Evan died; CodeCon

[Kragen Sitaker]

My friend Evan Doughty died in his sleep Saturday morning.  Nobody
knows why he died.  The autopsy, which finished Sunday, had
inconclusive results; no heart attack or cerebral aneurysm killed him,
but we don't know what did.  As I write this, I await my flight to
Ohio for his funeral.

You can read Evan's journal at
http://www.livejournal.com/users/evanfoo

He had lived 21 years before his death.

I miss Evan.  

He had the courage to take unpopular stands, but still listen to those
who disagreed with him; and he changed his mind when they showed him
he was wrong.  At Best Buy, he stood up against the store's policy of
illicitly signing up its customers for MSN when they bought a
computer, without telling them; this eventually led to his departure.

Among the Canonical Hackers, he had no fear to stand up for eccentric
beliefs and interests he pursued, namely, Christianity and
professional wrestling.  You might think that groups of hackers would
have a lot of tolerance for diversity, but until Evan forced us to
accept these aspects of his life, our tolerance for professional
wrestling fans and Christians sometimes included some condescension.

I didn't know Evan as well as the other CHs did, and now I never will.

You'd think that after such a sharp reminder of my own mortality, I
wouldn't get so irritated at minor frustrations, but so far, I remain
as irritable as ever.


I went to CodeCon (http://codecon.info/) this weekend.

Saturday morning, Beatrice and I had a long talk, after which we both
felt much happier.  But it lasted long enough that I missed most of
the first CodeCon talk, which concerned a pure-Python cryptographic
toolkit called CryptoPy, which seemed kind of fun; of course,
modifying crypto algorithms takes much less effort if you write them
in Python than if you write them in C, so CryptoPy makes algorithmic
experimentation much easier.

The OpenRatings talk, which concerned a web service for students to
share their opinions about professors, interested me.  Applications
like these could change our society dramatically, for either better or
worse.  In this case, the more efficient propagation of professor
reputation information gives students significantly more control over
their college experience, for better or worse.

The GNU Radio talk that followed impressed me much more.  Eric and
Matt have built software that can decode FM radio in real time (six
stations at a time), given the output of an ADC hooked up to an
antenna through a downconverter.  They also demonstrated, for the
first time, decoding of ATSC HDTV signals by the same means; they
showed us a short segment of "Law and Order" from a few days before.

This technology could give people much more control over their radios.
For example, your computer could listen to several radio stations at
once, playing only the songs you liked out loud.  Or it could blank
out commercials, or play all the songs backwards, or notify you when
your dedication to your girlfriend got played.

Extended to TV, it lets anyone build a super-TiVo out of their PC with
free software, but with support for HDTV.  (The software doesn't yet
run fast enough to make this practical; it needs 40 minutes to decode
each minute of ATSC.)

I won't say much here about bidirectional communications with
free-software-defined radio, but it should allow much more efficient
spectrum use and much more effective privacy.

The next talk discussed Neurogrid, a prototype flexible metadata
search system, filling some of the same needs as recent proprietary
photograph-management software: finding things by means other than the
filesystem hierarchy.

Saturday closed with a version-control panel, where Larry McVoy,
Jonathan Shapiro, and Greg Stein managed to avoid fisticuffs as Shap
and gstein explained how they had been stupid to think that version
control was easy, and how lm had explained to them that building a
version-control system would take three years.

The BitKeeper licensing policy came up only at the very end, as Shap
sandbagged lm with a really vicious promise to never screw OpenCM
users the way bk has been screwing some of its users (by yanking their
licenses).

Beatrice and I went to dinner with Mark Miller, Seth David Schoen, and
two other people whose privacy I will protect by keeping them
anonymous, although their MD5 checksums are
c16526ba8e12c687185a01e7382454a0 and fdd8a68d5f96fe0515cbd230bf3d7aab.

I enjoyed seeing Shap and MarkM and Mike Linksvayer again, among many
other people, and I was very happy to meet Ben Laurie for the first
time; some people, like Sofia Akber, I didn't even get a chance to say
hello to.

After dinner, we went to Club Ass IV, a party at our friend Laurel's
house (the House of Ass).  We saw lots of people there, including one
other CodeCon attendee, and danh, whom I hadn't seen in a long time.
Danh hasn't had as much free time since he got a full-time job again.

While I was there, I logged on to IRC and learned of Evan's death; his
friend David Sakmar logged on and told me and Jason.  My whole body
felt cold, then hot, then painful; I took a long time to believe Evan
had really died.

Beatrice and I went home around 3 AM.

Sunday morning, I arranged a conference call of the Canonical Hackers
to discuss how to dispose of Evan's effects and share what new
information we'd learned about his death.

Sunday at CodeCon, I missed the Alluvium talk, which discussed a p2p
media streaming system for non-live streams.  (I mostly know this much
from attending the Q&A section.)  Then the amihotornot.com folks got
up to talk, and really impressed on the audience the power of user
interface details to influence human behavior.

I went to a late lunch, then returned in time to see the end of the
Hydan talk, where Hydan's author described the various bits of
redundancy you can exploit to hide information inside executables.

We had a break in which Len Sassaman eulogized his friend Disastree,
to whom he dedicated this year's CodeCon.  He kind of slagged all the
so-called cypherpunks who, unlike Disastree, write no code.

Then a masked man came up to discuss Mixminion, a next-generation
anonymous remailer system to protect our right to anonymous speech on
the Internet.

Then Dan Kaminsky came up to give a really engaging talk about Paketto
Keiretsu, a collection of really ingenious networking hacks.  For
example, he demonstrated a TCP SYN/ACK retry time-signature
fingerprinting method, a way to tack extended information onto the
ends of IP packets without breaking backward-compatibility.

That night, I attended the California Community Colocation Project
benefit dinner, at which Mitch Kapor talked for a while about
Chandler.  I left to go home early around the time the trivia contest
started, so that I could spend time with Beatrice instead of listening
to trivia.  A reporter who'd sat next to me at dinner gave me a lift
home.  (So I could say to Beatrice, "I never did get a cab, but this
hot blonde gave me a ride home in her red convertible."  This amused
her a bit.)

It disappointed me that the elaborate deem sum dinner I'd paid for had
no vegetarian items, so I had to eat chop suey.

I missed Monday's CodeCon talks in order to go to work, which I really
regret, because Khashmir, DeepGreen, YouServ, Bayonne, and Advogato
all seem like more interesting topics than those of any talks I
actually saw, except for the version control, Paketto, and GNU Radio
talks.  But I went to a tapas place afterward with a bunch of the
attendees and Beatrice; at dinner there, Danfuzz gave me the inside
scoop on the Danger HipTop developer program and its limitations, as
well as some interesting tidbits about the platform.

I developed a sore throat yesterday, just as Evan did on Tuesday of
last week.  I hope I get better quickly.

-- 
<kragen@pobox.com>       Kragen Sitaker     <http://www.pobox.com/~kragen/>
Edsger Wybe Dijkstra died in August of 2002.  The world has lost a great
man.  See http://advogato.org/person/raph/diary.html?start=252 and
http://www.kode-fu.com/geek/2002_08_04_archive.shtml for details.