2018 Princeton University Invitational

[misaki]

It's great to hear that the invitational went well, and definitely a shoutout to the organizing team this year!

I thought I'd provide a bit of my perspective, as one of the event supervisors and organizers. As you might guess from my username, I was the 'supervisor' for Fermi Questions and Dynamic Planet; I put supervisor in quotes because I was away most of the time due to conflicts (thanks to the rest of the team for making up for my absence). It seems I am quite notorious for my tests, which I tend take 'artistic freedom' in creating—perhaps a bit too much.

Fermi Questions
This test was rather long, with 60 questions and 500 points in total—the first half would probably have been a fine test on its own. When writing, I estimated my score to be around 50-60% (with a partner), and anything above 75% to be nigh impossible; in contrast I think we scored over 90% on the Nationals test back in 2013. Other than a few unreasonable questions, it was pretty standard, so the event was fine. There were a few references to video games (BotW and DotA 2), so I guess you can use this as an excuse to play video games? Scheduling-wise, we were following the Nationals schedule. We were pretty confused as to why there was a need to put Fermi during impound, but for scheduling consistency we followed suit. Also, Fermi may be useful for job interviews later on (if you go into quant trading or something similar).

Dynamic Planet
First, I'd like to apologize for not clarifying some the multiple choice penalty. I should have made a verbal announcement, though I was also not physically present so perhaps I should have done away with it entirely. It was very difficult to get points on the section, and given the 'high' value of points there I wanted to minimize guessing. The entire test, aside from the warm-up section, was difficult—I actually added the warm-up section last because I realized the rest of the test was more suited for a seasoned geologist than a high school student. Some questions were downright unreasonable (e.g. obscure isotope ratios), though most I think was doable given enough depth in the subject. I was a bit surprised at some questions (each of the random plates in the maps section was answered correctly by at least one team) and a bit disappointed at others (e.g. calculating distance to earthquake, calculating strike and dip). For the tiebreaker question, due note that I never indicated that said anime was good. I tried watching it for the memes, but didn't get very far. The intended answer was that both are called 'Eromanga,' which I thought was funny. Forgive me for the minor grading errors (usually 0.5 points)—made quite a few of them since we were pressed on time.

Also I'm glad that Astro, Remote, and Rocks went well - I didn't write them but was the 'editor' (seems like I do know what a reasonable test looks like).

[Unome]

Fermi Questions
This test was rather long, with 60 questions and 500 points in total—the first half would probably have been a fine test on its own. When writing, I estimated my score to be around 50-60% (with a partner), and anything above 75% to be nigh impossible; in contrast I think we scored over 90% on the Nationals test back in 2013. Other than a few unreasonable questions, it was pretty standard, so the event was fine. There were a few references to video games (BotW and DotA 2), so I guess you can use this as an excuse to play video games? Scheduling-wise, we were following the Nationals schedule. We were pretty confused as to why there was a need to put Fermi during impound, but for scheduling consistency we followed suit. Also, Fermi may be useful for job interviews later on (if you go into quant trading or something similar).

60 questions -> 500 points?

Also very surprised to hear of a 90% score on Fermi - how easy was the test? I would imagine even on a fairly easy test, the top scores wouldn't reach over 80% due to the near-impossibility of consistently keeping a high 5 to 3 ratio.

[windu34]

I liked what Bill did for Dynamic and Fermi:

Remote Sensing
I was very happy with the score distribution in the end. Out of 179 available points (although there were 5 less available due to one of the images not printing), 1st place scored 120 (roughly 66%) and last was around 30 (<20%) and the scores in between were nicely distributed throughout that range, with more separation in raw scores within the top 6-10 places. As for time, I thought it was relatively long, but I had a feeling it was possible. I warned teams to move quickly and not waste time struggling with questions they were unsure of. The surveys revealed that 1st-3rd found the length to be perfect, while nearly all the other teams responded "too long". I adhered a little more strictly to the rules than I would have liked looking back on it - it would have been much more interesting (and fun) to have integrated all of the topics into singular questions that contained multiple parts which would gradually increase in difficulty between the 1st and last part of each question. The math questions I chose for the first three sections were all relatively obscure questions that I had either never seen tested or rarely tested. For that reason, I was pretty generous overall with partial credit due to difficulty. The last section (Energy Balance) was pretty easy and straightforward IMO and should have been easy points for teams that had a solid understanding of energy balance (highest score was 39/40 in this section). Of course I knew that the vast majority of teams don't understand energy balance at all so I tried to make the questions "guide" the teams. I placed a lot of emphasis on trying to help teams learn from this test in both the energy balance section as well as in the other sections because I feel the learning curve for this event is quite steep and I hope teams have a much better idea of how to proceed with preparing for this event after getting their tests back and looking at the key.

[pikachu4919]

Hmmm...I like this idea too. I hope neither of them minds that I'm stealing it.

Forensics
I supervised this event at MIT as well as Princeton, and both of my distributions were pretty much linear except for sharp drops at the high end for MIT and a basically almost linear distribution for Princeton, which I found interesting and was pretty happy about for the most part (because it means I have less ties to break! but in all seriousness...)

For Princeton, although it will be released online sometime soon, out of 587 raw points, the max was 263 (~44.8 percent) and the min (excluding NS's) was -7. The separations among the top 6 for Princeton were closer (~0-20 pts) than for MIT (30 pts between 1 and 2, 56 pts between 2 and 3, 43 pts between 3 and 4, and from there on out the scores were a lot closer together), but I guess I liked how that turned out for both of them anyways since the questions did a really good job of generating many different unique scores and separating the teams out much more.

The identification part of Forensics is the most important part, and my tests try to reflect that in terms of how the points are weighted (which is why my raw score amounts are always so high). At the same time tho, the biggest thing I like doing with my tests is trying to push the boundaries of my competitors' understanding of the objects in the SciOly Forensics realm (for example, tiebreaker 2 about calcium sulfate).

Of course, it often gets hard to come up with unique questions since the Forensics rules barely evolve from year to year (I mean, there are only so many ways that you can ask "What powder is this?"), but many Forensics tests out there often draw from memorization more than from critical thinking, for example, memorizing that cornstarch turns black with iodine or that sodium bicarb fizzes with HCl. While it is good to have those easy freebie questions that draw upon memory, again, it can get to be repetitive if only those questions appear. I try really hard to try to get my competitors to think beyond that since they truly won't deeply understand all the chemistry behind everything if they only memorize what happens (like my hint said, "Your preparation for today probably mostly involved learning about what happens. Today, I urge you to try to think more about why that “what” happens.”) This was heavily applied to the free response, and a fair amount of those questions were originally going to be worth bonus points instead of normally counting towards the score since I thought they were particularly challenging ones that would really stretch your minds, but both the Chemistry Events Director (no namedropping) and I decided that albeit challenging, they were still great and relevant problems that should be attempted, so thus they were upgraded out of bonus status.

Here's a lineplot of the scores, sorted from highest to lowest, not sorted by team number:



I didn't put a survey on my test because Forensics tests are already stressful as is, and maybe the survey results wouldn't be as well thought out as they could be. That being said, I created a Google Form survey for that purpose, so feel free to fill it out if you took Forensics at this invitational.

[kenniky]

Fermi Questions
in contrast I think we scored over 90% on the Nationals test back in 2013

Also very surprised to hear of a 90% score on Fermi - how easy was the test? I would imagine even on a fairly easy test, the top scores wouldn't reach over 80% due to the near-impossibility of consistently keeping a high 5 to 3 ratio.

I'm looking at the 2013 Nationals test right now and it is rather underwhelming. Only 33 questions and many of them are fairly simple (mass of Sun, Neptune, and proton are all in here)

[fanjiatian]

Any ideas when we'll be getting our tests back?

Digital copies of tests and keys will be released hopefully in the next day or two. Some supervisors needed time to recover and then make minor edits to their keys. We apologize for the delay!

[Raleway]

Just wondering for those teams that went to MIT and PUSO, which one did you prefer? Princeton was by far the best invitational I have ever attended. Everything ran very smoothly, on time, and directions (map/campus) were easy to follow. I heard about the Game On debacle at MIT as well as the Towers issue and how Code Busters actually didn't have enough copies for each team.

Just my two cents: PUSO was the better invitational but MIT is the traditional place for top teams. Where are you guys thinking of going next year?

[terence.tan]

My reviews:

Remote Sensing: a solid test overall; it covered all topics well and had an appropriate difficulty level. The math was a bit easier than I was expecting, but I liked how some of the math questions were really creative (#methaneCowsFTW).

Dynamic Planet: easily the most tragic test I've ever taken, out of Science Olympiad, out of regular classes, out of anything I've found online...it had so many random and tough concepts from geology that we ended up guessing on pretty much everything past the first page, and I didn't even realize that there was a guessing penalty on the multiple choice... I'm surprised that we did so well (20th!)

Experimental Design: well-run, interesting topic, not much to say here.

Game On: the topic was a little silly, to say the least. I felt pretty iced after that event. Nice touch with having two people grade the game then averaging the scores! I haven't seen this done at any tournament so far, but I would imagine that having two people grade substantially reduces the randomness involved in grading this event.

Hovercraft: the test was a little short/easy for 40 minutes, but it was a solid test that covered most of the mechanics event topics well. The last question in particular was really interesting and we spent most of the time working on that question, although I thought Scioly wasn't supposed to ask questions that require Calculus? I wish there had been a bit more fluid dynamics questions, but it was a fine test overall. Kudos to the really nice scratch paper the supervisor provided!

Am i allowed to ask for the theme for game on and the materials for experimental design?

[pikachu4919]

My reviews:

Remote Sensing: a solid test overall; it covered all topics well and had an appropriate difficulty level. The math was a bit easier than I was expecting, but I liked how some of the math questions were really creative (#methaneCowsFTW).

Dynamic Planet: easily the most tragic test I've ever taken, out of Science Olympiad, out of regular classes, out of anything I've found online...it had so many random and tough concepts from geology that we ended up guessing on pretty much everything past the first page, and I didn't even realize that there was a guessing penalty on the multiple choice... I'm surprised that we did so well (20th!)

Experimental Design: well-run, interesting topic, not much to say here.

Game On: the topic was a little silly, to say the least. I felt pretty iced after that event. Nice touch with having two people grade the game then averaging the scores! I haven't seen this done at any tournament so far, but I would imagine that having two people grade substantially reduces the randomness involved in grading this event.

Hovercraft: the test was a little short/easy for 40 minutes, but it was a solid test that covered most of the mechanics event topics well. The last question in particular was really interesting and we spent most of the time working on that question, although I thought Scioly wasn't supposed to ask questions that require Calculus? I wish there had been a bit more fluid dynamics questions, but it was a fine test overall. Kudos to the really nice scratch paper the supervisor provided!

Am i allowed to ask for the theme for game on and the materials for experimental design?

All the exams have been publicly released on their website. You can find them there.

As you might guess from my username, I was the 'supervisor' for Fermi Questions and Dynamic Planet;

Honestly I don't think I could've guessed based only on username lol heh, but then again we never met last weekend so...ye.

EDIT: maybe I also don't know enough about anime

[BTABC123]

Anyone know how the top teams scored in mousetrap?