To be fair to @nathanalderman.com, this does seem to be a claim that CACVB (indirectly) makes, based on Virginia Tourism Corporation data.
It's just... wildly implausible, and suggests their survey-based calculations are wrong (and were never sanity checked).
👀
bsky.app/profile/rory...
bsky.app/profile/rory...
uhh when did they remove all the seating in Union Station?
VTC publishes a vague methodology document (mostly a list of data sources). A visitor is someone who stays overnight or travels >50 mi. The primary source of claims seems to be survey data.
Methodology: vatc.org/wp-content/u...
Full report: vatc.org/wp-content/u...
bsky.app/profile/tjsl...
Rory Stolzenberg
Rory Stolzenberg
Rory Stolzenberg
Rory Stolzenberg
Rory Stolzenberg
Rory Stolzenberg
Rory Stolzenberg
Now, given VTC's 50-mile criterion & Cville's rather large commute shed you might be tempted to argue that a lot of that spend comes from commuters coming from over the mountain rather than tourists... but 50 miles is a long way!
Some napkin math on F&B spend:
CACVB says 1.52m hotel + ~220k STR listing-nights annually, avg occ ~65% = 1.13m booked nights * ~1.6 guests/room = 1.81m guest-nights
$348.8m / 1.81m = $193 F&B spend per guest-night. Add in some day-trippers, subtract some lower STR occupancy… there's just no way.
Can the new DiffusionGemma model help fix broken OCR?
In theory, denoising tokens in parallel could work better for OCR correction since context is seen upfront?
Pointed it at 19th-century newspaper OCR. It corrected better than the autoregressive baseline — at ~8x the speed.
Rory Stolzenberg
did they define tourists as “anybody who does not live within the literal city limits”
