Coulda, Woulda, Shoulda

Thursday 7th November, 2013

As a company Apple seems to court the naysayers. Of course, it’s been about to collapse for last 25 years, with plenty of examples of things not going to plan, technology that wasn’t adopted, products that failed to sell, or that died before version 2. But often during the last three decades those cunning designers at Apple have shown remarkably prescience. Their triumphs are clear. But how about this: examine those seeming failed products and technologies, and you’ll find that they naturally fall into three categories of failure: from style over substance, from ideas embodied ahead of the prevailing technology, or from trying to solve the right problem, but not necessarily in the right way.

Apple have produced a fair few Mac models that some might consider failures. Although any list of examples will therefore be subjective, I suspect there are examples that surely we can all agree with.

Affectionately known as the Luggable, the Macintosh Portable is generally considered a FAIL (with accompanying Family Fortunes’Agh-agnn' klaxxon.) Heavy and expensive, and until they added the v2 backlight it was unusable in daylight. The weight was due to lead-acid batteries: despite giving a (then) phenomenal battery life, they died permanently if fully discharged, and being wired in series to the supply, once dead you couldn’t even use it plugged in to the wall. Before consigning its reputation to the scrapheap, the defence begs the the jury’s indulgence with some choice pub-quiz facts. It was the first computer to send an email from space (how cool is that?), and it had an reconfigurable (read ambidextrous) trackball. More importantly, the hardware design effectively formed the basis of the Sony-built PowerBook 100 (with the trackball in the centre.) And as anyone (ahem) who owned one will tell you, the Pb100 was one of the greats.

The G4 Cube is the canonical style-over-substance Mac. It wasn’t even strictly a cube: a 200mm cubed CPU housing yes, but held in a 250mm high acrylic box. Under-powered, with the cd drive vertically mounted, external speakers, an external (and similarly sized) power supply, the box would crack if you didn’t use kid gloves. And there was no audio-in. Oh, but it looked good - good enough to share its own Simpson’s episode, and as product placement in dozens of movies and TV shows, and behind the scenes on Star Trek: Enterprise’s bridge. It failed because it was both too expensive, and slow. The form-factor was almost right: at introduction Mac Mini was treated as a miracle new design, but in essence it was just the cube done right. You can stack 5 new Mac Minis into the same volume, and the power brick is built-in. Despite it’s failure though, it’s still a sought after object of desire, not least because it makes an excellent MacQuarium.

With every new Mac’s arrival, old Mac hands ask the same question: Is this the machine engineering wanted to ship, or did marketing get hold of it? To some, the entire Performa series was a marketing disaster. They lacked oomph, the occasional maths coprocessor, had weird memory configurations (5Mb maximum), and there was a new version every 3 weeks (70+ models between September 1992 and November 1996.) But the idea of getting our grubby paws onto a cheap(ish) colour Mac was worth the minor flaws. Yes: definitely oddballs from the Marketing department. Let’s single out the oddest, the Performa 610 DOS compatible. This was a 68040-based old-school Mac with a 25 MHz Intel 486SX co-processor card stuffed into its Processor Direct Slot. Despite the name it shipped with Windows 3; you could toggle between OS’s on the fly, or run two monitors simultaneously. These days you can do it all in software, but back in the day this was bleeding-edge.

There are probably a number of 20th Anniversary Macs in the UK. A small number. Of which, at a guess, the bulk of them are in glass cases. It was an all-in-one unit (if you discount the large external sub-woofer) with a vertically mounted CD-ROM, built-in TV tuner, built-in radio, 2GB drive, an active matrix LCD display, but only 10in deep. It was the first desktop PC to use an LCD, and given a long enough ADB cable for the keyboard, you could actually wall-mount it. It was massively over-priced ($7k+), offered a track-pad instead of a mouse, and the screen was only 12.1 inches. It was ahead of its time, by about 7 years. It was the proto-iMac G5, and another example of the internet fixing problems that used to be dealt with by splashing the cash on hardware. Macs don’t need TV and Radio receiver hardware anymore, because we’ve got YouTube and iPlayer. If only large LCDs had been available, it could have been the first media centre PC.

A history of Apple is of course not just a history of Macintosh hardware. Apple attempt to innovate in pretty much every arena they enter, with hardware near-misses, cul-de-sacs, and failures aplenty.

Any list of purported Apple mis-steps will include the Newton narrative: Steve Jobs hated John Sculley’s baby, styli were evil, the hand-writing recognition sucked, Apple was spread too thinly, and engineers were better deployed elsewhere. So Jobs axed the Newton. But Newton was not just a PDA. The eMate 300 was Newton with a keyboard. Aimed at the education market and priced accordingly ($800), it was a pre-cursor to a number of future technologies. It was a tough little thing: entirely solid state, a practically water-proof keyboard, memory expandable, and shipped with basic office software. How far head of its time was it? If only it had been a wireless device! Given the cost corollaries of Moore’s law, the eMate 300 was clearly the steampunk embodiment of a One Laptop Per Child device.

Apple have a great track record for opening up new markets. Did you know that Apple produced the first consumer digital camera? Actually, no they didn’t, it was the second. But it was the first mass-market one. The QuickTake 100 was the pinhole camera of digital world: it could only take 24-bit PICT files, with a maximum resolution of 640x480, and then only 8 at a time. But it was so freeing. It cut out the whole’snap, develop, scan' process to which we were inured. For a design conference in May 1994, I managed to whip up the proceedings notes as PDFs with pictures of the speakers, in under an hour. For the delegates this was pure Arthur C Clarke magic. So what was the reason it failed? It awoke those slumbering photo incumbents to the fact they could no longer delay the move to digital.

Some Apple kit has a really short shelf life: case in point, the Apple Network Server (ANS). You’ve probably never heard of it. I suspect I’m probably the only person I know who’s seen one, let alone used one. The ANS was a PowerPC-based box designed to run AIX (see below) and was Apple’s rather confused attempt to break into the Enterprise Server market. In essence it was a hugely expandable PowerMac 9500, with the ROMs removed, stuffed into a tumble-dryer machine sized cabinet. It was designed NOT to run MacOS, and in hindsight to be a total FAIL. Until the new Mac Pros arrive (ahem), at $19,000 it remains Apple’s most expensive catalogue-priced machine. On second thoughts it does manage to succeed as the best tech bargain ever: a recent example was sold on eBay for $1.56. Shipping $120.

Apple actually made a games console. Well, they designed it, and Bandai made them. But they only made 40-odd thousand. The Pipp!n (yes, the ! means it’s yet another marketing fail) was also Apple’s attempt to create a network computer (NC). Oracle tried to establish an’NC Consortium', including Apple, to produce thin clients: diskless, internet-based computers. So Pipp!n is a double FAIL. It would have failed just for the usual reasons games platforms fail: it was too expensive ($600), used encoded CDs, used a TV as display, was massively underpowered. It also failed as an NC, with no network connection, and just a 14.4k modem. The icing on the cake? Just a single 3rd party developer. (Apple learned from this mistake, with developer engagement perhaps the main reason for iOS’s success.)

Apple has plenty of entries in the ‚Äö√Ñ√∫where are they now‚Äö√Ñ√π software back catalogue: Some are much missed, others deserving of their place in the pantheon of coding ignominy. Here’s a few noteworthy ones.

When Microsoft brought out Object Linking and Embedding (OLE), Apple felt it needed a’me too' to compete. Publish & Subscribe would be better than OLE, with all sorts of bells, knobs, and whistles. You could publish part of a spreadsheet document, say the sales chart, and then subscribe to it in a word-processing document. The word-processing document would then see any updates in the original, and offer to update as well. Yep, P&S failed on three levels; Developers found it difficult to work with, so it was poorly adopted. It would often break because the original was on a floppy disk at home (remember kids, not everyone had a network). Users couldn’t in fact see the point of it. The Mac OS had a great scrapbook, and copy & paste worked well. Apple relearnt the zeroth law of engineering.

Apple has had more than one stab at producing a Mac-friendly UNIX operating system. Why? For the same reasons OSX is based on UNIX today: UNIX is stable, has a huge back-catalogue, and gets you into enterprise. A/UX was their first go. Based on UNIX System V, it was POSIX compliant, had an X-Windows server, included TCP/IP networking, and would run on something as lowly as Mac SE/30. It allowed you to run Mac applications, alongside UNIX applications. To say it wasn’t mainstream was an understatement, but the technology it contained led on to many innovations we take for granted today. It ran Mac software transparently, but it did so by using a virtual machine, and offered the beginnings of a mainstream GUI interface to UNIX. This was before GNU/Linux, and it was very much closed source.

Apple were so keen to get the Mac into engineering and higher education sites (or at least not have them excluded from them) that they even tried the ‚Äö√Ñ√∫if you can’t beat’em, join’em‚Äö√Ñ√π school of software development. The Macintosh Application Environment (MAE) was an emulation package designed to run on Sun Solaris and HP ultrix that let UNIX denizens run Mac applications. In fact, MAE emulated pretty much the entire MacOS 7.5 in an X-Window. So you could do your finite-element-analysis work on your trusty SPARCstation, and then fire up MacWrite to produce the report in. You could copy and paste between UNIX and Mac environments, open files from NFS volumes, and perhaps most importantly run the sorts of applications not normally found on standard UNIX machines i.e. PageMaker or QuarkXpress.

Although dropped in 2004, HyperCard was tool for creating object-based documents (stacks) that contained both the code and data. It was easy to use, and fun to program using english-like HyperTalk. How important was it? It influenced or inspired the development of Javascript, AppleScript, HTTP, web-browsers, and even the concept of the wiki. Even the original Myst was written in it. For some it was the killer-app for the original Macintosh. So why even consider it a failure? Its creator, Bill Atkinson, is fairly clear on this point: a lack of network awareness: “I grew up in a box-centric culture at Apple. If I’d grown up in a network-centric culture, like Sun, HyperCard might have been the first Web browser. My blind spot at Apple prevented me from making HyperCard the first Web browser.”

Apple aren’t just boxed products. In the internet age online services deserve our consideration too.

You’ld have to be very old (40 or so) to remember AppleLink. It was Apple’s pre-web online community, accessed via client software and a modem. It was for distributors and developers only, and boy was it expensive. Apple tried using a 3rd party (Quantum) to create consumer - AppleLink Personal Edition, but it was even more expensive. eWorld was therefore Apple’s third go at an online service. It looked great, had a web browser, email, and FTP, but it was massively expensive, and Mac only. It failed, in part due to fierce competition from a recently renamed Quantum - America Online - or AOL. (If at first you don’t succeed, try try, try again, right. After their’success' with eWorld, it took another 4 years for Apple to dip its toes back into the online world. ITools/.Mac was really just about giving Mac users a @mac.com email address. And it was free. See, they had learned their lesson.)

Hindsight: it’s wonderful thing. What you thought of at the time as failure often turns out to be an important stepping stone; technology hailed as the next big thing may be next week’s chip-paper. Apple has a great track record for both. Without HyperCard the web might have taken longer to start. With no QuickTake we might still be scanning photographs. If they’d fixed the handwriting recognition before shipping we might all be using Newtons instead of iOS devices. Had eWorld worked, the internet might still be a walled garden, and e-commerce the next big thing. Be thankful for small mercies.

Clone Wars

In terms of the what-might-have-been aspects of Apple, the biggest elephant in the room is probably cloned. Whilst any off-the-shelf PC out there will pretty much run Windows or Linux, only an Apple Machine can legally run MacOS X. Until 1995, the only legal way to run Macintosh software on a non-Macintosh computer was to use a machine built using ROMs’rescued' from an old Macintosh. (There were plenty of these, including the Outbound, a lighter, $2,500 cheaper competitor to the original Macintosh Portable.)

In 1995, Apple launched a cloning programme, so that third party manufacturers could build and sell Macintosh clones, with a flat fee paid to Apple for every clone machine they sold. At the time this was seen as a win-win for all concerned. Apple won, because (a) it increased their market penetration, which at the time as between 6% and 9% depending on who you believe, and (b) given the issues they were having producing a replacement OS for the Mac they needed the additional revenue stream. Mac users won because the competition between the clones resulted in lower prices and/or improved specifications and expandability. The programme ran for two years, during which time eight or so manufacturer produced a selection of PowerPC based clones running MacOS 7.5.

Of course, Apple soon realised that this meant they were losing out on the sales of high-margin, high-end machines. And with the return of Steve Jobs to the fold in 1997, unsurprisingly they decided they wanted to end the deal. This is where it gets interesting. The programme was halted by the release of System 8. Apart from one manufacturer (UMAX), the clone licences were only good for machines running System 7, so when system 8 came out, the clone wars were over. Actually, with a bit of fiddling, most cloned boxes can be made to run anything up to MacOS 9.1. As an aside, it’s been claimed that even post 1998, Steve Jobs tried to licence MacOS to Compaq (remember them?). Now there’s a’what if' to ponder.

Alternative OSes

1994: Microsoft had announced Windows95, and Apple were playing catch-up. Despite managing the transition to PowerPC almost seamlessly, System 7.5, codename Mozart, lacked many of the essential features in its MS rival, most notably, pre-emptive multitasking and dynamic memory management, both of which meant that it was more likely to crash than a PC. So to counter this, Apple started work on not one, but two successors to System 7, codenamed Copland & Gershwin. (See what they did there?)

Apple was also suffering a shortage of engineers, with people being moved from non-essential projects to work on Copland. To maintain backwards compatibility, Copland was built with two environments: Bluebox emulator (for existing apps) and Yellowbox for’modern' apps. There was a lot to do. Even so, engineers brought their pet projects with them, and Copland’s functionality grew like Topsy.

Surprise! The ship date for Copland kept slipping. Apple, and the Copland project, had fallen foul of Fred Brook’s famous Second System Syndrome: ‚Äö√Ñ√∫the tendency of small, elegant, and successful systems to have elephantine, feature-laden monstrosities as their successors due to inflated expectations‚Äö√Ñ√π. Copland had those in spades. So much so, that then CEO, Gil Amelio (who’d incidentally helped develop the CCD image sensor!) looked to buy in an OS from outside. After toying with BeOS, they decided to buy NeXT and the OpenStep OS. The plan was to port it to the Mac, with the codename Rhapsody. Rhapsody was released, briefly, as Mac OSX Server. Copland was cancelled, as a failure. Except those cool pet projects were either back-ported to MacOS 8 & 9 e.g. the aforementioned OpenDoc and QuickDraw GX, or were brought in to MacOS X. Blue-box formed the basis of the Classic MacOS emulator, whilst Yellowbox, enamed Cocoa, now forms the basis of modern OS X and iOS apps. Spotlight search, iChat, and even the Safari browser - they all started as features in Copland. Apple also got a new employee: Jobs, S.