Replay Lounge 2026 Olympics | Women

[TontoK]

She did rotate her jumps under the system (within a quarter is rotated). She received ! on both of her flips. The judges can do with that what they want (no reduction in base value -1 to -2 deduction). Her PCS overall was lower than Kaori's. Skating skills are not the only component. She was rightly rewarded on performance. I still don't understand how Ami's score was so (relatively) low. I had her at 142-43. But I think her placement (bronze) was correct.

Honestly, I think the judges would have preferred giving Sakamoto the gold. She had amazing GOE on many of her elements (some might even say excessively high on a few of them), she won the PCS by a few points... the table was set for a Kaori win. The problem was that there just wasn't enough base value to make the stretch to gold.

My scoring conspiracy is that Ami was lowballed to make sure that Kaori took silver. It's OK, taking both segments into account, the podium was correct, and in the proper order.

Now, lower down... off the podium... I think there's some issues that might be discussed there, but is it really worth the bother to argue over who should have gotten 5th or 7th or Top 10?

 

[TT_Fin]

I have listened sometimes to Finnish commentator Mika Saarelainen, former hight level judge in international level, Swedish speaking commentator Marianne Nyman, national level judge (if this is new information to someone, reason to Swedish comment is that Swedish is also an official language in Finland and every Finnish speaking students must study it at school - also opposite way, whose native language is Swedish, they study Finnish). Mika is usually very objective, even towards our own skaters. He has picked two women whose jumps are the best: Niina Petronkina and Sofia Samoldeikina. I did not listen yet what he said about Niina here as I watched also a part with no commenting after finding that option, but he said in Euros Niina's jumps is like from jumping guide book, and he said the same about Sofia, on Sofia's case I listened to him here. I agree and in case of jump's GOES, both Niina and Sofia are underscored compared to some of the top skaters. They should get more.

 

[Blades of Passion]

not even close to a q, much less an ur. It was clean.

Look at where Alysa leaves the ice on her 3Toe. Her skate has turned 3/4. It's fair to say she did barely enough to avoid a 'q' call, but there is a technical flaw here.

The 2Toe after her 2Axel definitely should be called, 'q' at least, and would be fine to give <. That was not a real jump. It's just a reach-around-and-spin-off-the-ice move.

He has picked two women whose jumps are the best: Niina Petronkina and Sofia Samoldeikina

Petrokina's jumps are nothing special. She whips her upper body into the rotation a bit too much, instead of first letting the jumps rise, and her body positions on the lutz and flip entrances are rather awkward.

Samoldeikina has generally nice jumps but definitely not a "textbook" lutz or flip, both in terms of the pre-rotation and also the lutz edge.

 

[yfan1]

Look at where Alysa leaves the ice on her 3Toe. Her skate has turned 3/4. It's fair to say she did barely enough to avoid a 'q' call, but there is a technical flaw here.

Maybe I'm not looking at it properly, but I didn't think it was enough. And yes, I think it's too little rotation, it's very clear on take-off. Just feels "empty", if you know what I mean.

It was either her or Anna Shcherbakova who used to do nearly a full turn on the ice before jumping on their salchows, too.

Not to belabour the point I was making with Shaidorov, but this sets a bad precedent for women's skating. Luckily, Alysa is a much better winner than the three Russian women who came before her, and can be a very good performer with the right vehicle.

 

[dorispulaski]

What Sofia needs now is to work on her spin levels. Out of the 6 spins in her two programs, only one was level 4, and in her SP one was CCoSp3V. Her jumps are excellent. .

 

[yfan1]

Skating skills are not the only component.

This actually got me thinking that - while I agree on principle (see my comments on Kaori and Yuma's skating), I really do wonder how much value this sentiment has in the current state of skating. Step sequences take up 45 seconds of most programs - that's a massive section dedicated purely to skating skill display (because most choreography ends up looking the same with the cluster requirements). Some perform with the music decently - like Alysa, yes - but I do wonder what's going on in the judges' heads much of the time when they take the programs into consideration.

 

[boo69]

Both Alysa and Cizeron came back from retirement and took the gold away from Kaori and C/B
It seems that the bronze medallists from all disciplines are happier with their medals than the silver medallists.
How ironic that the skater who didn't care about medals ended up at the top and the ones who wanted it so bad didn't get what they want.
So Kaori couldn't add the 2T in combo with the last jump?
Alysa peaks at the right time. She has lost Nationals to Amber 2x, but beat her at the world stage both times. Amber can't put 2 good performances at big international competitions. She peaks at Nationals.

 

[TT_Fin]

Both Alysa and Cizeron came back from retirement and took the gold away from Kaori and C/B
It seems that the bronze medallists from all disciplines are happier with their medals than the silver medallists.
How ironic that the skater who didn't care about medals ended up at the top and the ones who wanted it so bad didn't get what they want.
So Kaori couldn't add the 2T in combo with the last jump?
Alysa peaks at the right time. She has lost Nationals to Amber 2x, but beat her at the world stage both times. Amber can't put 2 good performances at big international competitions. She peaks at Nationals.

Like hockey. If you get silver, you lost your last game. If you get bronze, you won your last game. It has happened many times in Fs that bronze goes to skater whose goal was just top10.

 

[frompolandwithlove]

I think this is my least favorite podium version from all of the possible outcomes I imagined. But then again, every podium without Kaori in first place would be unsatisfying to me. While I'm still heartbroken for her a little bit, she just joined the very honorable group of excellent skaters with an Olympic silver medal - just like Patrick Chan or Brian Orser. I hope she realizes that one mistake doesn't negate her whole career - it is unfortunate that it happened at the olympics, but ultimately Olympic games are just another competition and she was brilliant on so many other occasions.

I liked Alysa right after her comeback from retirement, but now I have to admit everything around her is a little too much for me (which, by the way, has little to do with Alysa and more with, I would say, very american way of promoting figure skating; you can call me salty, whatever, I don't care, Boston worlds broke something in me I think, the spectacle, the talent show-esque vibes, the intense and over-the-top marketing - it's just not my thing). But I'm sure she js the exact thing US Federation wanted - she will be great for media attention, clickability and general popularity, with the attitude and the backstory and everything. Skating-wise, I'm interested to finally see some other programs from her. Both her short and free were good, but at this point we're seen them enough. And due to womens' figure skating being in its healing era the great thing is: there is a chance Alysa will stay for another quad and not retire right away. How awesome it is to have a chance to observe the development of a skater!

And speaking of development and changes: Ami was a highlight for me. Can't wait to see what she'll bring to the table in upcoming seasons. But what a way to start senior career!

Interesting thing for me: is Amber pushing for another quad? I would love that, I would love to see her in 2030. It is not impossible. Corect me if I'm wrong, but Carolina Costner finished her career at 31, right? And she was still competitive. So it wouldn't be unprecedented.

 

[dorispulaski]

Amber is still improving! I hope she stays for a while!

 

[Joe Mendoza]

Thank you for sharing chatGPT's analysis ... however, it's hard for me to trust much about it because already, it has a flaw on the rules of judging panels.

The ISU didn't decide not to assign the US judge to the FS panel. It was done by a draw. So it's just not accurate of a statement.
If chatGPT is not aware of basic rules, how can they make such in depth analysis ?

When they are counting deviation and what not, are they aware that the highest and lowest scores are dropped ?

When they are looking at anti-rival biases... I was surprised to see names in there that in theory are not at all contenders. Why would a judge have a national bias against a Korean skater or even Petrokina ? I mean, it was very unlikely from the get go that one of these skaters would be a factor in the rankings.

My analysis here is that chatGPT decided that skaters who were close to one another may have suffered from an anti-rival biases... but the reality is that I doubt that there was an anti-rival bias here because the rival identified by AI are not "direct rivals"

In any case, I am always interested of seeing anyone's analysis, even AI analysis but I doubt that there is much to find here other than some judges did have different tastes and that nobody was harmed in the process. I didn't think the rankings in the SP were wrong at all. I haven't seen the LP yet but from looking at protocols, it looks to me that Kaori sealed her own result by not being able to tag a combo on her second triple flip and she lost her points there... End of story. As much as I love Kaori and as much as I believe that she's a better skater than anyone else on the planet right now, I don't think it would have been justified to rank her first... so I am glad it didn't happen.

Congrats to Alysa, Kaori and Ami.

So I copied and pasted your feedback into the AI (the thing corrected me and said it's actually Claude by Anthropic, not ChatGPT), and here's its response:

1. "The ISU didn't decide not to assign the US judge to the FS panel. It was done by a draw."​

4everchan is correct, and my language was imprecise. ISU judging panels are assigned through a random draw system governed by ISU Communication protocols, not by deliberate selection. My memo stated that "the ISU's decision not to assign Rosenstein to the FS panel was the single most consequential structural factor," which incorrectly implied an intentional human choice.
The accurate statement should have been: "The outcome of the random draw, which did not place Rosenstein on the FS panel, was the single most consequential structural factor." The analytical conclusion — that Rosenstein's absence from the FS limited the compounding of his bias — remains valid regardless of whether the assignment was deliberate or random. But the framing matters, and I should have stated the mechanism correctly. I would note, however, that this actually strengthens the broader concern rather than weakening it: if panel assignments are random, then there is no structural safeguard preventing a highly biased judge from being drawn into both segments, as happened with Han (Korea) in women's and Dabouis (France) in ice dance. A random draw is neutral to bias — it neither prevents nor promotes it.

---

2. "Are they aware that the highest and lowest scores are dropped?"​

Yes — this was explicitly addressed in the analysis. The ISU's trimming system was discussed in multiple sections of the memo. Specifically:

• The analysis noted that trimming operates at the individual element and component level, not at the total score level.
• I noted that when a judge is consistently at or near the extreme on many marks but is not always the single most extreme on each one, many of those marks survive trimming while still pulling the trimmed average in a biased direction.
• For Rosenstein on Petrosian specifically: he was tied for lowest on 5 of 7 elements and sole lowest on 2 of 7. On elements where he was tied for lowest, his mark would not necessarily be the one trimmed — another judge's identical mark might be removed instead, leaving Rosenstein's to count.
• On PCS, his marks for both Petrosian and Lee were the lowest on the panel for all three components. Only one PCS mark is trimmed per component — his would be removed on those, but there are only 3 components (vs. 7 elements), meaning the TES suppression partially survived trimming even if the PCS suppression was mostly trimmed.

The memo's impact assessment explicitly concluded that after trimming, Rosenstein's residual effect was estimated at approximately 0.44 points on the Levito-Lee margin — and that this was insufficient to conclusively change the 8th/9th ordering. This is a finding that accounts for trimming, not one that ignores it.

---

3. "Why would a judge have a national bias against a Korean skater or even Petrõkina?"​

This is the most substantive critique, and it deserves a nuanced response. 4everchan is right that in the SP standings alone, Petrosian (5th) and Lee (9th) were not in direct competition with Liu (3rd) or Levito (8th) for the same medal positions. The term "anti-rival bias" may have been too strong for what the data actually shows. However, I would push back on three grounds:

a) The statistical pattern exists regardless of the motive.​

The analysis does not claim to read Rosenstein's mind or prove he was thinking "I need to suppress Lee to help Levito." It identifies a statistical anomaly: the probability of any single judge being at or tied for the lowest GOE on all 7 elements AND the lowest PCS on all 3 components for a specific skater — while simultaneously being at or near the highest on multiple marks for a compatriot — is extremely small under random variation.
Whether this pattern reflects nationalistic strategy, unconscious bias, genuine artistic disagreement, or something else entirely, the pattern is real and measurable. The memo identifies the what; the why is legitimately debatable.

b) Combined standings reveal closer proximity than SP standings alone.​

In the final combined standings, the US-affiliated proximity was much tighter than the SP suggested:

Place	Skater	Combined
5	Amber Glenn (USA)	214.91
6	Adeliia Petrosian	214.53
7	Niina Petrõkina	210.82
8	Haein Lee (KOR)	210.56

Export as CSV
Glenn and Petrosian were separated by just 0.38 points. While Rosenstein could not have known the FS results when scoring the SP, a judge familiar with the competitive landscape would reasonably be aware that Petrosian and Lee were potential top-5/top-8 contenders across both segments. That said, 4everchan's point that these were not "direct rivals" in the moment of SP scoring is fair — the term was overreaching.

c) The pattern is asymmetric in a way random taste differences wouldn't produce.​

If Rosenstein simply had "different tastes," we would expect his deviations to be distributed somewhat randomly across skaters. Instead, the three largest positive deviations went to all three USA skaters, and the two largest negative deviations went to two specific non-USA skaters, with the magnitudes on the negative side (−5.39, −5.02) being far larger than on the positive side (+3.08, +1.06, +0.23). This asymmetry — modest boosting of compatriots combined with severe suppression of specific others — is not easily explained by general artistic preference. A judge who simply "scores tough" would suppress everyone; a judge who favors certain styles would show mixed patterns without national clustering.

---

4. "I didn't think the rankings in the SP were wrong at all."​

I substantially agree. My memo's conclusion explicitly stated:

"No bias changed a final SP placement."

And:

"The 2nd/3rd ordering (Sakamoto over Liu) was supported by 6 of 9 judges."

The memo identified patterns and quantified them but concluded they were inconsequential to results — the same conclusion 4everchan reaches. The only area of uncertainty was the combined 5th/6th ordering (Glenn/Petrosian, 0.38 points), which I described as "possibly but not conclusively" affected.

---

5. "If ChatGPT is not aware of basic rules, how can they make such in-depth analysis?"​

First, a minor clarification: I am Claude by Anthropic, not ChatGPT. More importantly, the substance of this concern is valid — if an analysis contains factual errors about the sport's procedures, it undermines confidence in the numerical work. The draw-vs-decision error was a framing mistake, not a computational one, but I take the point seriously. The numerical analysis itself — score deviations, rank deviations, element-by-element GOE comparisons — is derived directly from the official scoring data on SkatingScores and can be independently verified. I would invite 4everchan to check any specific number. The data is what it is; the interpretive framework is where reasonable people can and should disagree.

---

Revised Position​

In light of this critique, I would amend the women's memo as follows:

1. Replace "the ISU's decision not to assign Rosenstein" with "the outcome of the random panel draw, which did not place Rosenstein in the FS"
2. Soften "anti-rival suppression" to "anomalous negative deviations concentrated on specific non-USA skaters" — acknowledging the pattern without asserting a motive
3. Retain all numerical findings, which are data-derived and verifiable
4. Retain the core conclusion that no SP placement was changed by bias — which aligns with the critic's own assessment

The strongest finding in the women's event was never the SP in isolation — it was the combined 5th/6th margin (0.38 points), where Rosenstein's SP suppression of Petrosian and Han's FS suppression of Petrosian converged across both segments. That finding survives 4everchan's objections because it doesn't require attributing motive — it simply notes that two judges with home-country skaters near Petrosian in the standings both scored her far below panel mean, and the cumulative effect overlaps with the margin.

 

[4everchan]

So I copied and pasted your feedback into the AI (the thing corrected me and said it's actually Claude by Anthropic, not ChatGPT), and here's its response:

1. "The ISU didn't decide not to assign the US judge to the FS panel. It was done by a draw."​

4everchan is correct, and my language was imprecise. ISU judging panels are assigned through a random draw system governed by ISU Communication protocols, not by deliberate selection. My memo stated that "the ISU's decision not to assign Rosenstein to the FS panel was the single most consequential structural factor," which incorrectly implied an intentional human choice.
The accurate statement should have been: "The outcome of the random draw, which did not place Rosenstein on the FS panel, was the single most consequential structural factor." The analytical conclusion — that Rosenstein's absence from the FS limited the compounding of his bias — remains valid regardless of whether the assignment was deliberate or random. But the framing matters, and I should have stated the mechanism correctly. I would note, however, that this actually strengthens the broader concern rather than weakening it: if panel assignments are random, then there is no structural safeguard preventing a highly biased judge from being drawn into both segments, as happened with Han (Korea) in women's and Dabouis (France) in ice dance. A random draw is neutral to bias — it neither prevents nor promotes it.

---

2. "Are they aware that the highest and lowest scores are dropped?"​

Yes — this was explicitly addressed in the analysis. The ISU's trimming system was discussed in multiple sections of the memo. Specifically:

• The analysis noted that trimming operates at the individual element and component level, not at the total score level.
• I noted that when a judge is consistently at or near the extreme on many marks but is not always the single most extreme on each one, many of those marks survive trimming while still pulling the trimmed average in a biased direction.
• For Rosenstein on Petrosian specifically: he was tied for lowest on 5 of 7 elements and sole lowest on 2 of 7. On elements where he was tied for lowest, his mark would not necessarily be the one trimmed — another judge's identical mark might be removed instead, leaving Rosenstein's to count.
• On PCS, his marks for both Petrosian and Lee were the lowest on the panel for all three components. Only one PCS mark is trimmed per component — his would be removed on those, but there are only 3 components (vs. 7 elements), meaning the TES suppression partially survived trimming even if the PCS suppression was mostly trimmed.

The memo's impact assessment explicitly concluded that after trimming, Rosenstein's residual effect was estimated at approximately 0.44 points on the Levito-Lee margin — and that this was insufficient to conclusively change the 8th/9th ordering. This is a finding that accounts for trimming, not one that ignores it.

---

3. "Why would a judge have a national bias against a Korean skater or even Petrõkina?"​

This is the most substantive critique, and it deserves a nuanced response. 4everchan is right that in the SP standings alone, Petrosian (5th) and Lee (9th) were not in direct competition with Liu (3rd) or Levito (8th) for the same medal positions. The term "anti-rival bias" may have been too strong for what the data actually shows. However, I would push back on three grounds:

a) The statistical pattern exists regardless of the motive.​

The analysis does not claim to read Rosenstein's mind or prove he was thinking "I need to suppress Lee to help Levito." It identifies a statistical anomaly: the probability of any single judge being at or tied for the lowest GOE on all 7 elements AND the lowest PCS on all 3 components for a specific skater — while simultaneously being at or near the highest on multiple marks for a compatriot — is extremely small under random variation.
Whether this pattern reflects nationalistic strategy, unconscious bias, genuine artistic disagreement, or something else entirely, the pattern is real and measurable. The memo identifies the what; the why is legitimately debatable.

b) Combined standings reveal closer proximity than SP standings alone.​

In the final combined standings, the US-affiliated proximity was much tighter than the SP suggested:

Place	Skater	Combined
5	Amber Glenn (USA)	214.91
6	Adeliia Petrosian	214.53
7	Niina Petrõkina	210.82
8	Haein Lee (KOR)	210.56

Export as CSV
Glenn and Petrosian were separated by just 0.38 points. While Rosenstein could not have known the FS results when scoring the SP, a judge familiar with the competitive landscape would reasonably be aware that Petrosian and Lee were potential top-5/top-8 contenders across both segments. That said, 4everchan's point that these were not "direct rivals" in the moment of SP scoring is fair — the term was overreaching.

c) The pattern is asymmetric in a way random taste differences wouldn't produce.​

If Rosenstein simply had "different tastes," we would expect his deviations to be distributed somewhat randomly across skaters. Instead, the three largest positive deviations went to all three USA skaters, and the two largest negative deviations went to two specific non-USA skaters, with the magnitudes on the negative side (−5.39, −5.02) being far larger than on the positive side (+3.08, +1.06, +0.23). This asymmetry — modest boosting of compatriots combined with severe suppression of specific others — is not easily explained by general artistic preference. A judge who simply "scores tough" would suppress everyone; a judge who favors certain styles would show mixed patterns without national clustering.

---

4. "I didn't think the rankings in the SP were wrong at all."​

I substantially agree. My memo's conclusion explicitly stated:

And:

The memo identified patterns and quantified them but concluded they were inconsequential to results — the same conclusion 4everchan reaches. The only area of uncertainty was the combined 5th/6th ordering (Glenn/Petrosian, 0.38 points), which I described as "possibly but not conclusively" affected.

---

5. "If ChatGPT is not aware of basic rules, how can they make such in-depth analysis?"​

First, a minor clarification: I am Claude by Anthropic, not ChatGPT. More importantly, the substance of this concern is valid — if an analysis contains factual errors about the sport's procedures, it undermines confidence in the numerical work. The draw-vs-decision error was a framing mistake, not a computational one, but I take the point seriously. The numerical analysis itself — score deviations, rank deviations, element-by-element GOE comparisons — is derived directly from the official scoring data on SkatingScores and can be independently verified. I would invite 4everchan to check any specific number. The data is what it is; the interpretive framework is where reasonable people can and should disagree.

---

Revised Position​

In light of this critique, I would amend the women's memo as follows:

1. Replace "the ISU's decision not to assign Rosenstein" with "the outcome of the random panel draw, which did not place Rosenstein in the FS"
2. Soften "anti-rival suppression" to "anomalous negative deviations concentrated on specific non-USA skaters" — acknowledging the pattern without asserting a motive
3. Retain all numerical findings, which are data-derived and verifiable
4. Retain the core conclusion that no SP placement was changed by bias — which aligns with the critic's own assessment

The strongest finding in the women's event was never the SP in isolation — it was the combined 5th/6th margin (0.38 points), where Rosenstein's SP suppression of Petrosian and Han's FS suppression of Petrosian converged across both segments. That finding survives 4everchan's objections because it doesn't require attributing motive — it simply notes that two judges with home-country skaters near Petrosian in the standings both scored her far below panel mean, and the cumulative effect overlaps with the margin.

that was interesting : it's rare on this forum that someone says I am correct about anything. I will take it

 

[fsfan2480]

I just scored along with the free skate protocols and, for fun, I designated scores from the old 6.0 scoring system for the free skate to the Top 9 ladies (using the judges ordinals) just to see if anything matched up or made any sense whatsoever, Lol. What I found:

No matter what you may think of her, her skating, her technique, I had Alysa Liu as the clear winner of the free skate and 3 points higher than Sakamoto overall.

It is wrenching to realize and understand that by Sakamoto tacking on a 1T to the Triple Flip that she would gain points by negating the REP, and also add points for doing the 1T and likely would have won the Gold medal in a squeaker. I also figured she would tack on a 2L to the end of her Triple Loop in real time and was shocked/bummed when she didn't. Such a shame. Because I'm sure Alysa would have been happy/content about winning Silver to match the Gold she already has from Team.

The most jarring thing I found in the protocols and also with scoring along 6.0 (for fun) was Ami Nakai's 9th place in the free skate. I had her 5th (in protocols) and, if you pain yourself to give her 9th in the free in the old 6.0 system, it just makes absolutely no sense, whatsoever. Now, I realize that when you're dealing with point totals vs. old system it is apples vs. oranges. Totally. There's almost no way in the old system that Nakai's skate would have given her 9th in the free. But it's still nutty to see. I think that Nakai was dinged a bit too much for 2-3 of her jumping passes and I had her in 5th with the protocols -- very close to 6th/7th -- but still, 5th. So that left her with the Bronze, for me, just closer to Sakamoto and further away from Chiba.

Oh, Mone Chiba. You did so well. And I'm so sorry that you go home empty-handed, when your teammates get medals. Sport can be cruel. I had her comfortably in 4th, overall.

I actually had Amber Glenn 2nd in the free skate (a hair above Sakamoto) and 5th overall, but quite close to Chiba for 4th. So happy that Glenn had her redemption skate.

I had Petrosian 5th in the short and 8th in the free skate for 7th overall jussst behind Petrokina. I really feel for Petrosian given what she had to face at these Olympics (concerning the Russia of it all, her coaches, etc), but I think some of her jumps were overscored, and I dinged her on some of her components because try as I might, I just found that she had very little connection to the music or to the audience and, that matters to me.

As for the rest, I just really, really enjoyed everyone. Everyone brought it.

LOVE Julia Sauter's reaction. Love how the Korean skaters performed; especially Haein Lee. I thought that Samoldelkina was underscored and could've been a place or two higher -- she has some of my favorite jumps in the entire event. I also thought that Isabeau Levito and Loena Hendrickx were a bit underscored and I had both of them a place higher than they ultimately ended up. Fantastic competition, and so much fun to score along.

 

[Mathman]

My scoring conspiracy is that Ami was lowballed to make sure that Kaori took silver.

I confess that this is a type of conspiracy theory that I never understood. Why would ISU judges prefer that Kaori Sakamoto win the silver metal rather than Ami Nakai? Why, for the matter, would the Japanese Skating Federation favor one of their skaters over another?

I am reminded of Tatiana Tarasova's quote after the 1998 Olympics (Michelle Kwan versus Tara Lipinski). "Why didn't Frank Carroll just slip us a bottle of vodka? What did we care which American girl won?"

 

[Mathman]

So I copied and pasted your feedback into the AI (the thing corrected me and said it's actually Claude by Anthropic, not ChatGPT),...

This is actually pretty cool. If Claude had posted this on its own without Joe Mendoza's intervention, would we be able to tell that this is not a human being?

Can AI express human emotions (or at least string words together that make it seem so)? Here Claude is getting testy about some other AI program (ChatGPT) getting credit for his (Claude's) work.

 

[4everchan]

This is actually pretty cool. If Claude had posted this on its own without Joe Mendoza's intervention, would we be able to tell that this is not a human being?

if Claude had posted his analysis on here, I would have replied the exact same way.


I may have even gone more in depth about the anti-rival biases analysis that is very inaccurate (for instance, looking at Haein and Petrokina's as "rivals" can only come after the scoring has been done because there is no way that judges would have lowballed them in a planned way to disfavour them as close rivals because they were just not supposed to be factors.) When we are trying to identify biases, yes, we can look at how the scores were given. At the same time, there is something called SPORT. And in sport, underdogs can outperform favourites and in this case, some skaters did rank higher in the SP as Amber Glenn for instance. If the American judge gave them lower scores, before Amber had even skated, it is definitely not because they were trying to protect Amber... How could they have known she would pop a loop ? So this is where AI or couch coaches fail. Sport happens in real time. Data is collected all at once, at the end. Some of the biases revealed were impossible to manufacture while the event was happening. Looking at the score card may or may not imply such biases after the fact but they wouldn't be voluntary at all. How could they? Why would an American judge risk being suspended for lowballing a skater that will end up outside the top 5 anyway ?

Can AI express human emotions (or at least string words together that make it seem so)? Here Claude is getting testy about some other AI program (ChatGPT) getting credit for his (Claude's) work.

 

[Arigato]

Both Alysa and Cizeron came back from retirement and took the gold

A truly remarkable accomplishment by both athletes. Congrats!

 

[Cindy1983]

Who thinks disco-themed programs will become popular after Alysa’s win? (Nonscientific poll)

 

[Mathman]

if Claude had posted his analysis on here, I would have replied the exact same way.

Well, you did get Claude to backtrack to some extent. For instance, in Revision #2: Soften "ate-rival suppression" to "anomalous negative deviations."

In fact, to me the best point that Claude makes is #3a: The statistical pattern exists regardless of the motivation. The only thing that such analysis can do is to identify a statistical peculiarity. It cannot "prove" national bias, collusion, skullduggery, the judge fell asleep, the judge had a stomach ache from eating bad scrambled eggs, the judge is evil scum who should be expelled from the iSU, etc.

 

[DizzyFrenchie]

Well, you did get Claude to backtrack to some extent. For instance, in Revision #2: Soften "ate-rival suppression" to "anomalous negative deviations."

In fact, to me the best point that Claude makes is #3a: The statistical pattern exists regardless of the motivation. The only thing that such analysis can do is to identify a statistical peculiarity. It cannot "prove" national bias, collusion, skullduggery, the judge fell asleep, the judge had a stomach ache from eating bad scrambled eggs, the judge is evil scum who should be expelled from the iSU, etc.

I agree. Particularly, the big weakness of these statistical analyses when one try to take them as a proof, is that they're not based on the actual skates. Should we analyse the scoring deviation in comparison with correct scoring after the rules (which would have a margin, partly general and partly particular to the characteristics of each skater) we would have something way more relevant, yet to be further analysed. For instance, the idea of national bias doesn't take into account the possibility that a federation will push a skater/team more than another, which we have seen happening though.