Webspinner Academy Introduction to AI · Handouts

HandoutStanding practice log

Failure Log

The assignment that outlasts the course. Four fields, ninety seconds an entry. The pattern across twenty entries is the calibration — no single entry contains it.

From Lesson 14 §14.4 Entries per printed page 2 Review cadence Monthly, across all entries Verified 13 Aug 2026

The two disciplines that make this work

Write the expectation before you look at the output. Hindsight silently converts whatever happened into what you “expected,” which turns a calibration instrument into a diary. Re-read the whole log monthly. The recurring failure — the same mistake in different clothes — is only visible across entries.

Before you start

Your first project must satisfy all three criteria from Lesson 14 §14.3. If your candidate fails one, change the candidate rather than the criteria.

The three criteria, and what goes wrong without each
CriterionWithout it
You do it repeatedly Feedback stays anecdotal and never accumulates into a pattern
You can tell good from bad instantly No ground truth, so step 3 of the loop silently stops happening — the toy-problem trap
Nothing breaks if it is wrong ten times You check so heavily that you never find the boundary — the mission-critical trap

The loop, for reference

1. Pick a task you can verify · 2. Do it with the model · 3. Check the output against ground truth you already hold · 4. Write down where it failed.

Step 4 is the one everyone skips and the one that produces the skill.

My project

Project, and how it satisfies all three criteria:

Entries

Date ______________   Task

What I expected — write this BEFORE looking

What actually happened — include successes

What I will do differently

Date ______________   Task

What I expected — write this BEFORE looking

What actually happened — include successes

What I will do differently

Date ______________   Task

What I expected — write this BEFORE looking

What actually happened — include successes

What I will do differently

Date ______________   Task

What I expected — write this BEFORE looking

What actually happened — include successes

What I will do differently

Monthly review

Read every entry since the last review before answering.

1. Which failure keeps recurring in different clothes?

2. Where was the gap between expectation and outcome largest? What does that tell me about my calibration?

3. What can I now predict reliably that I could not last month?

4. What should I stop delegating? What should I start?

Why this form and not a different one

The four fields reproduce the components the expertise literature identifies as active in deliberate practice: immediate feedback, time for evaluation, and repeated performance to refine behaviour.1 The expectation-before-outcome discipline exists because retrieval and prediction produce durable learning where review does not, and because learners systematically misjudge which of the two is working — the illusion of competence.2

Endnotes

  1. Primary K. A. Ericsson, R. T. Krampe, C. Tesch-Römer, “The Role of Deliberate Practice in the Acquisition of Expert Performance,” Psychological Review 100, no. 3 (1993): 363–406. doi:10.1037/0033-295X.100.3.363
  2. Primary H. L. Roediger III and J. D. Karpicke, “Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention,” Psychological Science 17, no. 3 (2006): 249–255; and “The Power of Testing Memory,” Perspectives on Psychological Science 1, no. 3 (2006): 181–210. doi:10.1111/j.1467-9280.2006.01693.x