blog: The Loop Needs a Gate#61
Conversation
22bc14d to
6bb990e
Compare
ajbozarth
left a comment
There was a problem hiding this comment.
I haven't read/reviewed yet, but a couple comments on found on a glance. I'll dig into a review later this morning
| --exclude 'localhost' | ||
| --exclude 'mellea\.ai' | ||
| --exclude '^#' | ||
| --exclude '.*substack\.com' |
There was a problem hiding this comment.
why add this? This list shouldn't be adding external urls, if its failing, why?
There was a problem hiding this comment.
I dug into this and opened #62 with a better solution, so once that's merged you can rebase and remove this to make the PR cleaner
|
@planetf1 — this PR is scheduled to merge on |
ajbozarth
left a comment
There was a problem hiding this comment.
Blog looks solid and can probably go out today as intended, a few nits to clean up first though.
ajbozarth
left a comment
There was a problem hiding this comment.
Since we didn't get this out today it is now a blocker to update the date
|
All review comments addressed:
On the |
psschwei
left a comment
There was a problem hiding this comment.
I am fine with the overall piece (didn't do an in-depth review, but nothing beyond what @ajbozarth already mentioned jumped out)
|
@planetf1 — this PR is scheduled to merge on |
|
This is g2g, just need to remove the yml edit and rebase now that #62 is merged |
49ffb1b to
2ececed
Compare
Timely opinion piece riding the current loop engineering wave (Cherny, Steinberger, Osmani) and positioning Mellea's IVR loop and requirements-driven validation as the missing gate primitive. Signed-off-by: Nigel Jones <jonesn@uk.ibm.com>
Drop italic markers from the hashtag footer line in the blog post to satisfy MD036 (no-emphasis-as-heading). Signed-off-by: Nigel Jones <jonesn@uk.ibm.com>
Signed-off-by: Nigel Jones <jonesn@uk.ibm.com>
- Restore AlphaSignal quote to verbatim (full sentence) - Add cheap-gate paragraph: checking is narrower than generating, small models and deterministic requirements are enough - Replace origin paragraph with evidence links: Qiskit IVR benchmark (27.8%→50.3%), SOFAI cost routing, small-models-rock pipeline Signed-off-by: Nigel Jones <jonesn@uk.ibm.com>
Signed-off-by: Nigel Jones <jonesn@uk.ibm.com>
- US spelling: optimiser→optimizer, summarisation→summarization - Fix bold/link nesting on Making Small Models Rock link - Remove hashtag footer (belongs on social post, not blog body) - Remove trailing recap paragraph (content already promoted inline) - Update date to 2026-06-11 Signed-off-by: Nigel Jones <jonesn@uk.ibm.com>
2ececed to
dc8c763
Compare
Signed-off-by: Nigel Jones <jonesn@uk.ibm.com>
lychee auto-discovery of lychee.toml was added in v0.24.0; the action pin is v0.23.0 which requires the flag to be explicit. Signed-off-by: Nigel Jones <jonesn@uk.ibm.com>
CLI --exclude flags override (not merge with) the exclude list in lychee.toml, so the substack entry was being silently dropped while the YAML still passed localhost/mellea.ai/^# inline. Move all four excludes into the config file and drop the inline flags. Assisted-by: Claude Code Signed-off-by: Alex Bozarth <ajbozart@us.ibm.com>
470588b
Summary
Timely opinion piece riding the current loop engineering wave — Boris Cherny
(Anthropic), Peter Steinberger, and Addy Osmani's "Loop Engineering" essay (June 2026) —
and positioning Mellea's requirements-driven validation and IVR loop as the missing gate
primitive that every honest take on the topic admits is the hard part.
Test plan
/blogs/loops-need-a-gate2026-06-10and displays correctly/remind 2026-06-10