Twenty-eight years of work, seven recurring problems, and
not one of them runs continuously.
Every thread in this career disappears for at least a decade and comes back.
The shortest absence is eleven years. The longest is twenty-four: multi-tenancy,
worked on in 1998 and then not again until 2023.
The second surprise is the clustering. Five of the seven threads return inside
the same four-year window, after absences of eighteen to twenty-four years.
The recent work is where the early work comes back, almost all at once.
The absences
Thread
Appears
Longest absence
multi-tenancy
1998 · 2023
24 years
the wireframe
1998 · 2000 · 2023
22
R&D labs
2001 · 2024
21
argument as deliverable
2003 · 2024
21
the specification
2004 · 2005 · 2023 · 2026
18
accessibility
2004 · 2019
15
data visualization
2001 · 2014
11
How to read it
Black is declared. The bars are one person's account of where he
worked and what the work was about. There is no external register to check
them against, which is the opposite of
the refinery piece, where every figure is federal
data.
Red is derived. The absences are computed from those dates. Nobody
asserted them, and they are the only claim here that follows from the data
rather than from memory. If the declared part is wrong, the red is wrong
with it.
Grey bars are teaching and school, which run alongside the work
rather than instead of it.
What this cannot tell you
Whether the threads are real or retrofitted
They were assigned looking backwards. A thread named in 2026 and traced to
1998 is a pattern someone decided was there. The dates are checkable; the
categories are a reading.
Why the gaps happened
An absence could be a problem going out of fashion, or a career moving away
from it, or the same work continuing under a name nobody wrote down. The
chart measures silence, not cause.
Two entries carry no year
Blade Ventures and Anmahian Winton Architects are in the data and not in
the drawing, because an undated bar would have to be placed somewhere and
any placement would be invented.