Blog plugin I would like: Similar Old Books

Say I write a post. This one.

The plu­gin (which may suf­fer from CPU and stor­age issues, but screw that) com­piles a trigram-​​based lin­guis­tic vec­tor of the post. That is, it counts how many there are of every 3-​​letter sub­string in the entire text. It also notes the length of the post. Or maybe it cre­ates some other sort of com­pu­ta­tion­ally lin­guis­tic index­ing key, like a Sta­tis­ti­cally Improb­a­bly dohickey, or a dis­crete wavelet alpha­betotron. One of those.

The plu­gin then takes this tri­gram fre­quency vec­tor (or what­ever), and scans the entire cor­pus of Project Guten­berg text files for the most sim­i­lar pas­sages of the same length.

So in the end, you get a link like “Public-​​domain eTexts which may be inter­est­ing,” and it links to the eText, with a same-​​length passage.

Maybe a few pro­cess­ing and stor­age con­sid­er­a­tions. Like I said. But offline pro­cess­ing would work.…

Glitching Things Done: Stop! Following! Me! Wolfram!

So I’m try­ing to coor­di­nate Google Cal­en­dar cou­pled with iCal to man­age appoint­ments and tick­lers for ToDos. So far it seems nice, though I wish I had a MacOS client for Google Calendar.

The other day I was enter­ing the Google cal­en­dar item for the Wol­fram Tech­ni­cal Con­fer­ence, and some­thing hap­pened. I don’t know, maybe I was called away from the com­puter, but as far as I know I clicked to cre­ate the event, and then walked away and pos­si­bly closed my com­puter and broke the link, and when I came back I didn’t see it on the cal­en­dar so I made a new one.

When I synced my iCal sub­scrip­tion, there was a half-​​hour Wol­fram activ­ity, star­ing me in the face from the mid­dle of that day’s sched­ule. “Hunh,” say I. Go back online to Google, no sign of it. Look at the Agenda view, no item with that name on any day. Search? Nope.

I refresh my iCal sub, and there it is. Moved a bit.

I dis­miss it, and move along with life. It’ll scroll off, come Sat­ur­day, and I can just leave it there.

Ah ha ha, no such luck. This morn­ing I real­ized it: It’s always stuck at the time I refresh my sub­scrip­tion. So some­where in the cal­en­dar XML file is this weird par­tial abor­tion of an event, haunt­ing my days. Like a magic To Do that I don’t really want, and can never shake.

Mys­te­ri­ous. Strangely, pos­si­bly use­ful, if it could be repli­cated. But who knows how to do that?

Do they have a To Do item in Google Calendar???

Update: I squished the bug, some­how. I used the Advanced Search to find all men­tions of “Wol­fram”, saw an odd appoint­ment on Octo­ber 1, and deleted it. But for­got to look care­fully at the details of how it had been set up to induce the mag­i­cal Always! Now! behav­ior. Con­sider it a mys­te­ri­ous chal­lenge, since by their admis­sion Google Cal­en­dar is unable to alert you more than a week before an even.…

On Pulse

I’m increas­ingly taken with the por­tions of Robert Frenay’s Pulse: The Com­ing Age of Sys­tems and Machines Inspired by Liv­ing Things. Not least because I’ve run sev­eral engi­neer­ing work­shops and sym­posia on it, and even less because he thumps all the soap­boxes I’ve been stand­ing on for 15 years.

I’m not sure I’ve met Robert, but I doubt it; most of what he’s recounted so far I’ve pieced together on my own. I sus­pect we’ll meet soon enough, though.

Link there. Go read what he has to say.

Buzz this one; it will pay off in reci­procity for you. Because, read­ing me, I know you should read him.