Three new cards, no new operators. Two themes organized the day's output: how AI-era market uncertainty changes what buyers need from vendors, and how practitioners are building practical methods to separate durable gains from temporary ones.
Theme 1, Durable positions need authority behind them, not algorithm gifts
April Dunford's June 11 essay (Buyers navigating rapid market change rely on a vendor's market POV to validate their purchase; without one, reps lose the sale before the demo ends) surfaces a problem that grows with market uncertainty. Buyers in AI-adjacent markets now consult an average of seven information sources, but per Gartner research, they still rely on vendor reps as the most important validation source. The problem is that most reps are equipped to sell the product, not to offer a view on the market's future. Dunford's mechanism: buyers in rapidly changing markets are not just asking whether a product is good. They are asking whether this product will still be the right choice in two years. A vendor with a clear market POV gives buyers the narrative they need to make that judgment.
"Reps remain the most important information source when buyers are researching a business problem or need, identifying a preferred supplier and securing internal support, and finalizing the purchase."
The practical implication runs from marketing through to sales enablement. A market POV must be distinguishable from a competitor's and deliverable by a rep in a conversation, not just posted on a website.
Kevin Indig's June 12 data (Search visibility earned through algorithm shifts reverts to baseline unless underlying authority supports it) adds a parallel finding from search. Google's May to June visibility window shows Macy's gained a large position jump in March then gave it all back by June, while Yelp posted its all-time visibility record and held it. Indig's explanation: Google reallocates shelf space within categories whose aggregate demand has not changed. A site that gained positions from an update without a corresponding gain in authority will lose those positions in the next refinement cycle.
"Algorithm-gifted visibility is a rental."
Both Dunford and Indig are pointing at the same underlying structure. Positions built on underlying signals (a vendor's real view, a site's real authority) persist. Positions built on temporary pattern matches do not.
- April Dunford, Buyers navigating rapid market change rely on a vendor's market POV to validate their purchase; without one, reps lose the sale before the demo ends. Buyers facing seven information sources rely on vendor reps as their most important validation source; a market POV gives reps the narrative that closes the purchase decision.
- Kevin Indig, Search visibility earned through algorithm shifts reverts to baseline unless underlying authority supports it. Google reallocates search shelf space within stable demand categories; algorithm-gifted visibility reverts when authority-backed positions do not support it.
Theme 2, Practical proxies for what AI systems can actually do
Simon Willison's June 9 post (Asking a model about an obscure open-source project it has no incentive to fabricate is a practical proxy for its training depth) offers a deceptively simple evaluation heuristic: ask a model about a minor open-source project it has no incentive to fabricate. If it recalls accurate details, it has the training depth to retain obscure content. If it does not, it lacks the parameters or corpus coverage to do so. Willison tested this by asking Claude Fable 5 and a smaller model about his own low-traffic tools. Fable surfaced files-to-prompt accurately. The smaller model missed it.
"knowledge like this is a reasonably good proxy for model size—you can cram a whole lot more details into a larger number of parameters."
The heuristic is narrow but practical for teams who need to compare models without running full benchmarks. A model that recalls obscure, edge-case content correctly is demonstrating something that cannot be easily faked: it retained a low-salience fact that only matters because it is true.
- Simon Willison, Asking a model about an obscure open-source project it has no incentive to fabricate is a practical proxy for its training depth. A model's ability to recall accurate details about an obscure open-source project is a practical proxy for its parameter depth and training breadth.