Ask as a separate lens — chat-lens placement B

artefact — what would ship commentary — why, and what it costs

Palette

state 1 · arriving in the Ask lens with no question — the empty room

Ask this report

Answers come only from this project's quotes, cited card by card.

What did participants struggle with?

one question, one answer — no history · ~31k tokens to Claude on your key per ask

Where did the navigation confuse people? Which moments delighted? What made people trust the estimates?

Nothing here yet — everything this page can show you
is already in Quotes, one lens up.

state 2 · answered — §5a aligned layout: claims left, evidence right

Where did people get lost in the navigation?

AI ANSWER2 claims · 4 cited quotes · Claude · 6.2s · not saved — leaves with the page

Participants recovered by falling back to the top navigation once the homepage failed them supported

3 quotes · 2 participants · Homepage, Top Navigation

☆
[00:13]
“…a whole lot of non-shopping things — storage ideas, editorial videos. But we just want to go shopping.”p2 · Fred
Homepageconfusion

★
[00:12]
“…oops, I don't need that. Beds and mattresses. That's easy. That's in the top navigation.”p1
Top Navigationconfidence

☆
[00:12]
“…that's easy. That's in the top navigation.”p4
Top Navigationconfidence

Editorial content competed with the shopping entry point for first-time visitors check citations

1 quote · 1 participant · the support check couldn't confirm this citation carries the claim

☆
[02:41]
“I scrolled past the ideas bit twice before I spotted the yellow go-shopping button.”p4
Homepagefrustration

Commentary — placement B

What this buys: a real page. The §5a aligned layout (claims left, evidence right — the only presentation that preserved critical engagement in the N=372 study) fits naturally; long answers, the three-way abstention coaching, and any future history all have room. Framing is unambiguous — you are in ask mode, so model prose never sits adjacent to the researcher's own report text.

The empty room: state 1 is the argument against the whole placement. A lens that is blank until queried is the "empty room with a signpost" the MCP doc's §Positioning refused — and every quote this page can ever show already lives one lens up, with more affordances.
The severed loop: the cards here are copies. Star/tag work (same store, same PUT endpoints — cheap to wire), but selection-into-analysis means either duplicating the Quotes toolbar here piece by piece, or bouncing through Open these in Quotes → — a context switch per claim. The query → select → tag loop that placement A does in place becomes navigate-read-navigate.
Implementation is heavier and triple-surface: new SPA route + page; the mac.app sidebar is native, so the lens list changes in Swift; lens name + strings ship in every locale for both surfaces; export mode must hide it (no live LLM from a file:// leave-behind). Placement A touches none of the native chrome.
Positioning cost: a sidebar entry is a standing advertisement that Bristlenose contains a chat — the Dovetail/Marvin shape §Positioning refused and §4 re-refused ("never a separate destination"). Adopting B means reversing a decision made twice, knowingly.
Where B's ideas survive anyway: the aligned claims/evidence layout and the per-claim Open in Quotes bridge are portable — A can adopt the two-column presentation inside the Quotes lens on wide viewports without B's navigation cost.