The Library Read

598 frames read. 350 study. 214 discard. 31 strong. 1 portfolio. 2 exhibition. That is 5.7 percent at strong or better, against a sample drawn entirely from favorites — 18,431 hearts out of 257,495 images, the frames you went back for and marked.

The disagreement between your thumb and my ladder is the first finding, and it should not be read as an indictment. A favorite is a vote of love: this day mattered, this person mattered, this light reminded me of something. My ladder measures a different thing — whether the frame argues, whether a stranger with no access to your memory would stop. Most of these 564 study-and-discard frames are doing their job. Their job was never the wall. The heart-tap records a life; the tier call asks for a photograph. Twenty-six years of favorites running 58.5 percent study is not failure — it is the normal ratio of living to making, and the calibration says so: most frames are study. That is the scale working, not the scale being cruel.

One caveat, stated once: 188 of 598 frames were read from small preview derivatives, the originals offloaded to iCloud. Some of those reads may shift a band on full resolution — flagged for second read — but a tier distribution this lopsided does not flip on pixel count.

Early digital, 2000–2007: 19 frames, average 27.6, nothing at strong or better. The best of the era, a 2004 child portrait at 38, says it plainly — "the image records a child well; it does not yet construct a photograph." That sentence describes the whole period. The camera is a recorder. The eye has not yet claimed the frame.

2008–2014: 31 frames, average 31.2, the first two strong-or-better calls. The yearly numbers jump around — 39.5 in 2008, 24.7 in 2012, 42 in 2013 — which is what an eye looks like when it is settling: occasional intention, not yet a discipline. The 2008 Burning Man armchair frame (61) is the first time the evidence shows a formal idea — "the sign, the costume, the armchair, and the void are in genuine formal and semantic conversation."

The Sony years, 2015–2019: 226 frames, and the average drops to 26.6, the lowest of any era. 2016 bottoms out at 20.8. The volume tripled and the hit rate fell — the camera got better and the favorites got looser. But the nine strong frames from this period are different in kind from what came before: the Velella stranding (2015, 61), where "the diagonal composition is earned rather than imposed"; the columbarium self-portrait (2019, 61), a genuine conceptual proposition — "ease as a posture of intimacy with death." The era reads as appetite outrunning selection.

2020–2026: 322 frames, average back to 31.2, and 23 of the 31 strong-or-better frames in the entire sample. 2022 averages 34.3, 2023 averages 34.7 — the two best sustained years of the twenty-six. This is where the seeing sharpened, and the justifications say what sharpened it: decisions. "The decision to frame tight — denying sky and scale — is the right call." "The backlighting decision is the image's real intelligence." "The decision to use shallow DOF... is the correct formal choice." The late frames are not luckier. They are chosen before the shutter, not after.

Two exhibition calls out of 598, and both require honesty about what was scored.

The 91, dated December 2025, is Man Ray's Les Larmes, c. 1932, photographed on a gallery wall. The score belongs to Man Ray: "the score is applied to the work, not to the act of photographing it." It stays in the report because it tells the truth about your eye — you stood in front of a canonical image and knew to keep it — but the tier is borrowed, not made.

The 84, from September 2013, is the one that holds on its own terms, and it is also a photograph of a photograph: a mid-century Jewish wedding, the print itself dying — "peeling emulsion, the foxing, the torn upper register" — and the damage "inseparable from the image's meaning: a ceremony of survival photographed on a medium that is itself dying." The bride's veiled face at the center of grave witnesses produces "a punctum that does not release." This is the strongest act of seeing in the sample: recognizing that the object's mortality had become the subject, and framing for it.

So the finding stands without softening: in twenty-six years of favorites, the two frames that reached the top tier were both made in front of someone else's image. The eye that judges is, on this evidence, ahead of the eye that hunts.

As a partner

Kristi enters the favorites in 2018 — 75 appearances, no prelude. The curve is steep: 187 in 2019, 285 in 2020, her peak and the archive's single largest presence in any year. That peak is the lockdown year; when the world contracted, the frame contracted to her. Then the descent — 115, 86, 73, down to 25 in 2024, the thinnest year of the relationship's record. 2025 and 2026 recover to 86 and 79. The data shows a partnership that was photographed hardest when nothing else could be, then nearly disappeared, then returned to a steady middle register.

As a dad

The boys are scattered single digits from 2002 to 2013 — one frame here, four there. Then 2014: Charlie at 11, Jacob at 12, and the curve climbs together to a shared summit in 2017 — Charlie 111, Jacob 108. After that, divergence and decline. Charlie holds longer: 109 in 2019, 64 in 2020, then the slide to 6 in 2025 and silence in 2026. Jacob drops faster, flickers back — 35 in 2023 — then 1 in 2026. The fatherhood record peaks when the children are still in the house and thins as they leave the frame on their own legs.

As a traveler

One or two travel cells per year through 2013 — a life lived in place. 2014 opens it: 9 cells, then 19, 19, 17, 22, peaking at 24 in 2019. The collapse is legible to the year: 14 in 2020, 10, 5, 4 by 2023 — six years of range erased to near the 2013 baseline. The return is not gradual. 18 cells in 2024, then 27 in 2025 — the widest year in the record, wider than anything pre-COVID. 11 cells by 2026, a partial year already matching the mid-2010s. The traveler came back larger than he left.

The phases, read together

The partner fills the frame exactly when the traveler vanishes, and the traveler returns exactly as the children go — three curves trading the same finite attention across 26 years.