How close are you?

The mechanic

The goal-gradient effect

A percent-to-next-level indicator shows exactly how close someone is to their next milestone — “68% to Trail 2” instead of just “you have some points.” It works because of the goal-gradient effect: motivation increases as people see themselves nearing a finish line, not standing at the start of one.

It gives students — and parents — a concrete, visual sense of momentum rather than an abstract point total.

We all like numbers to an extent. Knowing how close we are often gives us the motivation to reach the finish line. This is one of the ways I think learning targets are effective. If a student understands what they should be learning, then they will know when they learned it.

---

Ethical AI

Explain it again, differently

Explaining concepts in multiple ways until one clicks is one of the easiest ways to use AI like a research assistant. Here is an example.

Almost every teacher will tell students something like, “You should have learned this last year,” or “Based on the standards in 8th grade, you should already understand…”

A common student response? “He/She didn’t teach me that.” Or, “I had a bad teacher.”

Incoming TRIGGER!

It doesn’t matter. Honestly. We’ve all had bad teachers. I have been a not-so-great teacher some years. And the reality? Unless you are fully done with education, you have the possibility of getting another one. College professors can be tough.

We live in a time of information. Everything we could possibly want to know is at our fingertips. So. Let’s live like it.

I teach my students to use AI like a research assistant. If you are expected to know 8th grade physical science standards so that I can teach you the ones for 9th grade, then use this powerful tool with fidelity and ethical intention.

You can ask an AI to review your 8th grade physical science standards with you. That would prepare you for the next level. Did you struggle with the 8th grade standards? Ask the AI to explain it to you like you are in 5th grade!

This is not a letdown or a negative regression. This is finding where you are and going forward. Stop using bad teachers as an excuse. To be fair, I would like to have some serious words with a lot of these teachers, but… I read this book that tells me I should not.

But wait! There’s more!

Does it still sound complicated? OK. Ask the AI to explain it to you in terms of your favorite hobby. Whaaaaaat? Ever experienced physical science as though it was explained through figure skating or underwater basket weaving? My football students start to get animated when it is shown using terms they almost worship. The year we used crochet terms was most interesting.

But that’s just it! Why live in chosen ignorance? Why not learn in ways that excite us? I tell you this: if we wanted my son to learn something, referencing NASCAR usually got his attention.

Whether you had a bad teacher or one of the too-often-rare good ones, you still control what you know. If you can spend hours watching someone else play video games, you can find ways to understand the information you need.

---

The read

The diagnosis beats the cure

Dario Amodei — We Must Pace the Frontier

An essay by Anthropic’s CEO making the case for “pacing” AI development.

What works

• The three-tier structure — unilateral, then industry, then global — gives the proposal escalating specificity rather than vague appeals to “go slower.”
• Embedded evaluators is a genuinely concrete, falsifiable commitment. Unlike most safety pledges, it’s checkable.
• It owns Anthropic’s own incidents rather than only pointing at competitors — it cites the OpenAI–Hugging Face incident, then acknowledges similar ones at Anthropic — which lends some credibility.

Structural weaknesses

• Thin on actual teeth. Only step 1 (embedded evaluators) is a real commitment Anthropic controls. Steps 2 and 3 are calls for others to coordinate — the same genre of aspirational statement the AI industry has produced repeatedly without binding follow-through.
• Key terms stay undefined. The “checkpoint” scheme — capability X triggering certification Y/Z — never specifies what X, Y, or Z actually are. For an argument whose central selling point is verifiability, leaving the verification criteria abstract is a real gap.
• Self-serving positioning isn’t addressed. A frontier lab proposing costly compliance infrastructure (embedded evaluators, third-party audits) benefits disproportionately if that infrastructure becomes an industry norm or regulatory requirement, since it raises the bar for smaller competitors. The essay doesn’t engage with this incentive at all.
• The China caveat quietly does most of the work. “Slow down, but not more than our lead over China allows” makes the actual policy ask much thinner than “pacing” suggests — it’s closer to “keep doing roughly what we’re doing, plus some evaluators,” while the rhetoric implies a bigger shift. It also seems to want to keep fear an important factor.
• Unsubstantiated urgency claims. “Since roughly this summer, AI has been advancing drastically faster” and the 6–12 month botnet-takeover estimate are asserted, not evidenced. I should just trust institutional authority rather than shown data. I don’t think so, Tim.
• The airplane analogy cuts against the proposal. Aviation’s safety record came from a regulator (the FAA) with binding authority and mandatory incident investigation — not voluntary industry pacing. Citing it actually argues for stronger regulation than what’s proposed.
• Costs of slowing are gestured at, not weighed. Given the essay’s own opening claims about AI curing diseases and driving abundance, a serious treatment would quantify what delay costs against what it buys in safety — this stays qualitative.

The diagnosis is more rigorously argued than the proposed cure, which leans heavily on voluntary coordination in an industry where voluntary coordination has a poor track record.

---

Field notes

Earning the grade

After reading issue 001, a teacher sent me this:

My first weed-out class at Tech was actually graded that way. Students hated it. I didn’t. I knew exactly where I stood and exactly how to get where I needed to go. I EARNED the grade I received by the skills I was able to demonstrate. I have always felt that being able to give students this alternative over a percentage grade is much stronger validation for what a student has actually learned. Often I’ll do scoring matrices for projects — they earn points for the things they do on the project connected to concepts and parts of the design process. They know what they’re working towards. — A teacher in the field

I would like to hear from you. Share your stories and experiences with me and the readers. No names will be mentioned. FERPA be upheld! But hearing from others always connects us. Too often teachers are isolated and unheard. Help me change that!

Just hit reply — it comes straight to me.