Postf. 2060, 89010 Ulm, Germany

Tel. ++49 +731 921-6931

Fax: (Goppold:) +731 501-999

http://www.noologie.de/symbol07.htm (URL)

How the Software Industry does its best to undo the productivity gains of computers...

The issue of keyboard layout and ergonomics will be enlarged a little more. Some people will remember the discussions around the virtues and drawbacks of cursor control schemes of an earlier era of computing before the mouse became popular. Notably schemes like the Wordstar diamond versus emacs and vi types.

The Wordstar diamond appeared on CP/M machines about 1978 and was still usable on the early IBM PC computers up to the advent of the AT MF-2 keyboard. This layout had the following codings:

cntrl-s := back,

cntrl-d := forward,

cntrl-e := up,

cntrl-x := down,

cntrl-a := tab left,

cntrl-f := tab right,

cntrl-r := page-up,

cntrl-c := page-dn.

For daring power-users there were even the control-q combinations which further amplified these keys. (See ILL:WD-1 , The Wordstar Diamond)

This key layout allowed the fortunate touch-typist to move the cursor while writing without interrupting the writing process. The positioning of the control key in immediate proximity of the cursor control keys made the coordinated touch of two keys at once extremely easy and allowed very fast cursor movement in a text. The sensible issue is that often used keys must be positioned in close spatial proximity to allow easy touch-typing. The same argument was also made in the discussions on alternative layout schemes than the qwerty model which never made it into common use, since the conventional scheme was too entrenched. Of course the Wordstar diamond was criticized because for people who are not touch-typist it makes less sense, and it takes just a little mental effort to memorize (and maybe one hour of practice).

Other cursor control schemes used another method: cntrl-b for back, cntrl-f for forward, cntrl-u for up, cntrl-d for down. This has the advantage of being somewhat easy to remember for english speaking people (and pure gibberish for all others), and for people who are not touch typists it may be an easier method. By positioning the cursor keys evenly distributed across the keyboard, the immediate feeling for control could never be achieved. This scheme and others were more used in the Unix world. It is probably safe to say that most Unix programmers are not touch typists.

As opposed to the QWERTY scheme, here it was not a bad scheme that stayed because it was entrenched. To the contrary: A fairly useful scheme was displaced by a vastly inferior solution. The MF-2 layout obliterated this type of cursor control because of the extremely uncomfortable new position of the control key which made this key much harder to use. It positioned this key at the lower left and right ends of the keyboard, way out of reach from the other, often used keys of the main field. The user has to bend the small finger of the left hand to a very uncomfortable angle to reach the key in its new position, destroying the close association needed for secure simultaneous operation of both keys. Instead, the caps-lock key was moved up. The question is how often does one use the caps lock key? In all my own typing practice of about 15 years, and about 12 Megabytes of text, I came to an average of maybe once a month. How often does one use the control key? In the days of the Wordstar diamond, when all cursor control was done with control keys, this key was the most important and most often used key of them all.

The decision to place the rarely used caps lock key in the central row of the keyboard adds another problem because this key is now quite often hit accidentally. And, to add insult to injury, we cannot toggle this key. That is, we must hit another key, the shift key, to undo the action of the caps lock key. I don't think that all the devils in hell are able to conjure up a more horrible ergonomic nightmare than the MF-2 shift lock key [30].

Here, a bad design decision wiped out a whole sector of ergonomy. The use of a separate cursor control key pad does not offset this disadvantage because now one has to move the writing hand away from the main part of the keyboard to reach the cursor keys. Bad news for touch-typists. And, more problematic, one has to move the eyes off the screen, interrupting the flow of work. For someone trying to think with the text while s/he is writing, this is desaster. Ten Million flies can't be wrong. Somehow the new scheme got adopted without so much of a question. It is nowadays impossible to find a vendor for the old type of keyboard and even the non-PC workstation vendors have adopted it. Since keyboard drivers are usually not user programmable, there is no easy solution for the problem save re-programming the keyboard with the soldering iron.

Now perhaps the MF-2 scheme was not a bad or sloppy, unergonomic design at all, but one that was carefully planned. It is not possible to find out who designed this layout - but if someone should have wanted to get some competition off the market that was using this scheme, they succeeded nicely:

The Wordstar diamond was used by this once most important CP/M text editor, which has now almost disappeared from the market, and a few followers. Most notable of those are the Borland Turbo tools. As it turned out, Borland also lost the market and guess who won? We don't need to delve into further speculations on this subject but it possibly serves to illustrate the point that not everything in the olden days was bad, and not everthing new is therefore better.

And we can be sure we will be in for "more of the same kind" of ergonomically strangling technologies. This is simply because some large players in the game want to keep a productivity advantage by not supplying the best possible tools to the large public.

We can refer here to the much discussed theme of the hidden secrets of Windows. This is just the small beginning of a whole industry of hidden trapdoors and intellectual mazes in software that makes the dungeons and labyrinths of the castles of an earlier epoch seem like child's play when compared to modern possibilities. We always can turn to "Neuromancer" to get an indication where the development is heading, and who will be the one to profit from it.

This brave new industry will be capable of giving us sensory and computing implants in a few years to come. At the Ed-Media 95 conference, we heard an invited talk about "the implantable workstation" (Gerald Q Maguire, ED-MEDIA95).

But should we trust this industry at all to let it come closer than a five-foot distance to our bodies and sensorium? I believe we should be extremely wary. We may have already let it come closer than is salutary.

How much blood, sweat, and tears would millions of computer users world wide have avoided if the innumerable "closed shop" binary configuration files of our modern software systems were in plain, readable ASCII? My guess is that the time and data loss caused by those unreadable or otherwise unchangeable, binary configuration files is in the billions of dollars world wide. Just the nuisance that configurations are only accessible through the circuitous manufacturer's "same procedure as everyday" mouseclick orgy, wading through mazes of menus, to change the parameters, adds to millions of man-hours lost.

A horror tale of personal experience tells the truth of life. Who would ever think of those unobstrusive Windows .grp configuration files? Probably 90 % of all Windows users don't even know about this kind of software beast that is controlling a vital part of their daily working environment.

One of these fine days, I had changed my system by putting in a EIDE harddisk to add to my SCSI harddisk. That is innocuous, would you think. Not so, my friend. Because the EIDE cannot, by any way, made disk D:, since the BIOS insists that it must be disk C:. Now comes the surprise: I had the Windows system on the SCSI disk, which was C: before, now residing on D:. Consequently all the .grp entries were invalid [31]. What should I do? Obviously the Windows designers hadn't intended that situation to appear, and there is no way to edit the .grp files (that I know of. I looked long enough in the manual, believe me). Should I copy the Windows to the new hard disk C:? That would have been the best thing to do, but I didn't want to. Re-install the whole system? I thought in my naivety that I might outsmart friend Billyboy by getting out my trusted bit-editor and patch in D: wherever it said C: in the binary .grp files. You can imaginge my surprise when the system came back at me with the message: "corrupted .grp file" and gave up. Checksum test, of course. Billyboy had shown me again what he thinks he can do with people who try to thread on territory that his software designers had decided "closed shop". And he gets away with it. Every day a million times.

I tell you how I managed anyhow. The procedure is called in German "Über den Rücken durch die Brust ins Auge". I don't know how to translate that. It just means verrrry circuitous. I have the Central Point PC Tools system. Its desktop system, which can substitute the Microsoft desktop has the nice menu entry for "import .grp files". Of course the designers hadn't envisioned the possibility that anyone would want to ever quit their fine system, and an export function for .grp files is therefore missing. Like so much of everything of these nice gifts of the computer industry is one way only [32]. Once you are hooked on the company product, you are done in, for good. Fortunately, CP had not been as tightly security-minded as Billyboy's men: The CP-.grp importer ate my binary edited .grp files with the D: patch without so much as a hickup. And converted them into the CP desktop data base. That is, you have guessed by now, just another one of those ominous, closed shop, binary only data base files that are controlling the whole central point Desktop system. Opposed to Microsoft, all the information resides in one big file, not many small .grp files. Now I had it in there, cleanly, but for some reason, CP desktop isn't entirely up to my liking. So what did I do? I created new Windows .grp files with the menu, but dragged the converted entries of the CP desktop into the new .grp files. That cost me in total probably as much as re-installing the whole system, all in all about half a day's work. Just so much money wasted because some lazy or over zealous systems designer had decided that the user is not supposed to change the .grp files with an editor.

You may get to know the details of these things when you subscribe to the "Microsoft Systems Journal" or get into the special mailboxes, or have a friend at the manufacturer. But if you don't you are stuck. And since any self-respecting software today lavishly produces control files in binary you can never be in all the mailboxes, subscribe all the trade journals, or have your friends at all the manufacturers, unless, of course, your name is Jerry Pournelle, and you work for Byte magazine. But not for the rest of us.

This kind of proprietary behavior (in Deutsch: Nach Gutsherrenart) clearly shows us that the software industry has already taken forms that would be called expropriation and road-robbery if it happened in another sector of society. Somehow, the user community thinks that this cannot be any better and puts up with it quietly. The cumulated losses are in the billions of dollars when we add them up. The damage doesn't fill the coffers of anyone profiting positively. It is just a general loss due to close-minded proprietary thinking and sloppyness, sometimes deliberately created to prevent people from creating their own interfaces to hook in subsidiary products that might infringe on some part of the market. The mafia extorts maybe at most a tenth of this sum.

The industry will never change that behavior unless it is forced to do so by legislature. In the earlier years of industrialization, the industry continued to produce woodworking machinery that took off fingers and hands of workers at liberty. Steam engines that exploded, scalding and cooking workers in the sweat shops. Transmission belts for machinery that dislodged and cut people in half, just like cheese. It took about 100 years of worker maiming before legislation was finally there to make machinery a little more safe. The factory owners themselves wouldn't have changed a bit.

Software nuisance is a problem that doesn't take people's lives, nor does it maim them. But it takes their lifetime, a little minute here, sometimes a few hours there, sometimes a week. The professional ethos of the software engineer and the systems administrator is to put up with this, otherwise, if things were so easy, he might lose his job, because users might be smart enough to edit configuration files themselves. That fear is unjustified, with systems getting more complicated by about an order of magnitude every 3 years.

How much more usable would the Next Interface Builder (NIB) have been if it wouldn't create a machine-only binary file but an ASCII readable file constructed with the Leibniz TLSI principle? (See below)

You could use a pattern of widgets that you had developed once and add other pieces with the editor, or exchange them, without going through the mouse menus every time. Using the NIB once or even ten times is nice. After you have done it ten or more times, you would prefer to be able to mechanize the process. Therefore, what was a boon in the beginning, becomes a millstone around your neck after ten times repetition.

[30]This is possibly the case only with the german encoded keyboard.

[31]Of course, all the .ini files also, but these are fortunately text files, so you can edit them with a normal text editor or you write a shell procedure to catch them all at once if you are a real smart hack. Of course the whole thing would be entirely unnecessary if you had Unix, because there you don't worry about C: or D:. Also, someone told me that the last versions of DOS may allow a re-assigning of drive names. I haven't checked into all the possibilities.

[32]This kind of tactics made the books when Rockefeller (or was it someone else) around the beginning of this century, traded the Chinese his brand new oil lamps for their old smelly lanterns. The awakening on the side of the poor Chinese was rude when they found out that the new lamps would only burn Rockefeller oil. This was, of course ten times more expensive than the old oil. And the old lamps were gone, traded in. The computer industry has infinitely refined and magnified on that time honored trick.