Bloglet, the gentleman's mock turtle soup --
Moss made it sweeter than myrrh ash and dhoup

I had a lovely New Year's. Just lovely. We'd gotten tickets for the Good Olde-Fashioned Future party several days in advance, so we spent the early part of the day walking in Fort Tryon park, K. taking pictures, me mumbling on my jew's harp and balancing on steep things. Then we had salmon and duck at a restaurant up the hill and ambled homeward for a shower and a spiffing up (pictures forthcoming). We brewed some extra sludgy coffee and set off for Long Island City at approximately 10:15 pm. We arrived there, after many subway adventures (including meeting a guy wearing a hat decorated with cutlery who wanted to take pictures of us and glimpsing a dude on another train who was the very spit of Neal Stephenson), at 11:58 pm, just in time to gather at the loading dock of a huge loft building alongside several hundred people dressed as chimney sweeps, electrified fish, Victorian gentlemen, mad prophets in tinfoil hats with neon head filaments, etc., and kiss in the New Year. Then we were informed that the party had just been shut down by the NYPD for lack of liquor license and violation of fire code and that we all had to go home.

So we got back on the subway and had many dramatic and foot-cramping adventures overland and underground and finally arrived home at around 2:30 am, whereupon K. took off her garter belt and made us each a whiskey, and we settled into bed with the Scrabble board in perfect happiness. _
respond? (4)
03:36:29 PM, Tuesday 1 January 2008

When I have spawn (gonna be a couple years there, Mom!) they're totally getting these. _
respond? (1)
07:07:39 PM, Sunday 30 December 2007

My favorite line from The Ghost Map, which my brother William gave me for Christmas:

"Attention then turned to the East London Water Company. Initially, company representatives swore that all their water had been run through state-of-the-art filter beds at their new covered reservoirs. But reports had surfaced of some customers discovering live eels in their drinking water, which suggested that the filters were not perhaps working optimally."

It's a fascinating story, though the book itself suffers a bit from excessive repetition. This is a kickass website about John Snow, the central figure in the case, and this (PDF) is an article about various methods by which Snow's data has been mapped over the years. Great stuff. _
05:34:23 PM, Sunday 30 December 2007

I just dreamed that I pet a lynx on a bus and it purred at me. Do lynxes purr? Anyway, it was kickass. _
respond? (5)
12:55:08 PM, Sunday 23 December 2007

Hell yes. Where skill and dexterity are wanting, patience and ingenuity are shoring 'em up. Also damn fun to watch. _
03:27:31 AM, Sunday 23 December 2007

My brother Robert is here, and this is freaking awesome. That is all. _
respond? (1)
02:45:18 AM, Sunday 23 December 2007

"Tua Bethlem Dref", current Christmas earworm. _
09:06:31 PM, Saturday 22 December 2007

"You make me gag banana!"

-- My mother. _
respond? (1)
12:42:29 PM, Friday 21 December 2007

Greatest Classical CD Covers, parts one, two, and three. Dude. Just... dude. Via Alex Ross. _
11:50:01 AM, Tuesday 18 December 2007

At long last, my essay on open source steno CAT software. I've had trouble composing this one, because my thoughts on it have been diverging along two paths: first, what sorts of features I want in order to do my job (and which my $4,000 software isn't at all designed to give me) and second, what sorts of features would be desirable in a program written for people who aren't professional realtime transcribers but who want to use steno to increase their computing efficiency, smooth out the process of converting thought to text, or communicate in text at speed.

I think I've necessarily got to start with the former, since if I'm going to invest a serious amount of time designing and then learning how to program the damn thing I'd better be able to use it to make a buck. Also, as I hope to prove, some of the elements that I can't live without would be superfluous in non-CART applications but might just be cool enough to rope in the geek crowd by sheer force of slick cache alone. That's the idea, anyway.

I'll start with Neal Stephenson's Cryptonomicon. If you ain't read it, get a copy. It's great. This essay contains only very minor spoilers, so unless you're hypersensitive about such things, you're probably safe to keep reading. If you have read it, I reckon you'll agree with me that the coolest scene in the whole damn book is when the main character has to decrypt some important information on his computer without letting it ever appear on his screen, because he knows the bad guys are watching everything displayed on his monitor, although they're not observing his actual keystrokes. He solves the problem in the most ingenious manner possible, and there's the threat of bloodshed and the promise of untold riches and all the stuff pulp fiction thrives on, but while my own daily battles with Tax Law and Urban Planning aren't waged for such high stakes, they require the same sorts of subterfuge, and my software simply isn't designed to handle it.

Frequently while transcribing I'll realize that I've misstroked a word -- which has translated as another word entirely, subverting spellcheck -- and will need to fix it manually when editing the transcript later. Currently I've got a dreadful kludge to mark the spot that's too tedious to describe here, but it's clumsy, awkward, and requires too much fiddling after the fact to be tenable. Sometimes I won't have misstroked anything, but will simply have thought of a new stroke definition for my dictionary, or a task for my to-do list. Even in a continuous lecture there are frequently pauses of several seconds during which I could throw in any number of definitions, corrections, or notes to myself, except that I can't, because:

1) All steno software (I've used DigitalCAT, Case CATalyst, OpenWrite, and Total Eclipse; if one of the other brands is an exception, I'd be happy to hear about it) is really court reporting software, and therefore designed to preserve the record at all costs, which means that all strokes which aren't mapped to direct commands will appear in the master transcript. You can't even use the keyboard emulator to type into another program's text field without it being echoed in the CAT program's main window. This drives me NUTS.

2) Virtually all strokes which are mapped to direct commands bring up dialogue boxes. Get this: I have to have two separate profiles for each student because my kludge (which basically boils down to "you might wanna filter through the last several sentences 'cause there might be something misstroked back there, unless it's a false alarm, which it well could be, 'cause this mark is indistinguishable from the software's guesswork indicator.") requires that I set the dictionary definition box's transparency to 90% (the maximum), squish it up so that it's illegible, and slap it in the corner of the screen so it won't distract the student when it pops up every time I need to mark a section. The other profile is for editing, where the dialogue box is normal-sized and legible; but couldn't I just skip the thing altogether? Or at the very least get just a simple, discreet status bar confirmation? Vim has spoiled me. It makes everything else feel as frustrating as Vigor.

So I need a way to type text without it showing up in the main document (and to switch in and out of visible/invisible modes with a single keystroke), to be retrieved and processed afterwards, and I need to be able to invoke commands without bringing up a lot of dross that obscures the screen, distracts my clients, and annoys me all to pieces.

What's more, I need to be able to invoke commands without the two-second lag that Eclipse uses to determine which strokes connect to which. This was one of the main reasons I gave Vim up as a loss for steno purposes. Eclipse keeps a two second buffer going (the time can be lengthened or shortened, but the shorter it gets, the more likely words will translate improperly; "catalogue" will translate "cat log", because Eclipse will have forced a disconnect between the first stroke and the second), and unless the buffer is manually flushed (which I've developed into a fast-twitch reflex, though it's a lot of strokes wasted), there will be a two second lag between the last stroke typed and the time it's released to be displayed on the screen -- or, in the case of Vim, employed as a command keystroke. I'm not sure why they've done this, or if a solution can be found (I suspect it can), but without one, there's no telling a computer what to do unless you've got the patience of a Fuke Zen sea slug.

This ties into one more area where CAT software currently falls flat: it can't control the rest of the computer. I'm able to record a macro that supposedly "presses" Alt-Tab, but does it then switch tasks? It does not. It does absolutely bugger-all.

Some other necessities, thrown out willy-nilly:

* Live rtf/cre management, no proprietary dictionary format. The ability to edit batchwise (regex, woo!) from the steno keyboard during realtime.

* The ability to quarantine the last untranslate; not simply correct it on screen but save the steno in a dedicated file for later inspection and (if necessary) systemic correction. Too often I'm tempted to leave the untranslate on the screen so I won't forget about it later, but that would be a disservice to my clients.

* The ability to toggle the immediately previous "does" to "is" or "can" to "can't", after contextual revision of ambiguous phonemes.

* Automatic text wrapping at every level of zoom. It's pathetic that none of the software I've tried is able to do this; I make the text bigger, and it starts scrolling offscreen. What century are we in?!

* The ability to reconstruct text documents from raw steno files -- including commands. This seems as if it should be trivial, but I've never seen it done properly.

* Some sort of plugin capability. I'd still love to be able to work seamlessly with Vim; it would save a lot of wheel reinvention. I'd love to be able to load in Bozzy with a keystroke, define a couple hundred words, and then send them off to be incorporated into my main dictionary, without having to go through a five-minute manual conversion process each time. I'd love to see my new briefs and problem strokes of the day automatically exported to a Typestriker clone so I can practice flexing my muscle memory before every job.

* Seamless modal switching. Command mode, insert mode, invisible mode, visual mode (death to mouses!), spelling mode, raw phonetic syllabic mode.

I've got a lot more notes on the subject, but that's my first attempt to lay out the major dealbreakers. There's also an unbelievable amount of bloat in most steno programs, useful only to court reporters and wholly at odds with the needs of most computer users. I would argue that CART -- since it's entirely screen-based and doesn't aim to produce a printed transcript -- is much more closely aligned with the geek ideals of power, discretion (or whatever word denotes "lack of handholding"), and efficiency. If I was able to hack together my own perfect CART application, I figure it wouldn't look so different from a generic steno-ready text editor, streamlined and flexible enough to be put to all sorts of bold new uses. _
respond? (6)
04:01:55 AM, Tuesday 18 December 2007

I am in Montana. It is, as always, a heady mix of bewildering and delightful. I miss my girl, though. Stupid finals week. _
respond? (2)
10:55:42 PM, Saturday 15 December 2007

My dear friend Sola and her husband have been making tea soaps. I got mine a couple of days ago, and they're so lovely I'm going to give them as Christmas presents (though I might order another set to keep). If you guys are into odiferous delights, you might wanna try 'em out. After leaving a dreadful sucky day job, Sola (who also binds incredible handcrafted books, a few of which I have the privilege of owning) is making a go at the artisan's life, and I admire her greatly for it. My favorite is the pear. _
08:58:19 AM, Saturday 15 December 2007

It's an odd feeling: often after a day of transcribing classes and conversing with my deaf students, I expect myself not to be understood when I speak. Three of my four students are Oral Deaf; they understand a bit of both ASL and ESE, but spoken English is their native language, so they speak to me and I answer them on my steno machine. My fourth student is an ASL-native and doesn't speak; we converse by writing back and forth on our respective computers. But since most of my day is taken up with writing both what I hear and what I wish to say, sometimes even after I've packed up all my gear and a hearing student or teacher asks me a question, I'm reflexively tempted to gesture, "Hold on, let me get my machine out, and I'll answer you," instead of just speaking in reply. _
11:03:44 AM, Monday 10 December 2007

"O fat liver full of oil, let us go and watch the moonrise!" _
10:58:29 AM, Monday 10 December 2007

Wow, I managed to break two pairs of earphones today. Go me. _
05:19:41 PM, Thursday 6 December 2007

Living in New York is so #*&$@ awesome. In January we're gonna see Quasthoff, Bostridge, and two singers I haven't heard of do Schubert Lieder at Carnegie Hall, and in February, the Magnetic Fields at Town Hall! Yeee! _
respond? (2)
11:30:06 PM, Monday 3 December 2007

I'm still mulling over what I want in free steno CAT software, and I'm rooting around to see if the Scratch Board concept or something similarly cheap and programmable might be feasible for a home-brew (non-mobile) steno machine, but I've decided what I want my ideal portable Steno Machine of the Future to look like:

Fingerless gloves. They come a little past the first knuckle, but not up to the second or third, and they're extremely light and stretchy. Ideally, they don't impair wrist movement at all, and have very little effect on finger movement. But hidden in the palm fabric is an extremely durable fan-folded array of flexible touch-sensitive panels. In their folded state, they allow for a clenched fist, a handshake, and all manner of ordinary manipulations without sustaining damage or preventing free moment. But if the wearer depresses a catch at the heel of the hand and then splays the fingers outward, the panels unfold out of the glove's palm pocket and each one takes its place under the corresponding finger. As a finger presses one of two areas on each panel, it registers the keypress, and the wearer experiences enough give in the panel to realize that the key has responded, but not so much that they're all flapping about obnoxiously. The mechanism should respond to conductivity, not merely motion, so that wind isn't a problem, but above all, it should be possible to stow it quickly and comfortably back into the palm with a single gesture when it's no longer needed. _
respond? (8)
01:34:40 PM, Monday 3 December 2007

This is the watch my brother gave me for my college graduation. I love it. Photo by K. _
respond? (2)
08:41:58 PM, Saturday 1 December 2007

The Tick vs. Arthur... muscle highlights?! _
06:46:49 PM, Friday 30 November 2007

Stenogeekery Revisited

I'm frustrated at the staggering lack of overlap between the steno and geek communities. The only real exception I can think of is Gary Robson. It's baffling; steno is an intensely geek-friendly technology. But the majority of stenographers I know are technophobic, or at least techno-incurious, and the majority of geeks I know have little interest in learning stenography. The former problem is not mine to solve, and I'm happy enough to let my colleagues keep their blinders if they want them. But geeks are my people, and there are good reasons why they should trade in their non-chording text entry systems for stenographic ones. There are two good reasons why they haven't so far: the learning curve and the cost -- but as soon as the latter is fixed, it'll obviate the former. If things go as they should, steno will become ubiquitous among the technological elite. That'll bung me out of a job, of course, but by then I'm counting on having amassed so much knowledge from the various courses I'll have CARTed that I'll be on to something new and lucrative before the cock crows.

Right now it only makes sense to invest in steno equipment, software, and training if you intend to make a career out of it. Even a used student machine costs around $1,000, and the software ranges between $2,000 and $4,000 unless you're in a certified steno school and qualify for student rates. It's just not something you can pick up for a lark and play around with. This is bad.

Before geeks will take to it, three things need to happen:

1. The hardware has to come down radically in price -- to the range of $200 or less. This is happening to some extent as the high attrition seen in steno schools (85% or more) results in students selling their learner models (which, for the first time, are mostly computer-compatible rather than manual) on eBay for peanuts. But even they aren't cheap enough for dabbling, and aren't likely to become so anytime soon. I'm not sure how to solve this one, but some sort of virtual/touch keyboard emulator would be a godsend. Or maybe a cheapo kit to assemble at home? If I had seed money, this is where I'd put it.

2. Someone needs to write free, multiple-platform CAT software and recruit a community to help develop it. More on this in a subsequent post.

3. Realtime text communication needs to become not just useful but necessary in daily life.

It would also help enormously if the whole package could be made to look cool and futuristic instead of clunky and clerical, but I'm not holding my breath.

Efficiency is elegance. The potential afforded by a chording system linked to a customizable lexicon is enormous, and steno doesn't allow merely for great speed of composition and revision; it expends vastly fewer ergs (and racks up fewer RSI points) in the process. I know autocomplete and abbreviation tricks save coders from actually typing out their boilerplate from scratch all the time, but even if you map every word to a two-letter-plus-space-bar expansion command, you're still doing three times the work of a single steno chord, with fewer mnemonic hooks to give you purchase.

Actually learning to plonk things out on the keyboard is not really that much harder than qwerty training, but there's admittedly a significant commitment to make in terms of compiling and customizing a personal dictionary. Still, there's an awful lot of people these days -- programmers, authors, bloggers, students, gossips, gamers, suits -- who spend most of their waking life inputting text, and they're wasting untold time and tendons going letter by letter.

Once a discreet, vision-permeable head-mounted display is developed, the advantages of text-based communication will blossom like a hedgepig. Text is permanent, pliable, less obtrusive and more overlappable than speech. We'll be swimming in it. We'll need to make typing as easy as speaking, or we're hobbled at the gate. Speech recognition, pace Star Trek, won't do the job. But steno would, magnificently, if only we can convince the geeks to start playing with it and turning it into something nimbler than it's been in the hands of paid single-minded transcribers.

Okay, next post: what the software needs to look like. Go ahead and argue with my premise in this one, though. It all seems blazingly obvious to me, which is always a bad sign. But why the hell shouldn't anyone who gives a damn about producing text learn steno? _
respond? (30)
06:34:08 PM, Wednesday 28 November 2007

Mirabai Knight

Older Entries
Complete Archives by Date
Search 'em.
Which Trouser Role Are You?
Blog Tracker of the St. John's College Blogmass
bloglet script by Moss Collum