What Does Data Think?
UPDATE: Tomato plant mystery: We are in a holding pattern. One plant remains. No sign of the thief.
In most of the media companies that employed me in the early years of the web, one of my responsibilities was supervising the metrics department.
In case that sounds like a big deal, this was well before the days of data scientists and multi-variable analysis; in most cases the metrics department in those media companies consisted of a lone individual.
And that person often felt like no one was listening to them.
After all, much more significant than the actual numbers he or she gathered was figuring out how to interpret that data. In and of themselves, of course, the numbers were neutral. But the people we worked with had a wide variety of opinions over what those numbers actually meant.
Was our audience growing? Which types of content were most successful? What was success in this type of media environment anyway? Which metric mattered most?
Occasionally, especially in the early years, we would publish a story that “broke the servers,” i.e., generated more traffic than our system could handle. There was little debate on those occasions over whether we had a winner, particularly because additional things tended to happen to support the data.
Things like attention from other media outlets, tons of email from subscribers and a boost to whatever financial metric we were tracking.
But these experiences caused me to eventually draw a few conclusions about people in general:
Many people are not very good at math.
Most of us see what we want to see in the numbers and don’t see what we don’t want to see.
Most of us don’t change our behavior or opinions even when the numbers say we should.
In the end, I wondered, what did the data itself think about all of this human frailty? That is one reason I have long been curious about the coming of generative artificial intelligence — we may find out the answer to that question.
Addendum 2026; I guess we’re finding out now.
(This is originally from 2023.)
HEADLINES:
Trump and Xi conclude ‘very successful’ talks but no deals confirmed (BBC)
Trump says Xi offered help on Iran — But how far is Beijing willing to go? (CNBC)
US-China summit: Trump says Xi pledged not to provide military equipment to Iran (CNN)
China will work behind the scenes to help reopen the Strait of Hormuz, Bessent says (CNBC)
Xi tells Trump U.S. and China could clash over Taiwan (PBS)
Cuba Says It Has Run Out of Oil (NYT)
Democrats get a last-minute reprieve on 2026 redistricting (Axios)
Senators approve withholding their own pay during government shutdowns (AP)
Abortion Providers Are Racing to Stay Ahead of the Courts (Mother Jones)
Trump administration officials are scrambling to contain the economic and political fallout of the war with Iran. (Reuters)
A ship anchored off the east coast of the United Arab Emirates has been seized and is heading toward Iranian territorial waters, the British military said. It happened hours after Israel said Prime Minister Benjamin Netanyahu had quietly visited the United Arab Emirates during the Israeli-U.S. war with Iran. However, the UAE swiftly denied that any secret visit had occurred. [AP]
Cuba says oil and diesel supplies have run dry under U.S. sanctions (CNBC)
Strategist Tied to Becerra and Newsom Pleads Guilty in Corruption Case (NYT)
How the Environmental Movement Lost Its Power (Atlantic)
Warning of record global temperatures as chance of very strong El Niño grows (BBC)
New Mexico politicians grapple with oil windfall from Iran war (AP)
You can reverse much of the damage alcohol has done to your body, science says (CNN)
I Work in Hollywood. Everyone Who Used to Make TV Is Now Secretly Training AI (Wired)
Anthropic tightens Claude limits and OpenAI courts defectors (Axios)
Frontier AI models don’t just delete document content — they rewrite it, and the errors are nearly impossible to catch (VentureBeat)
Silicon Valley’s A.I. Lobbying Reaches a Fever Pitch (NYT)
Trump Unwittingly Breaks Chinese Taboo Against Napping Facedown In Soup Bowl (Onion)
Whatever the AI data say we may not understand it. But we need to try. AI cannot be stopped but it needs guardrails. What does that mean? A threat to journalism or a boon to research and reporting? As you say, we are finding out. Soon.