The Saban-to-DeBoer Transition

The evidence pool for this chapter is again thinner than the argument would like, so this is going to be another short beat about the ambient signal — because the ambient signal has, through the first eleven days of October, done something new. Chapter 10 closed on a wire that had learned to spike and not to sustain: a count that could print 31 one day and 4 the next, with the Wikipedia floor holding beneath it. The read then was that the two curves had changed shape without closing — the readers were still there every day, and the press was arriving in bursts.

What the first week-and-a-half of October has done to those two curves is the whole chapter, and the honest version of this chapter is that the data pool for this window is smaller than the previous two. So let's keep this one tight, name only what the counters say, and leave the rest to inference.

“Keelon did a great job in the game, he played hard and showed he had the ability to hang in there with a top ranked team til the end. I still can’t believe that was only his 6th start at Alabama. There will always be…”

— a viewer comment on Alabama Crimson Tide on AL.com, YouTube, Oct 12

Here is the wire, per the public GDELT article-count series, across the eleven days that end this window. October 1: 2 articles. October 2: 5. October 3: 4. October 4: 2. October 5: 1. October 6: 2. October 7: 5. October 8: 76. October 9: 11. October 10: 288. October 11: 4. That is the shape Chapter 10 tried to name a month early — a floor that measures in the ones and twos, punctuated by two separate days that print in triple digits, with the second one almost four times the size of the first. On October 10 the GDELT counter logged 288 articles on an entity it had been logging two or three a day on for most of the preceding week. The next morning, October 11, it logged four.

Here is the counter, per the public Wikipedia pageview tool, across the five days this window gives us. October 1: 1,394. October 2: 1,574. October 3: 3,716. October 4: 2,541. October 5: 2,487. The floor Chapter 8 identified at 1,500 and Chapter 10 said was holding at a higher band is, in this snapshot, holding higher still — the lowest day in this five-day run is 1,394, and the next-lowest is 1,574, and the three days after that all clear 2,400. One of them, October 3, clears 3,700.

The honest reading of these two series together is the one this thread has been building toward for three chapters now, and it is worth stating cleanly, because the shape is clearer in October than it was in September: the fan curve has a floor and the press curve does not.

On October 5, GDELT logged a single article on an entity whose Wikipedia page pulled nearly 2,500 pageviews the same day. On October 6, two articles. On October 7, five. Three consecutive days in the first week of the month on which the readership of the Wikipedia page was running somewhere in the four figures while the daily article count was in the single digits. That is the same gap Chapter 9 named a canyon on September 5 — one article against 4,401 pageviews — and it is the same gap Chapter 10 said the wire was closing in bursts rather than sustaining. The October 8 wire count of 76 and the October 10 wire count of 288 are what the bursts look like now. The October 5 count of 1 is what the trough between them looks like.

The readers do not go away between the bursts.

Chapter 8 was the story of a floor at 1,500 holding against a wire that measured in ones and fives. Chapter 10 was the story of the wire learning to spike against a floor that had settled a notch higher. This chapter is the story of a wire that has learned to spike harder — 288 is a bigger single-day number than anything the September series produced in this thread — and a floor that has, in the days this window gives us, moved up another band. The October Wikipedia window never prints a day below 1,300. The late-August window Chapter 8 was tracking was the story of a program clearing 1,500 for the first time on back-to-back-to-back days. In this window, the lowest day is 1,394 and three of five clear 2,400. If a reader took only the Wikipedia column and no other evidence, the honest reading is that the baseline has moved again, and that the moves are getting shorter — August cleared 1,500, September held above it, October is sitting in a band that has 2,500 as a normal day rather than a notable one.

What is new in this chapter, and worth naming as the beat the thread is advancing to, is the asymmetry between the two spikes in the GDELT column. October 8 printed 76. October 10 printed 288. The two numbers are two days apart. Between them, on October 9, GDELT printed 11. After them, on October 11, GDELT printed 4. Chapter 10 described the wire as arriving in bursts and then falling back to nothing. In the October window the bursts are bigger and the falls are faster. The October 10 figure of 288 is the largest single-day article count that has appeared anywhere in the three windows this thread has tracked, and the day after it the counter printed a number smaller than the number it printed on an idle Sunday in the middle of September.

The temptation, writing this short, is to reach for context — to name what the wire was writing about on October 10, or to assume the Wikipedia bump on October 3 belongs to a specific game or a specific news item. The honest version is that this chapter cannot do that. The counters are the counters. What this window shows, held strictly to its own evidence, is that the Saban-to-DeBoer Transition is now eighteen months into a pattern where the fan curve has a floor and the press curve has only peaks, and that both of those things are getting more extreme — the floor is sitting higher than it did in August, and the peaks are printing bigger than they did in September.

Chapter 10 asked whether the canyon would close. Through the first eleven days of October, the answer is that it has not. It has gotten deeper on the bad days and narrower on the big ones, and the shape between them is the shape this thread will keep tracking.