This week I had occasion to revisit my remarks in Week 42 regarding almanacs (Feel free to read that first as a refresher.) In doing so I came up with a more universal theory of what happened when almanacs disappeared from easy public view.
The things that filled an almanac to its full size are exactly the things that were pushed to the side in circuit competition.
Statistics went away because that's a weird and hard memorization to do.
Current events
Geographic information political in nature
Pop Culture
Sports
General Knowledge
Most of these types of information moved from their closest point of contact with a writer being in the almanac to their closest point of contact with a writer being wikipedia. But once that information is absorbed into wikipedia, the information is not connected with other almanac information in the same way, effectively diluting and obscuring it, and the information was perceived as a lower reliability product, even if it the sources of the information from which the wikipedia page was compiled was the same source as appeared in the almanac. At the same time, wikipedia was seen as an unreliable and incomplete source by the circuit. These articles weren’t unreliable, or poorly sourced, but the fastest route to use that information was now suspect, and the trip to get to all the information a lot longer. Almanac questions no longer supplied the quick production turnaround they had long provided.
The occasion that got me on this was spurred by last week's excavation. As I said last week, there were a set of papers where I kept the first page of a website's printout for the express purpose of going back to it later using web.archive.org. In doing that I wanted to remind myself of the contents of a site that I felt might be lost when the owner changed, or the website changed. For several of the ones I had found in the papers, my fears have come true, and my faith in web.archive was rewarded.
There were many efforts at the time to compile what in retrospect could have been considered quiz bowl almanacs. That is the common answers and facts of a field, compiled down into one-line facts. It was one of those ideas which people would come up with, compile about 20 or so documents, and then either publish to the community or just to their team, or abandon in place. A lot of those first sheets I had kept had gone into that sort of place.
As an example: faqt.org which was compiled in the late 90's early 00's by David Levinson, a player for Minnesota (and I believe Berkeley). I kept a reference to it in those papers because it had an excellent summary of American diplomatic history that meant I didn’t need to create my own study guide. When I found the paper again, web.archive.org could find it, but it’s one of the lucky ones from what I remember that archive having.
Add to these: Chris Sewell's lists which were at one point adjacent to the Stanford Archives, and my reference desk site and a directory of a bunch of study guides in PDF format. Both are out of date, and I stopped paying for a website a long time ago. Both of them now have parts that are better suited to high school competition, and parts that aren’t useful at all.
The point here is that most attempts to create an equivalent of an almanac for quiz bowl purposes fail, or worse degrade over time. There is a necessary use case for such a site, as it would make a lot of study for quiz bowl easier, but such a venture would also require a lot of maintenance that is hard to justify as more essential than creating questions or organizing programs and events. The costs of maintenance ensure it’s a loss leader for the circuit.
The problem is not limited to quiz bowl almanacs, infoplease.com, the last remnant of the Information Please Almanac is a mishmash of busted links and abandoned information made inaccurate and incomplete due to little maintenance.
We have two weeks of data on the new program, as they've aired two episodes. In the book, I deem it essential that you chart matches of at least two episodes to figure out what patterns show up in the writing. In my viewing, I picked up some patterns which I'm trying to figure out how to exploit.
I’m more or less convinced the writer is compiling this from a clean slate, and is to some degree genre-savvy (that they have familiarity with both televised competition and with circuit quiz bowl.) The latter has influenced both what they have included and what to this point they have not included. I’m italicizing that clause deliberately because there’s not enough information for some sweeping statements about what’s not there, having only charted about ~120 questions so far. But while we can’t make some statements about the distribution, the conclusions we can draw from the format changes, are worth investigating.
The key change to the lightning round, which existed in both formats, is that the subject of the lightning round is now the same for all three teams, and how the category is formulated is different. Previously the format of the lightning round was based on some lexical characteristic of the answers, and the complexity of what was elevated as rounds progressed. So your first match might have had all answers beginning with M, win that match and you're playing a round where all answers are five letters with O as the third letter, then all the answers had "SEA" in them, then all answers were a certain number of syllables. This is about as fair as you can expect, as it still enables a variety of categories within the round, and allows the approach I recommended
The new version is simply that all three lightning rounds for the teams have the same broad category. "Places" and "History" being the first two. For someone like me whose mind is organized around the subjects of quiz bowl this results in it seeming very swingy.
I'm also wondering if this is a product of these being the first two rounds produced by the new writer for the new project. The fact of a format change implies that new questions fitting the new format was created. The evidence of the questions indicates that source material was approached repeatedly in the construction of these questions.
This is where Dum Dums come in. If you ever saw the candy known as Dum Dums, you’d know it’s a multi flavor lollipop that comes in bags that are a staple of pediatrician reception desks. In the bags you would often run into the mystery flavor. The mystery flavor is what is generated from the transition between batches of flavors in the machine. When enough of a flavor has been produced the Dum Dum company has a limited number of machines making lollipops, so to keep the production going the new flavor is added to the machine as the old flavor is going out. In that transition, you have a mix of flavors, which is sometimes pleasant, and sometimes not pleasant, and which can neither be labeled with either the old nor new flavor label. Thus to keep their system, running they developed the “Mystery” flavor label. This is kind of what is happening with the new format.
If you consider the process of writing as individual pieces of art, you don't understand how television or any format that produces a large number of questions functions. The proper metaphor is not producing a painting. It is a series of objects produced in an industrial process. And if you understand that it is an industrial process that has to start and to stop and to function in a regular method, you understand better.
If I know I'm going to be extracting a large number of questions from a particular source and I will need a large number of questions to complete the task, certain patterns would emerge in the startup phase.
- I'd start with that source early in the development process.
- If I was worried about repeating myself in questions, I'd proceed through the source in a rigid front to back manner. So if it were a book, I'd go through the first pages first, and so on to the end. If I had to break off, I'd start again where I stopped, and only retrace to things I hadn’t used before.
- Because I'd have certain sources I'd be using up front, if I was still creating by the time the first deadline occurred (and with a project this size I would be), my first packets would be sampled in favor of the sources that I'd already started with.
Any sufficiently large project of quiz bowl questions that starts from a blank slate will start the first day with writing of favorites. The writer will write the things you think are important and interesting enough that they must be included. Once that vein of research is exhausted, the writer will realize it's not enough to complete the task, and will begin taking a structured approach of going through sources to fill the demand. And if those sources are books or websites in alphabetical order, your first couple of rounds will oversample from the letter A. If your source for history is in chronological order, your first couple of rounds will oversample from ancient history. However these patterns only assert themselves at the beginning of production, when you are limited in what questions have been written, and thus can be placed in the first few rounds of the project. You fill with what you have available, and what’s available will have a pattern to it.
As for why I know this, this is the reasonable tack that NAQT used in starting out. If you were present in the quizbowl mailing lists of the 1997-98 era, you might have seen a theory floated that NAQT wrote the entire tournament set this way, as the third or fourth set of the year that one coach had moderated had more mentions of answers that began with the letters D and E than he had expected, and made the prediction to plan for F's in the next set. The coach had sort of caught us, but by that point the backlog of questions was enough to break the pattern up. I'm not going to say this was the reason that we changed from having high school sets distinguished by letters to distinguished by numbers, but it certainly helped mask the criticism.
The other behind-the-scenes method that helps break this up is having your questions with a well-defined distribution of categories and subcategories which spans the entire set of packets being written. I wouldn't have picked up on this being a set of questions started from scratch if a set of ten answers hadn't all begun with the letter A. (This is the sort of hidden pattern that’s bloody obvious the moment you start charting.) A lightning round with a broad category can reveal those patterns because so many answers fit in that category.
This is why new projects and the first event of the year are mentally freeing for lots of writers, because they are freed from the tyranny of repeats, and can put in what they want to write about, without the necessary regard for making sure it doesn't get bounced out of the set. The project isn't a grind yet, and you can be more productive than you think you need to be. It's important to take advantage of that time, because the end of the project the writer and editor will never be as productive as they were at the beginning.
The reason I don't mind this is because this represents the best available practice if you are concerned with repeating yourself, and starting from zero questions in the backlog. It shows you're efforting to work for the long haul, and intend to complete the task. No one should in quiz bowl should ever root for a project to fail once it’s begun execution, because the damage done by a such a failure would damage all forms of quiz bowl. It also means (as a necessary consequence of best industry practice) that these first few rounds have extreme results. We’re still in the Mystery Flavor part of the transition, but we should expect a regression to the mean, over the course of the year. If the first few rounds are unpredictable in a way that can be reasoned out, we can also reason out what it will look like a few rounds later, when it's our turn in the arena.
A note that doesn't fit anywhere in any of my social media. I ended up subscribing to a podcast about the prosecution of Sam Bankman-Fried, and I have to say, having the podcast sponsored by an online casino is just a [chef's kiss emoji] of cynicism.
I have a couple other deductions I have about the product which I will discuss next week, as it’s getting late. I would finally just like to note that I used the term “the writer,” singular, to refer to who wrote these two episodes I would be very much surprised if there will not be more than one writer involved in all the season’s episodes. But I’m equally set in my opinion that the first two episodes which established the rules of the format will not be entrusted to anyone other than the person who designed the format. If you want the rules of construction to be followed throughout the season, you better be able to follow them yourself.