions on a list. So-and-so is #4, thus-and-so is #5, thus so-and-so is better. Part of the brilliance of both the Fujishiro and Tōkō Taikan rating schemes is to instead operate on equivalence classes. Fujishiro's are coarser, the Tōkō Taikan's are finer, but in either case we can say that so-and-so and thus-and-so are both saijō-saku or 1800 man — and so who are we to say which is "better?" I think this is a more structurally appropriate view than trying to argue about whether Tomonari or Hisakuni is better.

 

I think this is spot on. Excess granularity is meaningless and error-inducing because of all the above-mentioned limitations. Noisy continuous measures should be normalized into distribution quintiles, for instance, and the 'score' presented with a large warning as to avoid misunderstandings. More broadly: I tend to think that works across certain schools cannot be compared: Bizen vs Soshu vs Yamashiro is apples vs orange vs milk. Who was better? Masamune, Mitsutuada, or Yoshimitsu? There is really no answer to this question. Comparison is meaningful insofar as the ideal that is driving the work is comparable. Another analysis that I find personally promising is lexical-based. Rather than relying on designations, analyze positive vs negative descriptors along the saeru/uroi axis, which is the central qualifier of artistry in the field.

 

In the end, we will probably land on combination of equivalence classes over elite factor, provenance factor, with qualitative honors, and lexical analysis along the saeru/uroi axis. 

 

Music for the future. 

 

Hoshi

 

 

[Jussi Ekholm]

This is very interesting subject. As a disclaimer I must note that appreciating the fine details in workmanship is not my thing. 

 

I do agree with GP's view that the rankings are depending on many factors. One interesting thing can also be found in another thread by @MassiveMoonHeh where he pointed us that there has been quite long time since any item achieved Jūyō Bunkazai status, however NBTHK keeps awarding new TJ every 2 years. Also if the item is Jūyō Bunkazai it needs to follow certain rule set meant for items of this designation, while Tokubetsu Jūyō have no such requirements. This is just my personal opinion after visiting several places in Japan, and I feel that some places and families got higher rankings perhaps more easily than others. Of course that is related to the provenance of the item. Often the Bunkazai items are preserved in current condition and many of them are not in pristine polish, if you compare them to Tokubetsu Jūyō which are often in pristine polish. Also as described earlier for TJ items tend to be of very finely made, as the artistic quality of these blades is often very high. However for me personally just artistic quality alone is not what I want to study. I love the rustic items at the shrines that are not that commonly known, things surviving in nearly original shape and size, regardless of their quality.

 

Now after I got back from Japan this morning I typed in all the swords I saw this summer to my data and being curious I wanted to see numbers and it seems so far I have seen 26 Kokuhō and 138 Jūyō Bunkazai swords. In my personal opinion there is a huge range among these swords. Just because something is a Jūyō Bunkazai as a designation the sword itself might not be comparable quality in my eyes to for example similarish Tokubetsu Jūyō sword. However if you would slap in the fact that it was dedicated to the shrine by Tokugawa XX in 17XX in koshirae, and it has been preserved ever since, now wouldn't that make it quite important item for example compared to perhaps slightly better quality item by same smith but with no known background? Also there are absolutely amazing items in Japan completely outside of these ranking systems.