It's a subhumanoid meltdown, comin' on through. Today we're going to journey back to the distant past of 2005, and have a look at a PC tablet from that year. I picked it up cheaply from eBay because I was curious about the concept; PC tablets had been a thing through my entire adult life, but I had never felt an urge to spend any money on one. In this respect I was not atypical. Pre-iPad PC tablets sold in very small quantities mostly to businesses that bought them on the off chance that they were the future, or to a few enthusiasts who presumably gave up in disappointment shortly afterwards.
I suspect that the better-looking, lighter models - such as the ThinkPad X61T - were status symbols for business class commuters. In the very narrow context of reading PDF files whilst crammed into an aeroplane seat a PC tablet was, for a few years, the smallest hardware that would do the job, but for almost everything else they combined the weight and expense of a laptop with the awkwardness of writing things with a pen. Neither the OS nor the applications were optimised for stylus operation and what was the point?
The weight issue really killed the concept, and was an obvious problem that seems to have been swept under the carpet by the PC industry, by manufacturers and the media alike. CNET's review of the generally good X60 Tablet includes the following, telling lines:
- "in our use, the ThinkPad felt light enough to carry around every day and even hold in one arm while taking notes in tablet mode."
Bear in mind that the quote dates from the second generation of the third wave of PC tablets. In fifteen years and numerous attempts since the GRiDPAD and IBM's 700T it seems that nobody was prepared to admit that the concept was fundamentally flawed, and that until the weight issue could be solved everybody should just go home. Except for Steve Jobs, who (grits teeth) was right yet again.
Ultimately the PC tablet was... a horse... that had been delivered dead. The PC industry periodically dragged its decaying corpse out of the stables to put make-up on it, and give it a thrashing, like something from a Peter Greenaway film. Like A Zed and Two Noughts or The Baby of Macon. Ribs poking out of the grey, leathery flanks, eyes pecked away by vultures, but the horse has beautiful lipstick and the eyelashes are just divine. A mechanism inside the head gives the illusion of life, and the head turns to you and its mouth opens and closes and sweet music emerges from a hidden speaker where its throat had been. You catch a whiff of its breath; the scent of Chanel No 5.
Around the horror a crowd of suited executives ask each other why this thing has not captured the public's imagination. Surely it can't be the perfume? So they change the perfume. Air pumps drive pneumatic tubes in its jaw and from the bloodied eye sockets there drips a single tear... of oil. That is what I think of when I think of the PC tablet.
Live music applications are amongst the few reasons to own a PC tablet nowadays, because the pen is more precise and faster than a trackpad, and they take up less space on a table. 1024x768 is just enough for Ableton, but higher resolutions are better. In this video I spend as much time moving Audiomulch's windows around as I do controlling the music.
Which has the positive benefit of making me look busy.
Which has the positive benefit of making me look busy.
The TC4200 has a melancholic air to it. Hewlett-Packard's earlier TC1000 and TC1100 were among the most forward-thinking PC tablets of the second wave that emerged in the early 2000s. At a time when most contemporary PC tablets were just laptops with a swivelling screen, the TC1000 resembled a modern-day Asus Transformer, with a keyboard stand that could be removed from the processing unit to save space. It was stylish, and to my eyes hasn't dated badly at all. The TC4200, in comparison, is visually drab and very conservative.
Unfortunately the TC1000 was hobbled by its processor, a Transmeta Crusoe. This was a non-Intel chip optimised for low power consumption that achieved compatibility with x86 software by emulating a Pentium. At the time, Intel's top desktop CPU was the Pentium 4, which was fast, but too hot and power-hungry for portable computers. As a consequence there was a brief window during which Intel's competitors might have cornered the market for powerful but cool x86-compatible mobile chips. The Crusoe was bouyed along on a wave of hype, and seemed just the ticket. In mid-2000 IBM announced that their forthcoming ThinkPad 240 was going to use a Crusoe, which seemed like a big win for the company. But in October of that year IBM dumped Transmeta and decided to power the 240 with an Intel Celeron instead. Nonetheless Transmeta's IPO, at the height of the dot.com boom, was a big success, and the company ended up with a market capitalisation of almost $6bn.
Unfortunately the Crusoe was very slow, and the TC1000 was not a success. Transmeta struggled on to 2009, by which time it had been reduced to flogging off patents. The company was eventually sold off for $255m to another company called Novafora, that went bust almost immediately afterwards. The sad thing is that there was a market for low-power mobile chips, a huge one, for mobile phones. ARM was in the process of dominating that market in the early 2000s and I doubt Transmeta could have beat them, but it was a huge market and there was room. By concentrating on x86 compatibility rather than the mobile phone market, Transmeta backed the wrong horse (a living horse this time).
HP replaced the TC1000 in 2003 with the TC1100, which was powered by an Intel Pentium M. The Pentium M was a breath of fresh air - and would have crushed Transmeta, even if the Crusoe had been decent - but although the M-powered TC1100 was a much more practical proposition than the TC1000 it was not enough to overcome the market's general lack of interest in PC tablets. This chap has fond memories of his, and purely in terms of bare specification it compares well with a modern iPad. It seems to have been a popular movie prop, appearing in a couple of films and The Big Bang Theory, amongst others. The TC1100 is something of a modern design classic, an unusually stylish machine for a company not generally famed for its cutting-edge aesthetic.
XP Tablet Edition's handwriting recognition is surprisingly not bad. Handwriting - with a pen, on paper - was a means of communication from before the iPad age; it still exists but is in decline. I hate it and always have, it's limiting.
In contrast the TC4200 is boring. It replaced the TC1100 in 2005 and was essentially just a Hewlett-Packard NC4200 laptop with a swivelling screen. It's as if HP's designers had been instructed to abandon their dreams and just create a PC tablet as quickly and cheaply as possible. As a consequence the TC4200 is too heavy to use as a portable tablet - it's based on a design that was originally supposed to rest on a lap, rather than be held aloft - and in that respect it simply doesn't work, just like all the other PC tablets. There was a "boom" in PC tablets in 2005, led by Microsoft and Intel, and I have to wonder if the TC4200 was simply a contractual obligation, a result of Microsoft asking its partners to produce something to put a relaunched XP Tablet Edition on, in exchange for cheaper licences for Vista, which was sure to be a success when it arrived. None of the TC4200's contemporaries stood out, none of them are fondly remembered by anybody nowadays, they are not remembered at all. And so they are quite cheap on eBay.
HP still makes PC tablets, most of which can be split apart from their keyboard, a design decision that ironically owes more to the TC1100 than the TC4200. The company also makes a range of Touchsmart laptops that combine a conventional laptop with a touch-sensitive screen. After using the TC4200 for a while I find that the stylus input works well as a complement to the keyboard - the stylus is faster and more precise than either a trackpad or a trackpoint nipple (the TC4200 has both). You can move the pointer from one side of the screen to the other almost instantly, and click very precisely on a point. In fact I have come over to the idea of styluses. They're deeply unfashionable, like onesies, but there comes a time when the warmth of a onesie outweighs the fact that you're wearing a big romper suit. You have entered the post-sexual phase of your life, but at least you're warm.
The TC4200 was part of the final generation of 4:3 laptops. It has a 1024x768 screen, which means that the edges of most websites are chopped away in portrait orientation (a few contemporary tablets used a 1400x1050 panel, which is much more practical). If I had developed a PC tablet I would have pushed for a square, 1024x1024 screen. Probably wouldn't have sold, but it wasn't going to sell anyway, so why not go down in a blaze of glory?
Modern tablets use a "viewport" model for rendering pages - they render the page on a virtual surface, and scale this to fit the display. Desktop Firefox can zoom in and out, but it's fiddly.
The Pentium M
The TC4200 is powered by the second-generation Pentium M. The Pentium M came and went without much fanfare, but it turned out to be one of Intel's most significant modern CPUs. The company spent the first half of the 2000s pushing the Pentium 4, which was the CPU equivalent of the fictional 6000 SUX automobile from Robocop - it was quick, but drew a lot of power and pumped out a lot of heat. Intel's original plan had been to keep the chip competitive by clocking it at higher and higher speeds, infamously predicting that it would reach 10ghz by 2011. A gain of 1ghz a year.
This rate of progress held for roughly four years, at which point Intel reached the practical limit of the design; the chip was simply too hot. The most tenacious overclockers managed to touch 8ghz with their Pentium 4s, but only for long enough to run a few benchmarks, and they needed liquid nitrogen cooling to do it. It's hard to remember now, but in the early 2000s nobody cared much about power consumption and heat, people were prepared to stuff nine fans into their PCs and use them as space heaters so that they could play Unreal Tournament 2004 at 121fps. Say what you like about Al Gore, if he hadn't graciously stood aside from the US presidency in order to concentrate on making An Inconvenient Truth, the fact is that we would all be dead by now.
Meanwhile Intel's arch-rival AMD was selling a range of chips that were just as powerful as the Pentium 4, but ran at a slower speed and consumed less power. Intel's marketing at the time seemed to emphasise raw clock speed as a sellable metric, and so the Pentium M was a surprise. It was based on technology that had been developed for the earlier Pentium III and Pentium Pro, and it ran at lower clock speeds than contemporary Pentium 4 chips. Desktop owners generally ignored it, but Tom's Hardware for one spotted that the Pentium M was actually faster, cooler, and less power-hungry than the Pentium 4. Unusually for a mobile chip it didn't seem to have any shortcomings, and for a brief period the Pentium M was a popular choice for small, build-your-own desktop machines (AOpen seemed to capture that niche).
Intel had tried to breathe some life into the Pentium 4 with the dual-core Pentium D, but the Pentium D itself was very hot, and its performance advantages over a single-core Pentium 4 were slight for games in particular, because mainstream multi-core processing was very novel at the time. That wasn't the Pentium D's fault, I suppose; the later Core Duo benefited from a couple of years of progress on that front. Still, Intel took the bold decision to axe the Pentium 4 and Pentium D in favour of a new desktop architecture based on the Pentium M. Core, it was called, and although there were a few single-core Core chips, most people remember the dual-core Core Duo. The Core Duo was a success not just because it was dual-core, but because each core was convincingly faster than a Pentium 4, and the chip itself was no more expensive. The Core Duo was such a success that Apple decided to give up on the IBM/Motorole PowerPC architecture - which was running into thermal problems of its own with the new G5 model - in favour of Intel's new chips, and even in 2013 a fast Core Duo machine with a good graphics card is still useful for most tasks.
The majority of laptops have their CPUs soldered onto the motherboard. The TC4200 is unusual in having the chip mounted in a socket, so it can be upgraded, which is what I did with mine. My TC4200 originally came with a 1.83ghz Pentium M, which I have upgraded to a 2.13ghz model at a cost of about £8. According to SuperPi my TC4200 now takes 1m 35s to calculate two million digits of Pi rather than 1m 43s, which is better but not astonishingly so. In contrast a 2ghz Core II Duo does the same calculation in 1m 02s, although it's hard to compare results because the Core II Duo machine was running Windows 8, and of course the surrounding hardware is different, and it's a dual-core machine, and I had more hair back then.
Nonetheless, if my maths are correct, a Core II Duo is roughly 1.6x faster than an equivalently-clocked Pentium M purely for calculating Pi with Superpi on a single core. And of course the Core II Duo is a dual-core design, not an insignificant difference. The gap would be closer with the original, 32-bit Core Duo, probably not by much. The fastest Pentium M ran at 2.26ghz, which inevitably has a price premium on eBay (£25, vs £8 for a 2.13ghz model) and doesn't make economic sense.
So, the Pentium M. When it was new it suffered from a nondescript name that was easy to confuse with the unrelated Mobile Pentium and the Pentium 4M, and as a mobile chip there were very few desktop motherboards available. For desktop users it was the other and has slipped into obscurity. It was quickly pushed aside by the Core Duo, and as a consequence Pentium M chips are essentially worthless nowadays and it's not economical to built a system around one.
I say this because I have a 1.83ghz Pentium M sitting in a box, doing nothing. For the cost of a new case, and a new power supply, and a new motherboard I could instead buy a cheap second-hand laptop, or simply keep the money and buy shares in Micron Technology (MU) or some alcohol so that I can tune my mind to the appropriate wavelength to begin negotiations with the electric voice. To attempt to engage in dialogue and perhaps climb that hill.
Centrino, and Other Subatomic Particles
The Pentium M was also overshadowed by Centrino, which was a generally successful attempt by Intel to dominate the market for laptop components in the mid-2000s. Intel hit on the idea of bundling an Intel CPU, an Intel motherboard, and an Intel wifi unit as a base that laptop manufacturers could build a laptop around; they would then have the right to put a Centrino sticker somewhere on the laptop's case, which would hopefully attract customers who had seen the Centrino name in adverts. Originally the Centrino platform was based exclusively around the Pentium M, but later iterations used Core Duo chips. Intel wanted to emphasise the platform rather than the processors, and as a consequence the Pentium M was buried under the Centrino name; it was quite common for people on the internet during the Pentium M's heyday to talk about the Intel Centrino as if it was a CPU.
The Centrino name was discontinued in the late 2000s. On a personal level I was never fond of it. The word sounds trivial. Pentium has a substantial sound, like Adamantium or Uranium; it puts me in mind of a heavy element. Centrino, on the other hand, sounds like a subatomic particle. A tiny little speck with a squeaky high-pitched leprechaun voice. That's how neutrinos speak, they have a squeaky high-pitched voice, and they say "yippee!" a lot. You've got to pick them up just to say hello. Higgs-Bosons, on the other hand, are very posh, like English butlers. Protons wear a little yellow cap with wings on it; electrons are blue.
XP Tablet Edition
When it was new the TC4200 came with Windows XP Tablet PC Edition 2005, which was essentially Windows XP Tablet Edition upgraded to Service Pack 2. It had an on-screen keyboard, a handwriting recognition engine, and a couple of tablet-optimised applications. Overall the package felt half-hearted. The tablet applications - and there were really only two of them, Sticky Notes and Journal - were unimpressive, and although the handwriting recognition engine seems to work I have barely touched it, and I suspect that most TC4200 users circa 2005 simply ignored it and saved their notes as pictures for transcription later on. Or, more likely, they saved their notes as pictures *once* and then flipped the screen around and used the TC4200 as a conventional laptop from that point.
For the tablet form factor to be a success Microsoft really needed to support third-party developers, and ask Adobe and so forth to add some tablet-specific features to Photoshop. Or, get this, the OS part of Microsoft could have asked the Office part of Microsoft to release a Tablet Edition of Office, which would have been a powerful killer app. Even if it only had a TABLET EDITION sticker on the packaging with no other changes.
Perhaps the company argued that there was no point doing so, because the stylus and text entry interface were just as functional as a conventional keyboard, and there was no need for tablet-specific applications; which raises the question of why anyone would bother with a tablet at all. Or perhaps Microsoft's OS team was unable to communicate with the company's applications team, and the company didn't have a developer outreach team. Perhaps Microsoft simply assumed that developers would flock to the tablet PC and colonise it with a rich ecosystem of pen-driven software on pain of excommunication. I don't know. As it stands the only two tablet-specific applications I can remember, off the top of my head, are the two that come with XP Tablet Edition. Sticky Notes is a Post-It simulator, and its pathetic:
6. "it", to the music
7, "it", TO JULIA! DON'T DO IT TO ME, DO IT TO HER!
Sticky Notes' wealth of functionality includes the option to export Notes as .snt files, and then import them again. Or you can copy the note as a picture to the clipboard. You can also record thirty seconds of audio as an audio note. So in theory you could make a whole animated film with Sticky Notes. If you were very dedicated. That's about it for Sticky Notes.
And there's Journal, which is essentially Windows Paint with ruled lines and a handwriting recognition component. To be fair, the handwriting recognition engine is better than I expected, although it's a shame that Microsoft couldn't implement it as an application-independent input component that could interpret any handwriting written in anything, perhaps even handwriting scanned from paper. Journal has the barest minimum of integration with external applications - you can send notes to your email client, that's about it - and it can only export documents as .TIFF images, plain text, or Journal files. It's a pale shadow of the highly integrated software that came with the Apple Newton or the Palm Pilot. I have the impression that Microsoft developed it quickly as a means of showing developers what could be achieved with a tablet, and if this is the case at least they didn't overreach themselves.
Journal still exists as an integral component of Windows 7 and Windows 8 (post-XP, the tablet components were folded into the OS, and so there isn't a Windows 7 Tablet Edition, for example).
This is why handwriting died off. If a computer can't even understand a simple word like "guzzle", what chance does a human being have? None at all.
The 2005 Tablet Boom
In 1995 top dead French sociologist Jean Baudrillard published The Gulf War Did Not Take Place, in which he argued that the Gulf War did not take place.
Arf! He argued that the "war" (in "quotes", because Baudrillard was the kind of person you could imagine enclosing everything in "quotes") had been a charade put on for "the media", and that although there had been "military action", it was not a "war2. The idea of war as a ritual rather than a genuine existential struggle was not a new idea in "1995", and I suspect that Baudrillard's isolation - he spent his whole life in academia - had the effect of stunting his intellectual development rather than broadening it.
Still, the 2005 Tablet Boom did not take place. Steve Ballmer appeared on stage to show off XP Tablet Edition on a ThinkPad X41T (reading the transcript of that event, the X41 is awkwardly shoved into a lecture about Exchange and Microsoft's "New World of Work", so it was presumably a paid advert). There was a wave of PC tablets, including the X41T and the TC4200, plus the Toshiba R10 and Tecra M4, the Acer C200, the Gateway M285, the netbook-esque Fujitsu T4000, and several models from rugged tablet manufacturer Motion Computing. The media was prepped, the submarines were manoeuvred into position, ambassadors were recalled, deadlines were met, etc.
The one thing missing was demand. Despite all the attention there wasn't any. According to this report, tablet PC sales reached a million in 2005, versus consumer laptop shipments of anything from 58 million to 65 million depending on whose figures you believe. The report predicts almost two million tablet PC sales for 2006. This article from late 2006 puts the number sold during the year at "more than one million units", which suggests that there hadn't been any growth at all. Perhaps some of those million units were replacements of existing tablet laptops, in which case there had been negative growth.
This is from Peter Goodwin's "Future World", a dull coffee table book about THE FUTURE as it appeared in 1979, e.g. maglev trains and wave power. In the ice age vs melting ice caps debate, Future World was in the latter camp.
And yet 2005 was the year in which laptop sales began to finally overtake those of desktop PCs, and so there was clearly demand for portable machines. It's just that nobody wanted a portable laptop that converted into a tablet. They were just a pointless waste of money. The TC4200 that CNET reviewed cost $2,099, versus $1,788 for an NC4200 that ran faster and had twice as much memory and a larger hard drive. Apple sold more laptops in 2005 than all of the aforementioned PC tablets combined, which must have stung.
Apple itself never launched a PC tablet. The company's Sculley-era "knowledge navigator" concepts evolved into the Newton, which was something else entirely. Post-Newton, Steve Jobs was convinced that tablets would not be practical until they were pocket-sized or at least portable.
For a short period in the pre-iPad era a company called Axiotron sold a range of "ModBooks", which were essentially MacBooks with the screen and keyboard replaced with a fixed pen stylus screen. They were expensive kludges - you couldn't rotate the screen, for example - although surprisingly they seem to have sold enough units for the company to stay afloat, perhaps because Apple's machines have traditionally appealed to the graphic design crowd, one of the few markets that might want a stylus tablet display.
Ah, but what's the TC4200 like? Used as a laptop it's a generic business machine from the mid-2000s. It's powerful enough to run the original Half-Life at a good clip, almost but not quite at the default 72fps limit:
Sorry chaps. Another time, another place we might be friends. But needs must when the devil drives.
But Half-Life 2 defeats it:
Shh. No tears. Only dreams now.
This is more a fault of the Intel 915 graphics chip than the CPU, though. Half-Life 2 running at 1024x768 with a 2ghz Pentium M and a Geforce 6800GT would probably be fine and dandy. Unfortunately, one of Valve's Source Engine upgrades totalled the lightmaps in some of the levels, which means that the eerie Nova Prospekt prison looks as if it being lit with a nuclear bomb:
Ignoring the visuals, Nova Prospekt is testament to the cleverness of Half-Life's designers. After an incredibly tough fight outdoors you burst into this awful torture-house and... nothing happens. It's empty. You take the Combine's forces by surprise, and for the first time in the game they no longer seem invincible. The tyrannical regime's authority melts away when it can no longer stop its subjects from saying "no".
For anything that doesn't require decent 3D graphics or a high resolution screen a 2ghz Pentium M is still, eight years later, sufficient for browsing Imgur, running spreadsheets, Word, etc. In a home office or small business context the PC reached a plateau of competence years ago, and although it made business sense to replace power-hungry Pentium 4 machines with relatively frugal Core Duos, there wasn't much reason for businesses to ditch their Core Duos in favour of i7s. The rise of the virtual desktop has extended the lives of older machines, by offloading much of the processing to The Cloud.
Over the last few years growth in the PC market had ground to a halt, and as I write these words sales have diminished as well, a trend which has generally been blamed on the rise of tablets. I suspect this is a factor, although it's surely more complex than that. I believe that growth in the tablet sector has just as much of a negative effect on smartphone sales as well, an effect that is not apparent because smartphones have been selling so well. And surely very few people who buy a tablet do not already own a desktop or laptop computer; and consumers who can only afford a cheap tablet were not likely to buy a PC anyway. But of course a new tablet might put off the need for a new laptop, so it's a complex picture.
I suspect that the decline in PC growth has been due to a general lengthening of the business IT replacement cycle, caused by the lack of a powerful incentive to replace Core Duo and early i3 and i5 machines; that the virtual desktop and cloud-driven return to the old-fashioned "dumb terminal" model has further reduced the need for new hardware; and that the typical consumer has decided to hold off replacing their perfectly decent desktop PC or laptop because they use it less often, because they have a tablet. For the less committed enthusiast hobbyist who upgrades his machine but doesn't build from scratch a blazingly fast SSD can breathe new life into an old PC and put off the need to buy a new one. And there is the economy. During the Happy Time, £699 was an impulse spend at the local supermarket, a perfectly sensible way of improving your credit score; but over the last few years there has risen a class of people for whom £699 is hard to come by. Until these people are dealt with they will be a source of economic malaise.
Still, I digress. The TC4200 is really held back more by the system than the CPU - the 2gb memory limit is tolerable, but the machine is further hobbled by its PATA IDE interface. On the plus side it has an SD card slot, which was unusual in those days. As a desktop replacement it has a VGA jack and three USB ports, one of which was powered to support an external "Multibay" that could house a DVD writer or a second hard drive. DVI output was only available with an optional docking unit that was otherwise mostly pointless.
The TC4200 has an unusually rubbish audio system. It has a single speaker, not a particularly good one. There's a headphone socket and microphone in, but no line in. Bah. In general, as a used buy the TC4200 is adequate but not much cheaper than its replacement, which was better in most respects, although still rubbish on the audio front. As a thing to browse Reddit in your lap it's entertaining. The biggest issue is the OS. In theory it'll cope with Windows 7 but that just adds to the cost, and TC4200 isn't worth spending much money on; XP support ends in April 2014, after which internet use is potentially hazardous. There is Linux, but the thought of getting Linux to recognise screen orientation on such old hardware doesn't enthuse me.
HP replaced the TC4200 in late 2006 with the TC4400, an altogether better option. In the end I found one just as cheap as the TC4200 and bought one too, so that I would have something to mix with (it's on the left):
The TC4400 was launched during the brief heyday of Vista, although I've wiped mine and installed Windows 7 from scratch. It's a 2ghz Core II Duo, and it calculates Pi to 2m digits in 1m 02s, exactly the same time as my 2ghz ThinkPad X61. I've also upgraded it with an SSD and 4gb of memory, which is the maximum quantity it can hold. Irritatingly you have to strip the machine down in order to reach the internal memory slot. It uses SATA 1.5 and is an altogether better machine than the TC4200 - it belongs to the modern age, whereas the TC4200's PATA interface and single-core CPU has an slightly Gen One air to it. As a used buy it's a better choice, and even if you're uninterested in music it makes for a decent albiet heavy toilet surfing machine (cough).
HP didn't replace the TC4400, at least not directly. It was the last of the TC models. HP's next laptop tablets were the consumer-orientated 2710p and TX1000, which had 1280x800 widescreen displays. The 2710p used Core Duos, the TX1000 had an AMD Turion 64. For 2009 the company launched the 2730p, which had a toughened chassis. The TC4200/4400's most likely modern heir is the Elitebook Revolve 810, which is staid and dull and hobbled by a 1366x768 screen. Dammit, this is 2013. I want more than 768 lines. The end.