Find tomorrow’s headline
in today’s data.
Companies report their results later. Alex studies alternative data—signals outside required company filings—to estimate what those results might be.
The Jobs Podcast · Alex + Nathan
Signal
& Noise.
Satellite corn. Seven-figure questions. An army of AI agents. And a guy who once dropped out of statistics.
01 / Follow the clues02 / Question the numbers03 / Keep the human04 / Go ride a bike
The job, translated
What does Alex actually do?
Find tomorrow’s headline
Is that a real signal,
or a very good disguise?
Think political polling, with business questions. Who is in the sample? Who is missing? Does the model tell us something useful, or just something flattering?
Explain it to the person
holding the checkbook.
His team evaluates potential data purchases. The person deciding whether to write a seven-figure check needs to understand both the opportunity and the caveats.
A tiny data playground
Four dots.
One big temptation.
A satellite can photograph a field every day. But if you only have a few reported harvests to compare it with, a great-looking relationship can be misleading.
Here’s an illustration: four cherry-picked dots from unrelated measurements. They look remarkably convincing. Now let in more evidence.
Simulated data, inspired by Alex’s example. These are not actual crop measurements.
4 observations
On the human part of the work
“The value of seeing a problem clearly and thinking through a solution clearly has gone way, way up.”
— Alex
New tools. New headaches.
AI does the knitting.
Alex checks the pattern.
He compares familiar coding work to knitting: methodical, satisfying, sometimes almost relaxing. AI now handles much of that execution. What’s left is a day full of decisions, experiments, and careful review.
01
The agents do the legwork.
A coordinating agent for each project delegates work to other agents. Alex reads their output and pulls the pieces together.
02
The weird mistakes need new eyes.
We learn to spot human errors. AI can get an unexpected part wrong—the very part a human reviewer might skip.
03
The priorities stay human.
Alex reads plans carefully, challenges weak choices, and shapes the questions before trusting the answers.
A few favorite detours
Wait, tell me more.
The data scientist who dropped statistics
Alex hated his high school statistics class enough to drop it—the only class he says he ever dropped. What changed? Discovering that statistics is full of elegant math and interesting problems, rather than just finding the right row in a table. A bad introduction doesn’t have to be the end of the story.
The satellite corn example
Imagine an investor studying agricultural companies. A data provider offers satellite images that highlight chlorophyll. Could those images help predict crop yields before the companies report them? The difficult work is translating a daily visual signal into a reliable estimate—and recognizing how little a handful of past harvests can prove.
“Do the data’s hair and makeup”
Nathan’s playful description of making a dataset look good gets an enthusiastic response from Alex. Underneath the joke is a serious job: work with providers to understand how a signal should relate to the real world, then test whether the evidence supports it.
The office, annoyingly, has a point
Alex admits to being annoyed by how valuable it can be to sit next to other people and talk through problems live. Relationships, facial cues, and shared context still matter—even in a conversation about extraordinarily capable AI.
Advice for the professional puzzle enthusiast
Alex describes his work as distilled problem solving. If early career life makes every job look soul-crushing, his message is encouraging: interesting problems are out there. Curiosity, clear thinking, and the willingness to really understand something remain valuable.
The takeaway
Stay curious.
Question the dots.
Keep the people.
A conversation about work that turned into a conversation about being human.
Back to the signal ↑