This shades towards plug-talk, except that it specifically
involves how we configure and use our Linux computers.

----

I use keyboards with clicky keys, sometimes in the same
room as devices with microphones. 

I read the mostly excellent "A History of Fake Things on
the Internet" by Walter Scheirer, 2024 Stanford University
Press, reminding me that everything we do leaks information.

The book points out that every pixel on a specific digital
camera imager has a different offset and gain - when you
post two photos, the pixel field can be analyzed to show
they come from the same imager, even if cropped or modified
in GIMP.  The techniques can easily detect image tinkering.

I was surprised to discover that the citation trail leads
to a paper I wrote for an integrated circuit conference,
decades ago (with a zillion cites, I've earned tenure of
I want it). 

All your web photos are belong to us.

---

Anyway, physical keyboard keys will also have these small 
variances, but mostly, so does your individual typing style.
A computer microphone hearing me type this would notice a
lot of backspaces; I type somewhat spastically.

After listening to a large enough corpus of typing, and
RECORDING ALL OF IT, and ANALYZING THE HELL OUT OF IT,
a smart-enough AI-like program could make some accurate
guesses of what specific keys I am typing. 

Also what keys I ALREADY typed in past sound recordings,
perhaps YEARS ago, with a long-enough audio recording file.

Including the SPECIFIC key sequences that I type entering
passwords.  Some websites and apps require that frequently.
MANY training opportunities for a clever program hooked up
to a microphone, perhaps a parabolic dish microphone
blocks away, pointed at the outside window of my office.

I just added some sound damping to that window. 

Yes, I've changed my passwords, but not the brain that
remembers them and the hands that type them; my mind and
muscles follow patterns that can vastly narrow down the
brute force search space for a password that works.  

The passwords may be machine-generated random strings;
my small hesitancies and mistakes while typing a random
string will also show up in an audio record.  Bracketed
by my grumbles: "type my password AGAIN???"

Typical phone conversations are less than 10 kilobits
per second compressed (with pauses); for a 2000 hour
work-year, 10% typing time, that is less than a gigabyte
per year.  With SSDs costing $30 per terabyte recently,
that is 3 cents a year per target.  Stored forever.

The surveillance microphone will cost a lot more,
but mass-produced electronics can be cheap as well. 
If the "microphone" is a hack on your smart phone,
perhaps government sponsored ...

... well, time to respond with "can't happen here" or
"why would they target me" or "xkcd/538 Security pipe
wrench", but then, that's what THEY want you to think.

It is amusing that some prefer that we waste our paranoia
on the poor and the foreign and the sexually different.
Or on the agro-Americans who suffer those sad paranoias.
But then, that's what THEY want you to think.

Sweet dreams!

Keith L.

-- 
Keith Lofstrom          [email protected]

Reply via email to