SecondSourceJudgment rebuilt from primary sources
Research · Jul 22, 2026

Yesterday's paper on whether agent-optimization gains compound lands on a bigger front this brief tracks: where should AI's memory live?

Research Notes · Tracking update Trend (paper 2026-07; trend placement today)

From the Jul 22, 2026 daily brief

Our July 21 Research Notes covered this empirical study (gains from optimizing an agent pipeline are one-shot by default; only built-in regression control makes them stick; arXiv, 2026-07). What's new today is where it lands: on a long-running fork this brief tracks — should "remembering you across conversations" live in external data (retrieved and stuffed back into context as needed) or be written into model weights? This paper is new evidence on that front: if even an agent's self-improvement gains won't accumulate stably on their own, that's one more point for "accumulation needs an external mechanism — don't expect it to soak into the system naturally."

Long context (settled 2024)read a large corpus at once (an entire codebase)
The memory war (in progress)remembering you across sessions — external memory or model weights?

Open ?

Opposing claimthe weights side: distill knowledge into parameters (cartridges / CL route)

Our current read: unsettled, with the evidence so far leaning toward the external-memory side — for writing knowledge directly into weights, even "it goes in and it sticks" has yet to be demonstrated. What would prove this wrong: we set a 12-month observation window on July 14, expiring mid-July 2027; if two or more frontier labs ship parameter-level memory-writing in production, this read is void and we flip.

Subscribe free — first issue lands tomorrow morning

Just an email address, unsubscribe anytime. This is the only thing we ask of you.

More in this section