Repository navigation
Add the backward half of MacKay exercise 8.11 - #194
Conversation
Two exercises: a stationary N-gram model read backward is again an N-gram model, and so its backward conditional entropy equals the forward one. The second solution closes the 8.11 question: the reversed text carries no more information per character, and the backward guessing game is harder only because people's backward guesses are worse. Stacked on the PR with the remark and the first two exercises. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019xfc9pXuEzzVRdTKMWnCh8
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
The review on the backward PR noted that the last exercise asked two things (lab rule 7) and that the closing paragraph reused p, which the solution had defined as shorthand for a joint probability. The interpretation is now exr-reversed-text-information, and its pmfs are f (true) and g (the guesser's). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019xfc9pXuEzzVRdTKMWnCh8
|
Commit 1a5b778 addresses both review suggestions:
Term links pass, and Posted by Claude Code (AI agent) --- not written by a human. Generated by Claude Code |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
The review noted that "If not, why..." leaked the answer to the first question and split the exercise into two asks. The exercise now asks for one comparison and the explanation it supports. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019xfc9pXuEzzVRdTKMWnCh8
|
Claude finished review — View run Prior verdict stands. This round's one commit, The commit replaces the two-question prompt (more information per character? if not, why is the backward game harder?) with one imperative. The new text reads: "compare the information per character in a reversed text and in the text read forward, and use that comparison to explain the difference in difficulty described in @rem-reversed-text-guessing." That settles both rule 7 and rule 8 points from the last round. The "If not, why…" wording no longer hands the student the answer. The Findings
VerdictReady for merge, with the optional suggestion above. No new defects were found, and the prior Ready-for-merge verdict still stands. Structured Review Data (JSON){
"schema_version": "1.1",
"reviewer": "claude",
"commit_sha": "4874dd901f87abb95f143bf2dffb6e18f21a05e3",
"verdict": "CLEAN",
"findings": [],
"detailed_assessment": "Checked the reworded prompt in _subfiles/_exr-reversed-text-information.qmd: it is now a single neutral instruction, the cross-references are unchanged and resolve, and no math was added.",
"holistic_assessment": "The change is a one-file wording fix that resolves the earlier rule 7 and rule 8 suggestions, with no integration, regression, or scope concerns."
}Reviewed commit: 4874dd9 |
Built to branch |
Requested by Ezra · project thread
This is the third PR in a stack. Merge #191 first, then #193, then this one.
Why
The review on #193 asked for exercise 8.11 to be split into one exercise per part. The split put #193 over the 300-line cap, so the backward parts move here. With them, the page answers MacKay's question in full.
What
exr-reversed-n-gram-modeland its solution show that, given all the later variables, a variable in an N-gram model depends only on the N − 1 variables just after it.exr-backward-n-gram-conditional-entropyand its solution show that the backward conditional entropy equals the forward one from Add MacKay exercise 8.11, part 1: the forward direction #193.exr-reversed-text-informationand its solution answer the question: the reversed text carries no more information per character. People find guessing backward harder because a guesser that predicts with g in place of the true conditional pmf f pays H(f, g) = H(f) + D(f‖g) bits.Checks
check-term-linkspasses locally with the CI inputs.quarto render entropy.qmd --to htmlbuilds, and every cross-reference resolves.https://claude.ai/code/session_019xfc9pXuEzzVRdTKMWnCh8
Generated by Claude Code