Saturday, December 31, 2005

[AoyW] The PC Was Never All That Personal

By design, the PC is not a mobile computer. In order to accomodate the per-unit tax levied for Windows & Office, PC designers jam more and more electronics into the box, keeping price and power requirements high. The result is a desktop or luggable supercomputer, rather than a true personal computer, which by definition should be effortlessly mobile. The few vendors who have tried to build pocket-sized PCs end up with devices that are underpowered for Windows, and terribly overpriced (see OQO). PCs get denser, not smaller.

Low-power x86 silicon has been around for years, and for cheap. Leveraging it requires a lightweight software architecture, which Microsoft will not embrace outside the PocketPC. Mind you, the PDA, with its poor input mechanisms, is a PC accessory, not an alternative.

The architecture of the web—rendering & input on remote displays—will let power-constrained mobile computers become full-fledged personal computers. Offloading the screen and rendering logic to local PCs allows reallocation of power to application processing and battery life. By this scheme, you don't need to carry a screen device in most work environments, as PCs are ubiquitous. For on-the-road I/O you may carry one, in the best size for your immediate needs.

Handheld-to-remote-display is a perfect application for Bluetooth; Intel Research prototyped it over 3 years ago (see Web 2.5 Pocket Servers). Unfortunately, handheld makers don't grok the potential of web architecture, consumed as they are with building so-called smartphones to head off the threat posed by ordinary mobile phones.

Real pocket web servers will eventually appear—they've been technically feasible for several years—if not by Bluetooth, then by UWB. They will likely evolve from flash drives. Contemporary flash drives are already a plausible deployment vehicle for always-on-you web servers. You plug in the drive, your personal/peer-to-peer web services run directly from it, they hit the local screen and possibly others via Wi-Fi, and you unplug the drive to quit.

Web 2.0 technologies, particularly SVG & AJAX, fill two big gaps in the web model, and thus open the door to Web 2.5, wherein web services run from your pocket, wherever you are, whenever you need them.

Wednesday, December 21, 2005

AoyW Pocket Servers

Web 2.5 services should run from almost any mobile device, but the ideal host is a wireless pocket server—something the size of a flash drive or credit card. Unfortunately, wi-fi doesn't allow for a battery-powered access point, which is what the pocket server should be in a wi-fi context. Bluetooth hasn't caught on as a replacement for USB, and a high-rate WPAN, based on some form of "ultra-wideband", has yet to escape the demo theater. That's why we haven't seen more of these guys.

There are two candidates, however, one a research project at Intel, the other a new offering from startup Realm Systems...

The Intel Personal Server is a Bluetooth gadget which has not yet escaped the laboratory. Its software is web-based, with some extensions, like UPnP and Windows file service. Its designer, Roy Want, is a disciple of the late Mark Weiser, visionary of ubiquitous computing. (I also count myself as adherent of guru Weiser.)

The (steel yourself for the heavy-metal website) BlackDog is a USB-connected pocket server. You plug it in and the host PC instantly fires up,, X Windows. (I am frankly embarrased to have mentioned the term in public.) I guess these guys just haven't gotten out on the web much lately. The OS on board is Linux, so X was an easy path to desktop apps. The connectivity between the X apps running on the unit and the X terminal running on the PC can should be used to link a web server with a browser on the PC.

Since wireless pocket servers are still rare, we're building airWRX to run directly from an ordinary flash drive, or on a wi-fi handheld, or (if you're the type who totes a laptop everywhere in standby) in the background on a laptop.

Thursday, December 15, 2005

Web 2.0 ASPs - The Emperor's New, New Clothes

The Silicon Valley hatchery is suddenly bursting with Web 2.0 ASPs (which I define as a service that hosts customer-generated private data, as well as logic applied thereto). They're hacking RoR & Javascript and scratching napkin—er, business—plans, while plotting the demolition of the desktop software cathedral. They wisely eschew the term 'ASP', embracing the more mysterious mixed-case palindrome, 'SaaS'. Is there anything really new under their chic 2.0 threads? Let's have a look...

New User Experience! Yes, AJAX lets you build a more responsive UI than DHTML. Flash and Java have also enabled flexible UIs for some years.

New Revenue Stream! Internet advertising is booming, no doubt. Google spends some $2 billion annually to place ads on other sites. But will you hand your customer's private, 'secure' content to Google for keyword scans? Will users click through often enough when they're trying to get work done?

New Bold Approach! Can the reinvention of Word, Excel, and Outlook really be regarded as innovation? Hasn't Linux made it clear that a cheaper, more manageable desktop replacement just isn't that intriguing?

New Market Leaders! Well, two, anyway: and WebEx. Together they account for 40% of the SaaS market. WebEx could also be regarded as a network service, whose software belongs in the cloud. Salesforce has profited from the spectacular flame-out at Siebel.

New Exit Strategy! Yahoo, Google, and Microsoft are expected to fill the role of the dot-conned Nasdaq, buying everyone in sight at high multiples of founder fervor.

So much for what's new. It appears to me that most of the Web 2.0 ASP proposition lies is in what's old—those virtues that Web 1.0 ASP hyped to no avail...

No software to install or upgrade. Great, but the era of manual PC software maintenance is passing. Local-code-on-demand isn't exclusive to browser-based apps.

Simple, effective web UI. There's no question that this is a huge benefit. Multi-format pages interlinked to form sites, viewable by multiple users, is a great way to organize and present content. In fact it's too valuable to remain an online-only mechanism much longer.

Accessible via any browser, anywhere. Anywhere the net connection is reliable, that is. What if you could put the web server in your pocket? Would you prefer its service to a network one? Is there any reason besides the previous two that this app should be online? If not, then Web 2.0 ASPs will be just as popular as their Web 1.0 forebears.

We might as well call this ASP 1.1

Monday, December 12, 2005

NYTimes: Can This Man Reprogram Microsoft?

Steve Lohr piles onto The Network Is The Computer bandwagon, describing Ray Ozzie's new role. Here's the note I sent him in reply:

The sudden strength of Google and successful IPO of have yanked attention away from lessons learned by dozens of ASPs after the bubble burst: that there are hard reasons why users want to control their own data and apps.

First, individuals reflexively dislike the idea of others owning their data, even if access is free. Next, electricity is reliable and ubiquitous, otherwise we couldn't depend on PCs. Internet access is some decades from being as prevalent as AC, especially because much of it will be wireless, unlike electricity. The internet isn't really "always-on".

Complex enterprise apps like CRM & ERP benefit from having an "on-demand" service offering so customers can get started without investing huge sums before seeing returns. Eventually, many of those customers will choose to move such systems in-house, if they find them to offer competitive advantage.

Personal and workgroup apps aren't complex or expensive to set up. Laptops or flash drives can be "always-on-you". The hidden reason why people wish for web-based personal apps is the web's info-model: multi-format pages interlinked to form sites. It's a much better way to organize and present data than folders-with-files-edited-by-apps.

You don't have to hit an internet server with a browser to use the web info-model. You can carry a web server with you. The real disruption coming in this decade is "distributed web 2.0", in which user mobile devices become web servers.

Friday, December 09, 2005

AoyW is Distributed Web 2.0

As handhelds get more powerful, and laptops lighter, and flash drives denser, something strange and wonderful will happen... the web will colonize them. Not as clients, but as servers! If centralized web services are cool, then distributed, mobile ones are a snowstorm; a shower of unique, lightweight systems, each tailored to its bearer.

Pocket web servers bring the advantages of the web information model into everyday computing: multi-format pages, interlinked to form webs; presented on multiple screens, in varying modes, to one or many users. But unlike cloud servers, you are never disconnected from pocket webs; noone can cut you off from your tools & data.

Software on the distributed web will be a service, in many cases. Pocket web servers will typically belong to two or more services (e.g. one public, and one private to your business) which provide software, backup, synchronization of shared data, and support. These services will also have centralized features, like semaphores for shared resources, and gateways to enterprise systems.

Pocket web apps can do tricks that cloud web apps can't, due to internet latency. They can handle all user input with one package of code, and do it fast; exactly what PCs do today to create responsive UIs. Pocket web apps can do tricks that PCs can't, due to PC architecture. They can deliver customized views of a page to a set of participating clients; exactly what web application servers do today.

The distributed web is the best of both worlds, and it's the natural site for some of the apps now being written for the cloud using AJAX. Office apps are much more alive in your pocket.

Monday, December 05, 2005

AoyW is the Personal Web

The web model has not yet penetrated an area that is arguably the most important in computing: personal and team authoring, the focus of the personal computer. We're all still banging in MS Word and hanging docs on file servers, or shoving them around in email. The web information model, with its interlinked, multi-format pages, and multi-user views, is missing in action. Why?

Because the focus to date, from ASP to AJAX, has been on pushing personal computing up to the web, rather than bringing the web down to the personal computers. And what's wrong with pushing up?

First, you don't get a fully responsive user experience from web services in the cloud, AJAX front-ends notwithstanding. Think about dragging a slider to set an effect level on a photo or video clip: with a web service, the effect is generated on the server, which sends a generated jpeg to the browser for every mouse move you make. Never mind how fast your broadband is, Internet latency chokes this UI. On a PC, mousing is caught immediately and routed to the effect code, which makes pixel-by-pixel revs to your display. Siting all the UI code near the data source is also a lot more sane than breaking it into client & server modules. (Wasn't the web supposed to spare us from client/server headache?)

Next, the religion of The Network Is The Computer has indoctrinated the congregation to ignore an obvious issue with connectivity: We depend on PCs only because electricity is ubiquitous and reliable. If the power is out, you've usually got acts of god in progress, and little divine inspiration for work. Only when net links reach powerline quality can we trust all computing to the heavenly cloud without real fear of downtime. I hear the choir cry, "How far off can that be? Wireless internet everywhere (via Wi-Fi & WiMax) is nigh!"

Has noone noticed that electricity is wired?! AC supply is so dependable precisely because it is not wireless! Really dependable wireless internet might take a while, like a decade or three. Wireless is hard.

There's a simple, cheap, effective way to pull your head out of the cloud and bring the web revolution home, without sacrificing always-on accessibility: Put it in your pocket! On a web 2.5 server swinging from your keychain. Hmm, that sounds like a USB flash drive... Bingo. Always-on is sexy. Always-on-you actually works.

Welcome home, web 2.0! I don't know how we got along without you.

Thursday, December 01, 2005

We Have Axes to Grind

This byte-stream covers the nascent world of mobile web 2.0 servers, as seen from one land-mass thereon: the pocket application server known as airWRX. We'll note developments all over the pocket-web planet, but make no promises to be objective. (You have been warned!)

This is the first post; April 2005 is the April Fool's Day archive.