Cloud Chamber 45
September 1993

A Diagnosis: '... he developed in his later years a telltale interest in science fiction, which is usually a reliable sign of imaginative bankruptcy.' (John Wain on C.S. Lewis, 1964)

Polemic in Progress

The coming CD-ROM edition of the Encyclopaedia of SF has been a source of excitement for some while. It actually impelled me to buy a cheap CD-ROM drive for my computer. Paul Barnett (Technical Editor) did likewise. Co-editor John Clute arranged to buy a new computer with CD facilities the moment the Fantasy Encyclopaedia advance came. Ahead stretched a bright vista of instant access to the book ... frequent updates ... in-depth searching and research....

Then, from the CD publishers Nimbus Information Systems, along came the demonstration software (covering entries from A to AM only, and thus small enough to go on an ordinary floppy disk). It rapidly became evident that Nimbus had never before taken on a real, major encyclopaedia with its necessary density of cross-reference, and that – putting it rather tactfully – their standard software for searching and displaying CD material simply wasn't up to it.

'Be our plenipotentiary,' Paul and John conveyed to me. 'You know this technical stuff. Get them to fix it.'

So I drafted a long and tactful letter pointing out the major deficiencies in the demonstration SF Encyclopaedia, such as:

(1) All the menus from which you choose things to search for – headwords, authors, titles – are limited to a width of 24 characters. This makes it a bundle of fun when (for example) you're looking for a specific, numbered anthology and are confronted with a long list of titles all reading YEAR'S BEST SCIENCE FICT.

(2) 'Searching' for a title or word does not, in Nimbus software terms, mean finding it. It means dumping you at the beginning of the entry (or the first of a succession of entries) in which it is found. For example, searching for Report on Probability A takes you to the ALDISS entry all right. After that you're on your own: you have to scroll through its lengthy text using the IBM's down-arrow and PgDn keys, and find the title by eye. But, as Nimbus point out, at least it is strikingly highlighted when at last you reach it!

(3) Report on Probability A is not strikingly or otherwise highlighted in the ALDISS entry. The title – like all too many longer titles – breaks across a line end, and Nimbus's software is too stupid to recognize and highlight a title which does this perverse (even if entirely predictable) thing. So the eye skids past....

(4) Another way in which titles can be made hard to find comes into play when they take the rare and unusual form of beginning with 'The'. All such titles appear alphabetically in the search menus as (for example) ALTERATION, THE ... but in the text as The Alteration. Because (I think) of this difference, titles beginning with 'The' are not detected and highlighted.

(5) Every hyphenated phrase which originally fell over a line-break has acquired an added space: for example, The Two- Timers by Bob Shaw. When such phrases happen to be book or movie titles, the result is that they too are not highlighted when searched for. (As above, the software takes you to the start of each entry containing the chosen title, and leaves you to locate it by eye without benefit of highlighting.)

(6) When in a long entry you do finally find a title you're researching, that isn't the end of it: it might appear again. Surely, I suggested to Nimbus, there needs to be a means not only of getting automatically to the first mention of a title (etc) in an entry but also of locating all subsequent ones with successive key-strokes, right down to the end? H'mm, they said.

(7) To make browsing in the Encyclopaedia more of a challenge, the headword (e.g. ALDISS, BRIAN) does not stay visible on the screen. It scrolls out of sight once you start moving through the entry text. When you've located a title in one of the less self-identifying entries (theme articles, for example) it is all too easy to get lost in the text and forget which entry you're in. This is doubly irritating when whole lines of the text screen are permanently devoted to reminding you that you are still, in fact, no kidding, reading The Encyclopedia of SF on CD-ROM.

(8) Well, if lost in the text surely one can always hit the IBM Home key and zoom straight back to the start of the entry where the headword can be read? This is a more or less expected function, just as the End key might be expected to take you to the end of the entry (there to find 'Other works by', who wrote the entry, etc). Sorry: in the Nimbus software, neither of these keys does anything at all. To navigate a long entry you just have to keep remorselessly hitting PgUp or PgDn.

(9) Cross-references are a major feature of any electronic reference work. Ideally one should be able to move the screen cursor to a marked cross-reference, press Return, and be whizzed straight to the desired entry. Sorry, Nimbus don't do it that way: if you press the Link key (they call cross-references 'hyperlinks' in order to sound more trendy) the software slowly, remarkably slowly, compiles a list of all the cross-references appearing anywhere in the entry you're reading, and displays it on screen as a menu from which you choose the cross-referenced entry....

(10) Well, perhaps choosing from an alphabetical list is still a plausible way of doing it. But I forgot to mention that the list isn't alphabetical: in the demo software, cross-references appear in the arbitrary order of their occurrence in the entry. The ALDISS entry has 84 cross-references by my count, but (when I complained) Nimbus couldn't see why anyone should have any difficulty in instantly picking the desired item from such a random-looking list. I forgot to say that only 10 such items are visible at any time in the cross-references 'window'.

(11) Nevertheless, browsing through cross-references is one of the most enjoyable and serendipitous things to do with an encyclopaedia, following a trail from ALDISS to ALIENS to wherever, as the fancy strikes you.... No. Nimbus offer only 'one level of cross-reference access'. This means that having followed a cross-reference from ALDISS to ALIENS you can go no further: the only way to follow up a second reference found in ALIENS is to 'back out' by repeatedly pressing Esc to return via ALDISS to the main search menu ... and then search all over again to get to ALIENS or to the cross-reference noticed therein.

(12) The heart of darkness of the cross-reference system is, however, that the references were generated by software with minimal human intervention. (In this way, a passing mention of Homo erectus in one entry has been marked up as a book title because it's in italics.) A reference to Martin Cruz SMITH is mechanically marked up by extracting the capitalized bit, and becomes a reference not to SMITH, MARTIN CRUZ – which would have meant work – but just to SMITH. This means that following a cross-reference to this author deposits you at the beginning of a list of 20-odd SMITH entries, and you have to keep pressing the Next key again and again until you've worked alphabetically through to the right one. I first realized this when I followed a reference to Martin AMIS in the entry AMIS, KINGSLEY and was taken, quick as a flash, to the entry for AMIS, KINGSLEY. Which seems, in John Clute's eloquent phrase, sort of dumb.

(13) A bundle of minor points. (a) Nimbus aren't too good on punctuation; they like to put spaces just before colons, not in the Encyclopaedia text but in their own software messages. (b) Likewise they like to indicate cross-references with '( Hyperlink to: XXXX)', which looks ugly enough but never more so than when a line break comes at the space just after the left parenthesis. (c) Various messily short lines also ensue from the software's failure to realize that text can be broken at a stroke (/) or hyphen. (d) They left out all the indentations of paragraphs' first lines. (e) The main search menu has further ugly truncations, with subcategories like 'Titles of works (by wor' (should be 'by word' but the menu isn't wide enough).

So much for the basic list of problems. After hot debate, Nimbus agreed fairly readily to do something about points 13d and 13e, and to sort out as many occurrences of 5, 13b and 13c as they could by tinkering with the marked-up text files (or better still, persuading Orbit to pay Paul Barnett to do this while plugging in updates). After a rearguard action in which their chap kept defending non-alphabetical lists as quite sensible really, they backed down and agreed with a certain reluctance to fix 10.

That leaves points 1-4, 6-9,11-12 and 13a Most of these have one important quality in common: they can't be cured by mucking around with the text files marked up for the software, but only by correcting the software itself. (Exceptions are 4 and 12. The first could be patched up by alphabetizing titles as THE ALTERATION and not ALTERATION, THE in the search menus ... but it's a crummy solution. 12 requires a vast amount of work spread over the entire text, and is therefore not to be contemplated, at least by Nimbus: all the 'link to' links marked as e.g. SMITH need to be made specific – SMITH, E.E., SMITH, MARTIN CRUZ etc, across more than 4,300 entries.)

When it came to making any correction to the software, even the most trivial, discussion stopped dead. Nothing could be altered, because 'the functionality of SimpleFace [Nimbus's name for their software] is fixed'. Explaining in more detail why the extremely minor point 8 (I would expect to be able to do the equivalent work on a piece of my own software in about ten minutes) was a no-no, Nimbus made it crystal clear: 'Keystrokes. These are parts of the SimpleFace software, so are not subject to change.'

John and Paul showed some tendency to frown over the Nimbus jargon, and I translated into publishing terms for them: it's a bit like a publisher saying, 'Sorry, your book is being printed in dyslexic broken English because the functionality of our mentally defective typesetting staff is not subject to change.'

Nimbus also made great play with the fact that people like Clute and Langford were evidently elitist professional researchers who were demanding luxury features. Only such ivory-tower swine could possibly wish to follow chains of cross-references, or expect a computer search actually to take them to the location of the title that was being searched for. Ordinary readers, the silent majority, would (it appeared) read the CD-ROM book laboriously but happily from end to end, their lips slowly moving all the while.

(And what, you ask, was the reaction of Little, Brown, whose Orbit imprint published the printed version? Their Colin Murray, another non-technocrat, said he believed every anguished word John Clute told him and sympathized all the way to the bank ... but unfortunately LB's lease of rights to Nimbus had given the latter total control over the CD-ROM software.)

Exactly why nothing can be changed remains mysterious. Was 'SimpleFace' written by a contractor who has since buggered off, leaving Nimbus up the creek? In cheerier moments they speak largely of their much more wonderful CD-ROM reader software which will come into being next year, or maybe the year after, and be dead good. Meanwhile they have scheduled the Encyclopaedia for October 1993. Jam tomorrow.

At this point I got angry and decided to prove a point. Most of you will know that the awesomely penniless software company Ansible Information consists of silver-tongued salesman Christopher Priest and technical dabbler D. Langford. One of its specialities is 'stealth' software that lurks invisibly in the background and makes other programs run better ('adds functionality', as Nimbus's chap would put it). I duly went into a debilitating four-day binge of programming and emerged with a tiny utility program that would run the SF Encyclopaedia as a kind of autopilot.

Specifically, this thing sorts out the following points on my list of shame: 2, 3, 4, 5, 6, 7, 8 and 13a. It makes a hell of a difference to be able to look through the titles list, press a magic function key, and watch the program not merely go to the first entry that mentions the title but scroll down through its text to the first appearance of the highlighted title (and all cases where the Nimbus software failed to highlight it have been cured). Press again for the next occurrence of the searched-for text in the entry, until you reach the end. Repeat ad lib in any further entries. I gloated something rotten over all this.

Then, with the enthusiastic endorsement of Messrs Barnett and Clute, I wrote to Nimbus with a demonstration copy of my quick fix, together with easy-to-understand instructions. 'Ansible Information proposes that since [it] remedies several of the most serious criticisms of the SF Encyclopaedia demo software made by the Encyclopaedia's own editors, it could usefully be incorporated into this particular Nimbus CD-ROM at a small royalty to be agreed....' [7 September 1993]

I can hardly wait for the reply. 'The trouble is,' said Chris Priest, 'you're telling them their own child is defective. That never goes down well....' We shall see.

Mailing comments. I fear there won't be many. It is a deeply knackered man who sits writing this:

Verbal tics. I'm sure I have my own. I hope they aren't too irritating. But the subject of today's tiny grump is that perennial favourite, the Remote Quotation. A typical example: we all know that 'Stone walls do not a prison make / Nor iron bars a cage', but I have had it up to here with sf reviews which inform me, without gaining the slightest useful resonance from the Lovelace poem, that 'two good stories do not an anthology make' (etc, etc – there is also a suspicion in this and countless similar cases that the writer is having a stab at saying that one swallow doesn't make a summer, but has been diverted by the greater gravitational pull of the other cliche). Likewise I cringe when told of anything whatever (but in this case Paul's Erickson article) that 'like Topsy it just growed'. How many people who trot out this all too well-known phrase or saying actually have Uncle Tom's Cabin in mind, I wonder? (It was, incidentally, regarded as a bit hoary when Brewer mentioned it in his Dictionary of Phrase and Fable around 1870.) As for the famous description of any kind of mutability, whether occurring in Parliament, in the midst of the Empty Quarter or halfway out to the Magellanic Clouds, as a 'sea-change' ... oh, never mind.

Killer reviews. Mark's assault on Spiral Terra seemed OK to me: short, effective, making good use of quotations from the awful book – I shall not lightly forget the sinister fellow who ate 'Live cats, some of them dead.' Further polish will come with practice.

Replying to reviews. It is not a dread cosmic rule (Vikki) that authors shouldn't respond to hostile reviews. It's merely wiser to avoid leaping into the fray – aside from questions of demonstrable factual error – because some hideous fate seems to decree that the author ends up looking a prat. I like Freda Warrington but thought her attempt at a Very Very Sarcastic response to a poor review in Vector was painful and diminished her. I quite like David Wingrove, but his continued level of over-reaction (I still cringe at the blustering circular to BSFA members that began by saying that he was contacting us directly because on past form the wicked BSFA wouldn't allow him any right of reply – prompting the response that as a former editor of a BSFA magazine, he should know) has made him a bit of a laughing-stock in several quarters, leading to such absurd backlash as Joseph Nicholas's 'Wingrove moratorium' proposal. Can I recommend two sensible essays by Paul Fussell: 'Being Reviewed: The A.B.M. and Its Theory' in his The Boy Scout Handbook (1982) and 'A Power of Facing Unpleasant Facts' in Thank God for the Atom Bomb (1988)? A.B.M stands for Author's Big Mistake....

Robert Jordan. I got an inkling of how Jordan sustains such long fantasy novels when Paul Barnett, with his Jordan proofreading nearly over, wrote in relief:

'Champagne corks will pop. Well, sort of pop. Almost. The more I think about it, "pop" seems approximately the right word. I tug on my hair. Must stop tugging on my hair! I think. Well, perhaps I think. I'm so tired of thinking. Over in a shadowy corner of the room ... but could it be? Perhaps. Back in Emond's Field, old Lini used to say, "Take it as it comes; release it as it goes." Take it as it comes; release it as it goes! "Pop" – it's a word. Champagne corks are released as they go – Lini was right about that. I'm tugging my hair again! I must stop doing that! Otherwise I'll be releasing it as it goes, like Lini said! As good as.'

Reviews. Thought you might like to see a piece recently written for the New York Review of SF, who may yet bounce it. Reacting to NYRSF's occasional slight pomposity, I like to send something unexpected or slightly teasing ... my last piece there was the 'sf foodie' speech seen in Cloud Chamber 41. The erudition of CC's readers being notorious, I needn't add – as I did for the NYRSF editors – that this is not one of those Borgesian hoax reviews of a fictitious author.

Key Reading

• From the above it is possible to infer that I'm not entirely sure why I like FK's dotty little books of not-quite-sequiturs. In a letter the great man himself corrects a minor puzzle: '"Obsequies ..." has been around for ages as a title; I didn't get around to actually writing it until recently.' [7 September 1993] With greater honesty than many a writer – well, that the average Langford in the street – he agreed with me that 'Unspeakable Desolation ...' was a truly wonderful title and that he'd pinched it from a travel book which would duly be credited as inspiration.