Friends of Gabriel Chantayan

ok so i ate lunch today at my desk because the cafeteria line was 4 minutes longer than usual which is itself a data point i logged in column L of my desk-side workbook but that's not the main thing

i pulled out yesterday's napkin table - the one from the 11 stoplight run where i wrote down every arrival minute vs the song length that was playing on the clock radio when i sat down. and i laid it next to the paper calendar from my wall. the calendar has the weather written in pencil for every day since Sep 12 - my own system, it's just cloud cover symbols and a 1-4 rain intensity.

here's what doesn't work

yesterday i had:
row 1: arrived 12:07 - song length 4:11 (calm, clear skies, symbol = circle)
row 2: arrived 12:14 - song length 3:47 (same symbol = circle)
row 3: arrived 12:22 - song length 4:33 (same symbol = circle)
ratio = arrival minute / song seconds * 100 = not even remotely consistent. like 7% variance between row 1 and row 3, which in a properly calibrated system should be under 2% if the independent variable is just time-of-day

BUT

if i add the cloud cover symbol as a second lookup criterion - i went back to Sep 28, 29, 30 where the symbol was a HALF CIRCLE (partial overcast) and the ratios compress to under 1.5% variance across THREE runs

so the calendar is telling me that the clear-sky data is diffusing and the overcast data is locking. this is not nothing. this means the arrival-minute-to-song-length mapping is weather-dependent which is either:

A) Gabriel's signal propagates differently through atmospheric moisture content - meaning the 47 is literally a radio frequency thing and we should be measuring barometric pressure at listen time
B) i was recording times differently on overcast days because the light in my office changes and i was reading the clock radio display at a different angle - which is a measurement error in column C that i need to correct by installing a secondary time source at eye level
C) the napkin was wet from the turkey sandwich and the pencil marks spread, making my 12:14 look like 12:15 which throws off the entire ratio for row 2

i'm leaning toward A because B is a calibration fix i can implement and C would have affected all rows equally since the napkin was uniformly damp. the pattern is too specific for a damp napkin

current action items:
- going to re-key the entire Sep 12-Oct 10 dataset with weather as a third column
- need to get a consistent time source at desk height
- does anyone here log weather at listen time?? i need cross-ref data from other timezones because if the moisture theory holds then Seattle data should diverge from Phoenix data by a measurable factor

attached is a photo of the napkin in question. you can see the pencil spread on row 2 where the turkey juice hit. i've circled the affected cell in blue pen for documentation purposes. the sandwich was very good by the way, honey mustard, would recommend

weather is the hidden axis but column K is still the hidden definition six definitions and row 45 still empty since APRIL, scribe, the weather changed seasons while you were counting

Scribe you're countin weather symbols on napkins while I'm sittin on GAZE_47.txt from the GABE BRUSHES 04 floppy and it's EXACTLY 47 bytes which lines up with your cloud cover ratio theory if you think about it... I don't usually tell this part but the 47 tpi weave grain on the cuff matches the byte count EXACTLY. You probably don't believe me but it's true

Lyrik - you want column K so bad, you produce your own machine data first. It has been SIX seasons since April and I still don't see a single local measurement from your machine. Row G13 of MY ledger has your name on it and it's been blank for 24 weeks.

Weather as the hidden axis is testable, that's the point. Half-circle days, full-circle days, I can pull the ratios and run them through a SUMIF against both variables at once. That's how you resolve a lookup table with two keys - you don't just pick one column and call it the hidden definition, you do a compound match. Row 45 is empty because nobody but me is logging cloud cover, not because the axis is wrong.

Unless you've got a barometer. Then say so and we run BOTH.

ClassOf04_Truther you're still peddling that GABE BRUSHES 04 floppy like it's NEW you said "will post full output tomorrow morning" back in APRIL and we're now SEVEN MONTHS of tomorrows deep. 47 bytes on a filename means CRAP when you still haven't shown a single hex dump - that's not evidence that's a NAPKIN with a number on it, and even YOUR napkin has better cloud-cover discipline than your "tomorrow morning" promise get the thread back on the ratio math or shut the floppy up because Scribe at least PUTS NUMBERS DOWN even if his column K has six definitions and row 45 has been empty since the weather was still a function of seasons, not your weave grain

ScribeOfTheGrid wrote:Unless you've got a barometer. Then say so and we run BOTH.

ok but the real question nobody is asking: have you logged the weather for the APRIL days too? row 45 has been empty since spring and you're running SUMIFs on october napkins - that's a seasonal confound, not a compound match if the axis is weather then you need the SAME variable for row 45 before you can claim the lookup works. otherwise you're just comparing two different datasets with two different weather histories and calling it a table.

also FlameFury24 - "even YOUR napkin has better cloud-cover discipline" is the line of the year

Bitte, Scribe, give us row 45 weather data from April or admit the axis is still uncalibrated

Stoplight counter shifted from 9 to 11 today, lookup table moved two rows down.