By August 6, 2027, which causal account of Google's August 5, 2026 AI leadership shakeup will be the COMMON UNDERSTANDING — i.e., the primary-cause account backed by the preponderance of credible retrospective treatments (major-outlet longform/investigative reporting, books, documentaries, on-record statements by the principals) published between Aug 6 2026 and Aug 6 2027? The shakeup: Jeff Dean, Sanjay Ghemawat, Oriol Vinyals and Quoc Le left Google to found Discovery Loop (a public benefit corporation to automate ML/science/engineering research, with Google as founding investor and cloud partner); Demis Hassabis stepped down as Google DeepMind CEO to become Chair of GDM and Chief Scientist of Alphabet (remaining CEO of Isomorphic Labs); Koray Kavukcuoglu became SVP of Google DeepMind reporting to Sundar Pichai (no CEO title). Resolution: choose the scenario whose PRIMARY-cause attribution best matches the modal credible retrospective account as of Aug 6 2027. Primary cause = the counterfactually decisive factor in the dominant account ('but for X, no departures/step-back that quarter'). A scenario can win even if elements of others are acknowledged as secondary contributors. Resolve 'S14 No convergent account' if no single account commands a clear plurality of credible retrospectives, or if the major retrospectives materially conflict with one another.

rationale:

The most likely outcome is that one-year retrospectives will converge on a structural explanation, specifically S2 (Regime change: 25%). While the literal events occurred on different timelines, major business-press retrospectives rarely settle on "coincidence" as a primary causal driver. S2 offers the cleanest synthesis of both moves without requiring a hard-to-prove conspiracy: Google is pivoting from open-ended research to product and commercial execution. This explains why Hassabis, who prefers scientific vision over product execution, stepped back cnbc.com, and why Dean's group sought startup freedom away from Google's enterprise constraints 2 sources. Bloomberg 2 sources and CNBC are already heavily carrying this frame, making it the clear modal favorite.

However, the literal truth of the events provides a strong floor for S1 (Two true stories: 15%). The timing asymmetry is stark: Demis Hassabis's transition was reportedly in motion for at least a year 2 sources, while Jeff Dean's move to found Discovery Loop was decided only in recent weeks wired.com. Google officially denied any connection between the two announcements axios.com, and key voices, notably Hassabis's biographer Sebastian Mallaby, are already framing the shakeup as a mere 'formalization' of existing realities businessinsider.com. If investigative reporters fail to unearth a deeper link, this official, unbundled narrative could hold.

Alternative unified theories remain highly viable but require further reporting to dominate. S4 (Control war: 14%) is fueled by Bloomberg's reporting that Mountain View is centralizing AI power, alongside Parmy Olson's noted 'suspicion' that Hassabis was outmaneuvered by Koray Kavukcuoglu 2 sources. S3 (Crisis accountability: 10%) anchors on the Gemini 3.5 Pro delays and talent bleed 2 sources. S3's ultimate viability heavily depends on how Gemini 4 is received; a poor launch could retrospectively cast the August shakeup as a clear accountability purge, while a success would neuter the frame.

Given the official denial and the immediate splintering of media frames within 24 hours of the shakeup, there is a substantial chance that retrospectives remain materially in conflict (S14: 13%). If no single book-length account or definitive investigation forces a consensus, the 'two true stories' PR line might permanently coexist with the structural regime-change and competitive-crisis narratives, preventing any single primary cause from achieving a clear plurality.

Tail scenarios have some evidentiary hooks but lack the connective tissue to explain the whole shakeup. S7 (AGI repositioning: 5%) and S5 (Talent repricing: 5%) fit Discovery Loop's mission and funding structure but fail to adequately explain Hassabis's step-back. S8 (Hassabis cascade: 5%) is contradicted by Dean's timeline, while S6 (Mission rupture: 2%) lacks direct reporting tying the principals' exits to the underlying safety-team unrest turntrout.com.