Lederberg 3: Specific Question Discussion

[Sima Guang Hater]

Feel free to discuss specific questions here. Any criticism, no matter how small, is appreciated! A copy of the version of the set played on 4.5.2019 is available in the General Discussion thread.

[khannate]

One correction: in the tossup on the Laplace transform in packet 8, there is an error in the clue about moment generating functions. The moment generating function evaluated at t=x is the two-sided Laplace transform of the PDF of evaluated at t=-x.

Also, this may be because they were never covered in any class I took and I mostly learned about them myself, but I found the tossup on random forests (which I thought was a terrific answer-line!) absolutely brutal. Except for the Microsoft Kinect clue, every single clue in power basically requires having read papers about random forests (as opposed to having seen them in a class). The Tin Kam Ho clue comes from a paper that has 1944 citations, and half of it is outside of power. Also, this is kind of nitpicking, but it says that "this method measures the error of each model using inputs that a given model has not been trained on," which this is also true of k-fold cross-validation (that models are evaluated on inputs they are not trained on). I've seen the term "out-of-bag error rate" used in this setting, and it might help make that clue more buzzable.

[otsasonr]

On the topic of the Laplace transform tossup, the clue about the transform of a step function should be clarified to indicate that this is describing general step functions, since in engineering contexts, unqualified use of “step function” usually just means the Heaviside unit step function, the transform of which is just 1/s. The clue is further confusing because the constants a_n of the corresponding general Dirichlet series would actually be functions of s, which would be odd, since they are supposed to be, well, constant. Can I get a source for this?

[setht]

I already mentioned this to James at the ICT site, but in case he lost track of my comments in the whirl of Lederberg + ICT: I think the tossup on accretion (round 1, tossup 3) needs some revision. In particular, the second, third, and fourth sentences say that turbulent viscosity (generic or specifically due to the magnetorotational instability) "impedes" and "disrupts" accretion. I think that's the opposite of the typical view, which is more like "(turbulent) viscosity provides a mechanism for transferring angular momentum outward, thereby allowing accretion to proceed." (So it would be better to say something like turbulent viscosity "facilitates" accretion.)

Also, giving the alpha disk viscosity as "full-width at half max times sound speed times alpha" seems a bit strange to me. The standard formulation is alpha times sound speed times scale height.

[The Sawing-Off of Manhattan Island]

Unless I'm misparsing the clue in the _potentiometer_ tossup, I think it would be helpful to rule out encoders before the clue about their being used for positional feedback for servo mechanisms - I'm fairly certain its possible to use optical encoders to keep track of servo position as well. I thought _perovskite_s and _dimension_s played kind of transparently (super trendy modern crystalline structure and quantity that makes your machine learning algorithms perform worse respectively). I really liked the tossups on _edge detection_ (and really most of the other machine learning topics that came up) and _Rydberg_ (rydberg polarons are super cool :O). Also, I thought _mangrove forest_s was a really cool idea, but it felt like it switched to using clues that rewarded geography knowledge pretty early.

[naan/steak-holding toll]

Also, I thought _mangrove forest_s was a really cool idea, but it felt like it switched to using clues that rewarded geography knowledge pretty early.

This was noted by Ike Jose when he reviewed the tournament. I agreed with his criticism, but unfortunately Cody Voight had written a tossup on mangrove forests for VCUO 2015 that had all of the more "pure science" mangrove forest clues I had found. I was going to scrap the question, but a bunch of my fellow writers said they liked it, so I decided it would stay. Also, I think in a set with a zillion pure science questions, it's fine to have some clues that will reward you for some alternate modes of engagement that still talk about things scientists care about - I think protection from tsunamis is something you'd very much care about as an environmental scientist, for instance, even if it's something of a "geography clue." It kind of sucks that this happened in a set with only one pure earth science tossup per round, and ideally we'd have found maybe one or two more pure science clues. However, I still really don't like these sorts of purist and provincial attitudes - as long as your questions are rewarding things that scientists care about, and not every single question is as "extracurricular" as this mangrove tossup (or the giraffe tossup from SGI), then I don't really see the problem.

[Wartortullian]

Unless I'm misparsing the clue in the _potentiometer_ tossup, I think it would be helpful to rule out encoders before the clue about their being used for positional feedback for servo mechanisms - I'm fairly certain its possible to use optical encoders to keep track of servo position as well.

Agreed. The only thing that stopped me from buzzing with encoders was realizing that the clue would have been way too early.

[Judson Laipply]

I already mentioned this to James at the ICT site, but in case he lost track of my comments in the whirl of Lederberg + ICT: I think the tossup on accretion (round 1, tossup 3) needs some revision. In particular, the second, third, and fourth sentences say that turbulent viscosity (generic or specifically due to the magnetorotational instability) "impedes" and "disrupts" accretion. I think that's the opposite of the typical view, which is more like "(turbulent) viscosity provides a mechanism for transferring angular momentum outward, thereby allowing accretion to proceed." (So it would be better to say something like turbulent viscosity "facilitates" accretion.)

Also, giving the alpha disk viscosity as "full-width at half max times sound speed times alpha" seems a bit strange to me. The standard formulation is alpha times sound speed times scale height.

Agreed and I’m sorry about the mistake.

The alpha disk velocity is written like that for pyramidality sake but I can see the argument for writing it in the conventional manner

[Carlos Be]

On the topic of the Laplace transform tossup, the clue about the transform of a step function should be clarified to indicate that this is describing general step functions, since in engineering contexts, unqualified use of “step function” usually just means the Heaviside unit step function, the transform of which is just 1/s. The clue is further confusing because the constants a_n of the corresponding general Dirichlet series would actually be functions of s, which would be odd, since they are supposed to be, well, constant. Can I get a source for this?

Applying this operation to a step function yields a general Dirichlet (“deer-uh-CLAY”) series.

Apologies, you are correct. The clue is referring to the special case of the Laplace transform used to convert the general statement of the Wiener-Ikehara theorem to the statement that directly applicable to number theoretical Dirichlet series. In that context, the extra factor of "1/s" is irrelevant since you just move it to the other side, so the source I used didn't mention it. Of course, in any other context that factor is quite relevant and the clue absolutely should have mentioned it in some form.

[Fado Alexandrino]

The AGB star tossup seemed like a great idea but poorly executed. If you pull any player who said in their LetterOne interview that science was their weakest subject, they will surely still know that at the end of the sun’s lifetime, it will become a red giant and swallow the Earth. Everything in the tossup is technically correct, but it felt unsympathetic to the player.

I PMed other specific content to Eric already.

[Jet Fuel Can't Melt Steel Dreams]

The AGB star tossup seemed like a great idea but poorly executed. If you pull any player who said in their LetterOne interview that science was their weakest subject, they will surely still know that at the end of the sun’s lifetime, it will become a red giant and swallow the Earth. Everything in the tossup is technically correct, but it felt unsympathetic to the player.

I PMed other specific content to Eric already.

The answer of "red giant" is wrong. First off, the term is ambiguous. It could be interpreted as either:
1. red stars that are "giants" i.e. stars that have a luminosity classification of III in the Yerkes classification. This would be wrong because "red giant" isn't a phase of stellar evolution
2. the RGB/red giant branch, which is a phase of stellar evolution. It, however, is an incorrect one.


Nowhere does the tossup mention the end of a star's (main sequence) lifetime. In any case, if the tossup was asking what phase of stellar evolution the sun enters after exiting the main sequence, the correct answer is the subgiant branch/SGB.

I also dispute the validity of asking hypothetical "player(s) who said in their LetterOne interview that science was their weakest subject" in judging open level science questions.

For reference, here is the text of the tossup:

The presence of high-atmospheric clouds in a type of variable star that are in this evolutionary phase is responsible for the so-called "washed-out" appearance of their spectral lines. The fact that evolutionary population synthesis overpredicts the amount of stars in this phase is called the “boosting effect.” The efficiency of the third dredge up which occurs in this phase is enhanced by convective overshoot. The evolution of stars in this phase is divided into an early phase and a phase during which these stars undergo periodic thermal pulsations. (*) Mira variables are in this phase of stellar evolution. After evolving beyond the “red clump” or “blue loop,” stars exit the horizontal branch and enter this stage. Stars during this evolutionary phase are the primary site of the s-process. After evolving beyond this stage, stars develop a superwind and eject a planetary nebula. For 10 points, identify this phase of stellar evolution named after the fact that the path traced out on a H·R diagram during it is almost aligned with the red giant branch.

[Bloodwych]

Oh right, for my neg I buzzed on the MGMT methylation clue and negged with glial cells (after being prompted and insisting I was right) because it's not entirely established that GBM comes exclusively from astrocyte precursors, just GFAP+ cells (which apparently includes neural stem cells as well). OOF

[Sima Guang Hater]

Oh right, for my neg I buzzed on the MGMT methylation clue and negged with glial cells (after being prompted and insisting I was right) because it's not entirely established that GBM comes exclusively from astrocyte precursors, just GFAP+ cells (which apparently includes neural stem cells as well). OOF

The sentence said "secondary tumor arising from these cells"; secondary GBMs arise from pilocytic or anaplastic astrocytomas. You're correct that the cell of origin for primary GBMs is unknown.

[Amizda Calyx]

I really liked the bio here, even if I had a million stupid negs. Mechanosensation was particularly rewarding.
Couple things I've already mentioned to editors but might as well bring up here:
• Hypusine is found in eIF5A, which is now considered to be an elongation factor, not an initiation factor.
• Luminal breast cancer subtypes comprise half a page in Campbell (page 387, 10th ed.) and appear to be very common terms for ER(+)/PR(+) cancers among lay literature and publications describing molecular classifications. They probably shouldn't be the first words of the tossup.