Guy Kewney's Weekend Diary

The hotel problem in San Francisco over the weekend was easily solved. I went to Santa Rosa. There, in a "resort" hotel called the Flamingo, were gathered together the most extraordinary collection of famous computer people I've ever been involved with. Names like Ken Thompson, creator of the "C" language; writer Neal "Snow Crash" Stephenson; astronomer and "Cuckoo's Egg" author Cliff Stoll; Roe Adams, writer of Wizardry and Ultima IV; Mike "Rexx" Cowlishaw; Steve "Behemoth" Roberts... oh, the list goes on. And I was there!

As a result, I didn't sleep much.

These guys call themselves "hackers" - a term which formed the title of a book by Steve Levy. They do not mean "people who dial illegally into other people's computers". They mean hackers in the canonical sense of "those who produce utterly brilliant hardware and software solutions in an afternoon, and describe it pseudo-modestly as a 'quick hack' when they actually know that nobody else could have solved the problem at all, never mind as elegantly."

What "hackers" actually means, is people who don't go to bed. I met Richard Greenblatt, one of the legendary MIT hackers on my arrival on Friday night around half past midnight. In his youth, he was notorious for actually making "nests" for himself in the false ceilings of the MIT computer rooms so that he didn't have to sneak in after hours to get some "hands on". I'd missed the evening's opening sessions, so I asked: "How's the conference going?"

"Dunno," Greenblatt said candidly, "just finished sleeping."

Very, very little of what went on is publishable; the Hackers did once invite journalists to the conference, and the next thing they knew, the TV station was broadcasting an item about the worldwide terrorism threat from computer break-ins, illustrated with footage from the conference. "Never again," said they. So it's all "off the record" and all I can tell you is that one of the Hackers found me a spare bed, on Sunday night, down in Silicon Valley. It cost me a damn expensive dinner. Worth it, mind.

Tuesday

A contact from the Hackers conference, asking for anonymity right now, shows me his newest hack. It's an advert in the New York Sunday Times, and it is complete gibberish; a sequence of 40 letters, characters, upper, lower, and other printable rubbish.

"They wouldn't print it, at first," he tells me; "they thought it was some kind of Mafia scam."

What it actually is, is signatures of signatures. He is recording "original files" so that you can prove that your file existed on such and such a date.

Obviously, the foolproof way would be to print the hex dump in the Financial Times. After all, it would look much like the share prices anyway. Apparently, the enormous cost of buying 20 pages of FT small-ad space seems to deter poor programmers.

So what this chap does is to create a signature. Each second, he takes all the files people sign, using his software. They send the signature to him, and at the end of the second, he creates a signature from all the signatures he has, including the previous second's signature.

At the end of a week, he creates a signature from all the signatures, and prints it. You can, if someone says they invented it, prove that the signature published can only be derived from your file, and that no other file could have provided it; and that therefore, your file existed before that date. You can actually pin it down to the second, if you go to my hacker friend and ask him to unstitch it all.

Apparently, it takes about a gigabyte of disk to record a year's signatures. Not much, when you think about it. But it all hinges on the fact that his algorithm will produce 40 bytes of completely unique signature for any data given to it, never repeating itself. No, I don't know how you prove this.

Wednesday

Up early to catch a flight. It would be easier if it hadn't been for an invite to another Hacker. He wanted to play me some music.

To be precise, he wanted to play all his music. He has it all on hard disk. Including decompression software; something like 3,500 songs ranging from the 50s to the current date, all crammed into 9Gb. All the music in the world, he says, ever recorded, is compressible into 50Gb or so. It isn't unchanged; it's merely "transparent" - taking advantage of the uselessness of human ears.

For example, if there's one loud instrument playing, you don't have to record exactly how loud the others are. You assign an arbitrary level to them, less than 50 per cent of the loud one. Nobody can tell the difference without a recording instrument and display. Again, if there's a loud noise (a drum beat) you can ignore all waves for a few milliseconds after it as the ear can't detect anything else. You do realise that this means you can put your entire collection of CDs onto one compressed DVD? Next year? Using shareware?

The day passes in flight to Chicago, to London. The dawn comes up as the taxi reaches Hammersmith. I suspect I'm not going to get much done today, which is now.

Thursday

An important phone call. "He's out. He'll call you back." I give them the mobile phone number, and lie down to rest my eyes. He doesn't call back. I wake up around 6.30pm when the family returns.

Friday

Playing with Windows CE handhelds. A quick check with David Potter, arch-rival of the CE folks - he's chairman of Psion. Potter confirms my original impression - that the CE design is not just remarkably similar to the Series 3 pocket machine, but is designed after considerable study of the machine. He won't reveal exactly how many Psion 3a's Microsoft bought during the work, but does admit it's more than 100.

We take a Psion 3C and line up the "Q" of its keyboard with the "Q" of the Casio model CE Windows machine. When the "Q" keys touch, so do the "P" keys. A coincidence, of course, based on ergonomics. The Psion is arranged to be exactly too small to type on, and so naturally, when Microsoft looked to design a keyboard too small to type on, they chose the same pitch.

Potter points out that he was invited to join the CE Windows project. "We were part of the [original Microsoft handheld OS] WinPad project," he reminds me. "We actually had one ready to build, if that had gone ahead."

I suppose all we need to wait for is next year's press release from Microsoft, in which they point out that they "invented" pocket computing. You think I'm joking, don't you?