YDC convenes after Aki 2025

[Kintamayama 47,192]

The Yokozuna Deliberation Council convened in Tokyo today. Chairman Oshima Tadamori (former Speaker of the House of Representatives) praised Yokozuna Oonosato, who won his fifth yusho since his promotion to Makuuchi, saying, "Just one word: Bravo. He has the awareness and responsibility of a Yokozuna. I hope he continues to do so." Houshouryuu, who lost the first championship decider between two Yokozuna in 16 years, won his match against Wakatakakage on the 14th day by henka. Regarding this point, he said, "I understand that there are various opinions," but added, "Winning is an important goal. In addition, when the yusho is right in front of you, this is the result of a comprehensive judgment." The two Yokozuna livened up Aki basho. Chairman Oshima expressed his hopes, saying, "The 'Taihou era' has finally arrived."

Edited September 29 by Kintamayama

[RabidJohn 2,056]

On 29/09/2025 at 10:47, Kintamayama said:

... "Winning is an important goal. ..."

The definitive word on henka.

I appreciate that he didn't use the damning "unfortunate" anywhere.

Edited October 1 by RabidJohn

[Jakusotsu 6,389]

2 hours ago, Kintamayama said:

...praised Yokozuna Oonosato, who won his fifth yusho since his promotion...

Um, no.

[rokudenashi 343]

Taihō era

[RabidJohn 2,056]

1 hour ago, Jakusotsu said:

Um, no.

Since his promotion to makunouchi, yes, but you'd think they'd make more of 1st as a yokozuna.

[Kintamayama 47,192]

2 hours ago, Jakusotsu said:

Um, no.

I meant to Makuuchi. 

[Kintamayama 47,192]

2 hours ago, rokudenashi said:

Taihō era

Indeed, it was pointed out to me by Akinomaki as well.

[Gurowake 4,151]

6 hours ago, Kintamayama said:

the first championship decider between two Yokozuna in 16 years

While I admit that you're probably only translating this and aren't considering what it is that they're trying to say, it's the first *playoff* between two Yokozuna since 2009.  There have been other yusho-deciding bouts between Yokozuna on Day 15 since then, most recently in Haru 2020 (the last time we had two Yokozuna face each other at all) when Hakuho vs. Kakuryu as the last match of the tournament was a "championship decider" as both were on 12 wins going in and no one else had at least that many.  In an interesting coincidence, Takanosho also had 12 wins that tournament, winning on the last day and so not in contention at that point.

Edited September 29 by Gurowake

[Koorifuu 1,110]

Y'all are haters, Oshima did say that we're getting the Oho Era and I'll have none of your "corrections".

[Kintamayama 47,192]

5 hours ago, Gurowake said:

While I admit that you're probably only translating this and aren't considering what it is that they're trying to say, it's the first *playoff* between two Yokozuna since 2009.  

Yes, you are right. I wasn't specific enough. I meant in a playoff, but somehow failed to convey it in writing. The original says playoff, I went with championship decider, because that's what it was, and I was sure it was clear, but I got that wrong. Original was correct.

[dingo 1,477]

I'm just just waiting for Oho to make yokozuna so they can start calling it the O-ho-ho era. 

[Jakusotsu 6,389]

Thank goodness Kotoshoho will never make it...

[Bunbukuchagama 964]

6 minutes ago, Jakusotsu said:

Thank goodness Kotoshoho will never make it...

There was zero chance of that.

[Gusoyama 104]

Can someone explain to me why it is being called the Taihō era? Is it a kanji thing I don't get? 

[dingo 1,477]

Taiho -- 大豊, the first kanji of both Onosato and Hoshoryu. Together they can be read as taiho. It can also be read as Oyutaka, the name of a former sumo wrestler and ex-Arashio oyakata. https://en.wikipedia.org/wiki/Ōyutaka_Masachika

[Akinomaki 41,774]

38 minutes ago, dingo said:

Taiho -- 大豊, the first kanji of both Onosato and Hoshoryu. Together they can be read as taiho. It can also be read as Oyutaka, the name of a former sumo wrestler and ex-Arashio oyakata. https://en.wikipedia.org/wiki/Ōyutaka_Masachika

The YT idiot translation always calls Hoshoryu Toyonoryu and accordingly the era the Otoyo era.

[BigKahunaMan 18]

I doubt this entered into the YDC's calculus, but it does feel a bit appropriate to call the era "Taiho" considering we've now got two Ukrainians in makuuchi, his grandson is also in the joi, and Taiho himself was half Ukrainian.

[Akinomaki 41,774]

There are always connections: Taiho's era was the Hakuho era 柏鵬 (Kashiwado-Taiho) 

[Koorifuu 1,110]

On 30/09/2025 at 20:15, Akinomaki said:

The YT idiot translation always calls Hoshoryu Toyonoryu and accordingly the era the Otoyo era.

Shikona are machine translation's natural enemy, due to their fanciful-decorative nature with all the uncommon readings. The algorithm will always go with the most likely option, and that certainly isn't the super obscure reading oyakata love scraping for shikona. It isn't a coincidence that the vast majority of its correct guesses are for rikishi's real surnames - much larger sample pool to draw that info from.

This will only change when translation AI becomes so advanced it's able to intuitively understand how ozumo & its shikona work, and where to apply that knowledge.

Edited Friday at 16:08 by Koorifuu

[Gurowake 4,151]

1 minute ago, Koorifuu said:

This will only change when translation AI becomes so advanced it's able to intuitively understand how ozumo & its shikona work, and where to apply that knowledge.

Which also highlights why they're building huge data centers and trying to find good sources of electricity to run them, because in their current form they need to be retrained constantly from the ground up any time they want to incorporate new information into an AI system.  This just doesn't seem sustainable to me, but given that no one really knows exactly how neural nets actually encode the data, it's impossible with the current paradigm to just simply add information without redoing the entire thing, especially since that new information might change things that it already learned rather than just add to its knowledge.  It says a lot about the weaknesses of LLMs and suggests that true general artificial intelligence (I don't like the term "artificial general intelligence" because people then say AGI, which means something completely different to me as a tax preparer - Adjusted Gross Income) will likely have to come from a different paradigm.

[Kachikoshi 52]

Do Japanese speakers even know how to pronounce shikona? The TV broadcast always includes the pronunciation next to the kanji so I assume they do not.

[Kintamayama 47,192]

1 hour ago, Koorifuu said:

Shikona are machine translation's natural enemy, due to their fanciful-decorative nature with all the uncommon readings. The algorithm will always go with the most likely option, and that certainly isn't the super obscure reading oyakata love scraping for shikona. It isn't a coincidence that the vast majority of its correct guesses are for rikishi's real surnames - much larger sample pool to draw that info from.

This will only change when translation AI becomes so advanced it's able to intuitively understand how ozumo & its shikona work, and where to apply that knowledge.

Google translate, for instance, can be "taught" to translate certain words in certain ways. I wrote to Google a few times specifically pointing out the Houshouryuu thing. We shall see.

[Reonito 1,774]

6 hours ago, Gurowake said:

Which also highlights why they're building huge data centers and trying to find good sources of electricity to run them, because in their current form they need to be retrained constantly from the ground up any time they want to incorporate new information into an AI system.  This just doesn't seem sustainable to me, but given that no one really knows exactly how neural nets actually encode the data, it's impossible with the current paradigm to just simply add information without redoing the entire thing, especially since that new information might change things that it already learned rather than just add to its knowledge.  It says a lot about the weaknesses of LLMs and suggests that true general artificial intelligence (I don't like the term "artificial general intelligence" because people then say AGI, which means something completely different to me as a tax preparer - Adjusted Gross Income) will likely have to come from a different paradigm.

I completely agree on the general point about LLMs, but they can "fine-tune" trained models without retraining from scratch. It wouldn't be hard to tweak the translation models to work correctly in the sumo context, someone would just need to do it. Presumably it could even pick up from the rest of the prompt that it should use the sumo-specific terms instead of a more common translation in a more general context.

[Koorifuu 1,110]

On 03/10/2025 at 18:15, Kintamayama said:

Google translate, for instance, can be "taught" to translate certain words in certain ways. I wrote to Google a few times specifically pointing out the Houshouryuu thing. We shall see.

Hope you got more luck than me! I've written at least 5 reports across 3 years on how it acts like scabbard-fish and swordfish are interchangeable when translating between English and Portuguese - and it's still wrong most of the time. 

[Bunbukuchagama 964]

5 hours ago, Koorifuu said:

Hope you got more luck than me! I've written at least 5 reports across 3 years on how it acts like scabbard-fish and swordfish are interchangeable when translating between English and Portuguese - and it's still wrong most of the time. 

"Peixe espada" is what they commonly call scabbardfish in Portugal (it's especially popular on the islands); "espadarte" is the word for a swordfish. Apparently, Brazilians also use "peixe espada" for swordfish, which might explain the whole confusion. 

Edited Sunday at 05:11 by Bunbukuchagama