Fog Creek Software
Discussion Board

Left field competition to MicroSoft?

Hi everyone,

I just finished reading Don Norman's book 'The Invisible Computer'.  In it he imagines a new world of 'information appliances' that are basically software/hardware devices specialised for a particular job/s (like the Palm Pilot, but maybe even more specialised - maybe a dedicated web viewing appliance or dedicated MP3 player).  Norman feels this evolution of computing is very likely to happen.

One interesting thing with information appliances is that Operating Systems become much less important - dedicated devices don't need a great big Operating System to manage them.  Instead, Norman suggests what will be needed is a Universal Communication Method so all the small devices can easily swap information between them, if appropriate.  This could be a threat to Microsoft - the diminishing importance of Operating Systems, and the fact that a competitor might be able to come up with this Communication Infrastructure that would be more dominant than an individual O/S.  Also, if there was a universal communication medium with no standard O/S then maybe people would be freed from having to only develop Windows style apps.

Any thoughts?

Sherlock, probably out of his depth ;0)

Thursday, April 18, 2002

Small devices? Like those Windows CE/PocketPC ones?

Interchange of information.. SOAP/.net

Microsoft gotcha covered.

Robert Moir
Thursday, April 18, 2002

The technologies to do this kind of thing already exist. Mix some bluetooth with some JINI and you've got a distributed, flexible system in which independent devices can communicate.

But what are they going to talk about? You need standards for interoperability. Say I want to enable my new cable box to communicate with a credit card to charge for a pay-per-view movie. I'd need to handle authentication and authorization, possibly handle multiple currencies, etc. All of which would require standards...and complex ones at that.

Norman has interesting ideas, but he almost always oversimplifies things because he doesn't understand implementation.

Matt Christensen
Thursday, April 18, 2002

The obliteration of operating systems is an absurd concept. If you have a device, and it has a CPU, it needs an operating system.

Of course, I could be wrong (I've never programmed for an embedded system), but even teeny tiny little devices benefit from having an operating system to handle hardware-related issues. That way, application developers can work with a level of abstraction that they couldn't before (using malloc() to allocate memory is simpler for app developers than writing brand new memory allocating bytecode from scratch).

Besides, the devices that Don Normal is referring to are still several years down the road. At that time, the teeny tiny devices are likely to be easily as powerful (or more so) than the desktop computers from a few years ago. If your PII-266 needed an operating system, why wouldn't the snazzy new devices of 2009 need an OS too?

Benji Smith
Thursday, April 18, 2002

My friends and I imagined the day that Photoshop would require it's own hard drive, CPU and graphics card to run.... Sort of like a Playstation, but for Photoshop.

I don't completely agree with Don here (not having read the book). I can certainly imagine an expansion from one machine to multiple machines, but certainly at some point we'd want a contraction of multiple systems into one. To turn our MP3 player, Game Machine, CD Player, Web Browser, DVD Player, Stereo Receiver, VideoPhone, TV/Cable/Satellite receiver, etc. into one integrated machine... sort of like a computer.

After all, who wants all that clutter?

Mark W
Thursday, April 18, 2002

Exactly, Mark.  I remember the introduction of the BlueTooth idea a few years ago. People were saying "Imgagine: Your pager, cell phone, pda, mp3 player, digital camera, and laptop will all talk to one another!!"

Today the focus has changed. Who wants to carry all that stuff around? I don't want to carry more than one gadget and I want it to do everything

For that, you need a fairly sophisticated operating system. Embedded systems will continue to get more complex, not simpler.

Benji Smith
Thursday, April 18, 2002

maybe. one can buy an integrated stereo or separates. if one buys an amplifier, should it be an integrated amp or pre+power? there is room for all of those in the market. for myself, i tend to prefer one thing that does one job, but perhaps i take "high cohesion, low coupling" too seriously. still, it does make change easy...

Thursday, April 18, 2002

I do program embedded systems, and yes, you will need some type of operating system.
The good or bad part of this idea comes up when you want to upgrade. A PC is great, I can configure it, buy a new CD R-W or DVD player, or even install Linux.
How do I upgrade my Internet Appliance? What do I do when I get sick of my ISP and want to change? Do I throw out my web pad finger touch browser and go buy a different one? No, that costs $250. Much cheaper than a PC, too many $$ to easily throw out.
Think of small dish satellite TV. To change networks, buy an new dish and a new box. To change to cable, just stop using the $99 dish.
I do not want to sound like a /. Poster. But, what do I do when Big company X releases a new copy protected format and my old MP3 played stops working? I am SOL.
If I make these devices, this is a good thing!

Doug Withau
Thursday, April 18, 2002

note - I agree with you. As a musician I know all about pre and power amps and the benefits of having them seperated. On the other hand, lots of college dorm rooms (I'm guessing) have a computer with a 3 speaker system (l/r/woofer) as the only stereo system. Computers as DVD players never did take off, nor did TV's as computers (WebTV), but they're out there, and maybe oneday someone'll get it right.

As a specialist in both computer & audio and with a video background, I might opt for the best individual components, but that doesn't mean the low end integrated TV/VCR won't sell.

Besides, I fail to see how a computer every 3 feel constitutes 'invisible...'

On a different note, I sent an e-mail to Don Norman a while back on his website ( he sent back a response within a week or so and when he redesigned his site, sent another e-mail asking what I thought of it. I read The Design of Everyday Things too. All in all, I'd say Don Norman is one of the good guys, and I'm glad he's out there. Too bad he doesn't have a message board on his site... mwa ha ha...

Mark W
Thursday, April 18, 2002

I used to work in the embedded space ... can't say I was entirely impressed with the idea of Information Appliances.  Benji is right -- there is an inexorable trend towards making embedded devices complex.  The conversation usually goes something like this:

1. The idea: "Hey!  let's make a device that just does web surfing!  It's only going to need 4MB of RAM, so we can sell them cheap!"

2. Feature creep: "Wait a sec.  No one's going to buy this unless it can parse DHTML and view Flash animations and run Java applets.  It'll also need to do email, instant messaging, file sharing, play mp3s and oh, let's have it toast bagels too.  We're going to need to add 60MB of RAM, upgrade the CPU, and add a hard drive."

3. The realization: "Hey ... we've just reinvented the desktop / laptop.  What a concept!"

Meanwhile, management is oblivious to anyone who suggests that maybe this new monster isn't going to be any less expensive than your average laptop.  After all, it's an EMBEDDED DEVICE.  It's got to cost less!

Like what Matt said ... the idea is interesting, but completely oversimplified.

Thursday, April 18, 2002

As sherlock points out, maybe this is a left field threat for Microsoft ( which is neither here nor there for me.) 

I actually like Microsoft's work on Pocket PC and consider that it will supplant Palm and the other variants. However ... as mentioned above, Microsoft has indeed "beefed up" its platform for the new .NET version (CE.NET), to the extent that it seems it may be sluggish on current devices.

Also, apparently Microsoft and Intel are preparing to move into phone technology, at the chip level according to a report I read. The traditional phone companies apparently consider the duo won't have a hope of mastering the nuances. This will be interesting to watch.

On the other hand, the phone companies are currently trying to nurture independent software developers, and a lot of them have a lot to learn too. E.g SDK's that don't install.

Hugh Wells
Friday, April 19, 2002

I'd like to do a quick response to some of the replies to my original post.

1) I should have been more careful when I said there would be no need for an O/S.  What I actually meant, was that the O/S would become less visible to the user, and not such a feature of the device.  If you are having dedicated devices supporting 2 or 3 key tasks you don't need a monolithic Windows style operating system with taskbars, start menus, icons, control panels, printer mangers, etc.  You will have something closer to a Palm: I press my nice physical 'notes' button, and I write my notes - I don't worry about the O/S.  All the latest work with XP and the new Mac O/S is just trying to put window dressing on these horrible large O/S systems that intrude on the user's work.  Sure, if we make it 'pretty' users won't mind it so much - they may even come to like it...

2) Dedicated devices Vs. convergence (or modern PCs)
I don't think this is an EITHER/OR argument - there is room for both.  A quick analogy - a handyman may use a swiss-army knife for some jobs, its handy to have around, it gets you out of fixes and it combines multiple features - nice.  However, the same handyman would never give up his toolbox of nice dedicated tools, each that fits the job far more perfectly than the pocketknife ever could (and does the job ever).  Same with technology - sure, we may want to keep a dedicated workstation for office tasks - however, we may also want dedicated appliances for things we do very often, where ease of use, pleasure, convenience are at a premium.  The problem with combined devices is that the more jobs something does, the less well it can do each individual task - the device can no longer be specialised - you also then need an additional bit of software, the O/S, to let you juggle between all the tasks the device can do.

3) Technology problems - yes, information appliances may be hard to do, and a universal communciations protocol may seem an insurmountable hurdle - but this may just be a matter of time.  Just because something is difficult to do doesn't mean it isn't worth doing, or can't be done.

4) A final note - the world Norman imagines is a world where the computer technology has matured and computers are seen as products, not technology.  I.e. they are robust, reliable, fun, stylish, dedicated - taken for granted.  They fit the task beautifully.  You don't worry about looking after them, defragging them, learning an O/S, etc.  They become 'invisible', just as products do.  They fit our lives so well, we stop thinking about them and get on with what we want to do.

Thanks for your replies, it has been interesting to hear a set of views mainly  opposed to my own!

Sherlock, the minority

Friday, April 19, 2002

> universal communciations protocol may seem an insurmountable hurdle - but this may just be a matter of time

I do communications applications for PCs and servers. Re having no O/S, sometimes I see dedicated devices that I can't easily compete with, because they've put some crucial algortihms (e.g. image processing) onto a dedicated chip, or an intelligent card. Still, there are good economic reasons why you don't use dedicated custom chips for every job... hardware prototypes are fantastically expensive, lacking the economy of manfacturing scale... therefore, we write software too. Re devices interoperating via standard protocol, yes they do that... but do you know how many different devices there are, and how many different protocols for different purposes... there are all the internet protocols, all the hardware protocols, all the telephony/ITU protocols... and those are mostly just the low-level (hardware, transport, and network), let alone the higher levels that involve understanding the data. There isn't just one species of Life in the world, and there isn't just one protocol either. Old protocols fade away, more and more new protocols are implemented. There is market/evolutionary incentive towards interoperability, but on the other hand each new application ("Voice over IP", for example) seems to call for a new protocols being added to the existing suite, and corresponding protocol converters (e.g. to connect phone switches to the internet).

Christopher Wells
Friday, April 19, 2002

Apparently 3Com believed in Don's vision as well:

but without any success. The products you mention - mp3 player, palm are all unique devices, only the internet appliance is something the computer does natively. (By mp3 player, I'm referring to a walkman sized device.)

I'm guessing the products that would benefit from being small and unique would become so, other products that are large to begin with, and share a lot of common components, will remain in one place. For example word processing and internet and photo editing and perhaps one day TV watching, though since computers tend to be solitary and TVs social I'm not sure how that will play out.

In the audio industry I'm seeing a move towards people with two computers - one for audio editing, and one for everyday use. Mostly because you don't want to clutter your speed & memory dependant real-time audio editing with unrelated stuff.

I'm still not sure I follow Don's line of thought here (again, not having read the book). He seems to be saying "dishwashers and refridgerators will be seperate things in the future." What things that we currently use one device for will be seperated?

I was in Florida recently and they have Comcast there. A couple of dozen radio channels are bundled in with normal TV channels. If your cable is hooked up to your stereo, it becomes fairly easy to turn that on instead of the radio. Why? Because you're looking at the channel listing already. There is no "radio guide" though that might change if people actually pay for XM radio.

As more and more people have home theaters, they integrate the TV, DVD, VCR, CD and Stereo, items that were in two seperate camps until recently. Even the DVD now can do dedicated audio - a storage medium is a storage medium. DVDs also does video games. One round disc, multiple uses. I just read about a new technology that will allow you to put 600gb on one DVD, enough for 20 movies they say. 20 gb on a postage stamp sized storage medium.

I already mentioned college kids with their computer-as-stereo. Now imagine the computer hooked up to the home theater system as the only DVD/CD/MP3 player, it's not too hard to do, one of my co-workers downloads movies off the internet and watches them on his computer. One day he'll be able to do that with DVD quality audio and video.

Also, as processors and memory becomes cheaper and faster, there's no reason not to embed Linux as the operating system in your home alarm. No reason for proprietary chip-based OS's anymore. I just read about a memory system that's cheaper faster than Flash ROM, allowing you to embed an OS on it to be instantly called up.

Will you have a seperate OS in your alarm, or will it utilize the power of your home PC to show a 3d image of your house and all of your zones & entryways, even with video output to your computer, which you can record in mp3 format or stream to your alarm company or the police. Webcam meets home security. And never mind motion detectors, computers with fuzzy logic can tell the difference between a curtain in the breeze, your cat, and an intruder.

Seperete device? With it's own UI? Windows, Linux or Palm powered? It almost seems counter intuitive to make someone learn a new UI for every device they own, but then, not every UI matches every situation. "Press 1 to leave a message. Press 2 for Crushed Ice. Press 3 for the Spin Cycle..."

Again, what applications will be exploded into different devices and what devices will be brought together into a single machine?

Well, that was certainly longer that I expected it to be.

Mark W
Friday, April 19, 2002

*  Recent Topics

*  Fog Creek Home