The Philadelphia Phillies reached out to let us know they are looking to hire a “Lead or Senior Quantitative Analyst, Player Evaluation” for their Baseball Operations Department. And they’re particularly interested in people who have experience with probabilistic programming languages like Stan and PyMC! For more details check out the job posting.
This is burying the lead. You’d get to work with Chris Fonnesbeck (co?)-founder of PyMC. Plus, the Phillies currently have the best record in baseball about 1/3 of the way through the season, and it’s always nice to work for a winning team.
I guess it’s de rigueur these days to give job applicants a quiz. I’ve never seen one in a job ad. This one reads:
and you get 250 words to respond. Defense is the cutting edge of analytics, so this makes sense. Everyone who knows baseball knows that catcher is the most complex defensive position: you’re the defensive captain because you can see the whole field, you’re calling pitches, you’re framing pitches, you’re blocking balls in the dirt, and you’re throwing people out on the base paths. You have to catch the occasional foul ball and make put outs at the plate. You have to run down the base paths to back up first or third on throwing plays. You have to scramble after bunts and you have to throw or tag batters out after they strike out. That’s a lot of requirements that are unique to catchers. And that’s just off the top of my head!
Hey, neat, we should all do this exercise and respond here in the comments tomorrow.
Phil:
This is like the classic “use a barometer to estimate the height of a building” problem, which as all those offbeat solutions (drop the barometer from the roof and time how long it takes to reach the ground; ask the building superintendent what is the height of the building, and offer him a free barometer if he’ll answer the question; tie the barometer to a string and reel it down the side of the building, etc.)
In this case, there are fun solutions like “conduct a survey of MLB managers asking them to rank the top 5 catchers in baseball,” “see who’s paid the most, after adjusting for the way that salaries go up with age,” etc. But I’m pretty sure that none of these joke answers would help a job applicant.
Since my team’s first string catcher has been injured since pulling an oblique on opening day and has still not been on the field, I would rank them by innings caught. Doesn’t matter how good you are if aren’t out there.
Interestingly, I once participated in an X-files esque mystery where mice were suddenly dying after coming out the door of a vivarium (ie. within 30 seconds). It was truly bizarre. No one had any idea what was causing it. At one point to prove that it wasn’t a sudden pressure change I created a barometer… It was essentially a couple of feet of thin silicone tubing with a binder clip holding one end closed and a bubble of water injected a few inches in past the open end of it using a pipette with a long thin tip that could be forced into the end of the tube…
So then I rode the elevator from the bottom floor of the building to the top floor of the building and watched the bubble of water move back and forth as the pressure changed (and made some marks on the tubing in sharpie). Then when we went in and out of the room where the mice were dying we could show the pressure difference wasn’t more than … a floor or two in an elevator.
So, you really can estimate the height of a building with a barometer and it works just fine.
PS: the final answer to the mystery my wife and I confirmed at about 11PM after staying in this trailer in a run down wearhouse all night attempting to rule out everything… that there was an ultrasonic motion detector on the lights in the vestibule of the vivarium trailer and the particular strain of mice involved had ultrasound induced seizures if exposed during a particular range of ages (a month to a few months old). The seizures would cause them to be unable to breathe and they’d die. You could revive them from the seizure by petting them rather heavily and essentially performing CPR. We notified the vet department the next morning that we’d figured it out and they removed all the ultrasound motion detectors replacing them with infrared.
One of the most direct serious *science* experiences of my life. Experimental results meant *literally* life and death.
Daniel, that’s a great story… sort of the science equivalent of an entertainment-industry story you’d find in Which Reminds Me, by Tony Randall and Michael Mindlin.
Thanks Phil, I really love that story. It was like an episode of House MD combined with The X Files except totally real-life. Even involved walking around a spooky dark run down warehouse with flashlights, because outside the trailer the warehouse it was in was basically an X-Files set.
Wow, what percent of the mice were susceptible to this? Makes me think of a friend who had to put down their dog for seizures, they definitely would have tried turning off any alexa or whatever in the house even based on hearsay like that comment. No doubt it depends on intensity/frequency and would hard to pin down.
Also, what made you think of that?
Anon. Apparently it’s common in swiss-black mice that were recently weaned, so several weeks old. This was a large percentage of this colony so basically every time a lab tech would leave the facility carrying weaned mice in cages for experiments they’d all run around like they’d been shocked electrically for 10 seconds then go stiff and die. It was extremely traumatic for the technicians. and of course the technicians began to wonder if *they* were safe.
The phenomenon was published already because another lab had induced seizures in their mice with an ultrasonic tattoo machine they were using for IDing the mice. We didn’t know about that, but once we identified the ultrasonic light detector we looked for papers on audiogenic siezures and found that result.
We discovered the motion detector was causing it because we were able to reliably bring the mice into the vestibule if the lights were switched off. Then I think we unscrewed the fluorescent bulbs and even with the lights themselves not turning on if the switch was on the motion detector itself would trigger the seizures.
It took literally like 6 hours to figure it out. we started after getting some take-out dinner or something and went deep into the night trying to determine what conditions caused the problem, what conditions didn’t, etc. This was already after a full day of brainstorming things we wanted to try before we got there.
Thinking back a little more now I don’t think we took the fluorescent bulbs out I think we brought the mice into the vestibule in a opaque cardboard box to rule out the lighting itself
I’m guessing its this:
https://www.jneurosci.org/content/22/23/10088
Ultrasound is > 20, so seems like its just loud (~ as loud as a motorcycle) noises that trigger it. Very interesting. Or is it those specific frequencies?
I tried it using this tool, and that is a very irritating sound. Listening to that at 100 db for a minute I’d think it would be a form of torture, even for a human.
https://onlinetonegenerator.com/multiple-tone-generator.html
Yeah, I think that’s the article we found. The motion detectors used sounds that are very high frequency. Above 19kHz, typically like 25-40kHz so perhaps they’re more susceptible at the higher frequency, or perhaps those things put out very high volume we just can’t hear it at all!
When I was younger I always had very good hearing well into the higher human ranges. Flyback transformers on CRT TVs would drive me crazy for example, Frequency about 15.7kHz according to wikipedia. https://en.wikipedia.org/wiki/Flyback_transformer and I definitely couldn’t hear the motion detectors.
I wasn’t trying to do the exercise, just brainstorm about the components. For that, you might want to do things like measure framing with a kind of Bradley-Terry model where you have different pitcher/catcher pairs. Or you might want to measure throwing by a combination of reaction time, speed, and accuracy.