I was part of the panel on on-demand computing at TiE Bangalore's event. This is the presentation I made: PPT - PDF.

Atanu Dey discusses two truths in his inimitable style:

- Incentives Matter: Given the right incentives, people who are bound to the Wheel of Becoming, who have not yet gone across, behave accordingly. If you want them to be good, then reward them for good behavior and they will behave well. If you want them to stop bad behavior, then give them negative reinforcement and they will reduce their bad behavior. The former is called the carrot approach and the latter is called the stick approach. That's about the sum of it: carrots and sticks.

- Markets Work: Seemingly enlightened behavior emerges from apparent individual selfish actions.

Fred Wilson blogs about business models:

I think the action is in the advertising platforms and meta data and content distribution systems that are growing up around bloggings and RSS. That's a mouthful, so some examples might be helpful:

Advertising platforms - Adsense, ContentMatch, Kanoodle, AudienceMatch

Meta Data - Technorati, Blogpulse, Findory

RSS/content distribution - Feedburner, Bloglines, Newsgator

These are just examples, not anything near a definitive list, and certainly not a "best of" list.

I believe that content creation via blogging is not a business in and of itself for almost everyone. It's a labor of love, a hobby, an outlet, a networking exercise, or an extension of a brand.

I believe the software business, like all software businesses, eventually turns into a feature war and gets commoditized. The hosted software business, TypePad for example, may well avoid that fate due to the lock-in it has over its customers by virtue of URL, RSS feed, template, layout, etc. I think that it is in many ways a hybrid between a software business and a data business.

There's money in blogs, for sure, and I think the big bucks will be made in scalable platforms that leverage all that is happening in the blog world.

Innovation Weblog points to a post by Nick Dufill on 3 basic types of mind maps: library, presentation and "tunnel timeline" maps.

Bob Cringely writes that we are ready for Internet appliances and wonders where they are.

An Audrey, nee Ergo Audrey, nee 3Com Audrey, is, or was, an Internet appliance -- a small computer intended solely for web browsing. Several such boxes appeared in the 2000-2001 era, though none to my knowledge are still manufactured. And that's a shame because the world is finally ready for Internet Appliances just at a time when the computer industry is darned sure it isn't going to make any more.

If you saw my "Nerds 2.01" show, you saw Oracle's Larry Ellison ranting about how stupid PCs are, with their hard and floppy disks, and what most of us really want is a simple appliance for doing e-mail and web browsing with storage and all software housed somewhere on the Net where Larry's kid would be in charge of keeping everything current. So many people saw that show (Bob claimed) that a little Internet Appliance industry came into existence with products coming from vendors like Compaq and 3Com. Only nobody bought them, simply because they were too darned slow.

But that was then and this is now. Now we have broadband in a large percentage of homes and we have cheaper memory and faster/cheaper processors. The irony is that four years after they died, the market is just screaming for such products to extend the Internet to the final few places it doesn't yet belong, like my mother-in-law's house.

She had a computer, but never noticed that it had been fried by a power surge. When we arrived for Thanksgiving it was still "on" but dead, the monitor an unblinking eye. She has an e-mail address, but doesn't check it. She wants to make airline and hotel reservations over the Internet, but instead does it over the phone to my wife, her very own Internet travel agent.

This woman needs an Audrey. Smaller than the PC it will replace, Audrey can sit on her desk next to the telephone answering machine. When she gets an e-mail both a big button and the translucent stylus will glow and blink. There's a touchscreen (no mouse), and the little keyboard is wireless and can be hidden in a drawer. You can also mount your Audrey on a wall. It would be the perfect device for a telco to bundle with some entry-level DSL service, don't you think?

CRN has an interview with Michael Terner, CEO of KnowNow. Excerpts:

We put in an RSS solution for ING, one of the largest banks in Europe, that allows them to utilize, manage and control RSS within their corporation. They have over 100,000 employees, and they've been using IBM WebSphere portal to distribute information to employees. They also have news feeds they wanted to distribute to employees through the portal. You can get a portlet from IBM for RSS, but today it only handles one feed. Of course, they wanted it to be able to handle multiple feeds, and they didn't want to install code on the client. Finally, it also helps them address a key problem that is emerging from the deployment of RSS. With users doing updates every hour or every four hours, you're bombarding the infrastructure internally. With our server, we deliver information only when there's new information. It eliminates all of the pulling requests. KnowNow remembers who you are and what you're interested in and keeps the link open in such a way that there's no traffic on the lines. It's a publish-and-subscribe model.

You can build an application that doesn't require software to be installed. E-mail and Excel are an integration problem that hasn't been resolved yet. We built an application that is on our Web site that looks like an RSS reader. We set that thing up so the headers change color as it ages. Because we have a publish-and-subscribe model, management can view activity such as responses to incoming calls by simply being a master subscriber. It requires good business partners to be able to build these applications, but it doesn't require a complex application developer because there's no client code to install and most of the work of getting the information moved around is being done by the system. In many ways, this allows you to treat the data sources as a service for the client, whether it has software running on it or not. So you enable that whole software-as-a-service model.

There are two types of networks which are relevant in the context of inter-connecting the network commPuters. The first is the LAN, and the second is the WAN. In the local area, Ethernet holds sway globally. In fact, even in the emerging markets, the state of the LANs is nearly as good as the developed markets. With 100 Mbps and 1 Gbps LANs, there is plenty of bandwidth to service the needs of the access devices in enterprises, educational institutions, government offices, and the like. Where there is a huge lag is in the wide area network. More specifically, the issue is in the connectivity available on the last mile.

Let’s take India as an example. There are two possible connectivity options for Internet (and grid) access at present: via the telephone line, and via the cable connection. Both will not go far in India, and that is where the short-term for real broadband looks bleak. With BSNL and MTNL controlling most of the 40+ million local loops and the government deciding not to unbundle it, we are at the mercy of the two (soon to be merged into one) government telcos to provide us with high-speed xDSL connectivity. (High-speed does not mean shared 128 or 256 Kbps access, but genuine multi-megabit connectivity.) The cable companies are doing their best to fill the gap, but are hampered by the quality of the cable, other technical challenges which hinder two-way access, and the investments they can make.

What the “singleton” network commPuters in India need is broadband wireless connectivity. We need to leapfrog to next-generation networks which can blanket urban and semi-urban population clusters and deliver high-speed (512 Kbps-2 Mpbs) connectivity over the last mile, and then hub these connections into fibre back-haul networks, which already exist in plenty across India.

These broadband wireless networks can come via multiple technologies and providers. IIT-Madras has developed a cable wireless technology, which uses cable for the downstream connection (1 Mbps) and wireless (via CorDECT) for the upstream connection (100 Kbps). Mesh Wireless networks could use the unlicenced WiFi spectrum to provide last-mile connectivity. Then, there is WiMax. The mobile operators could roll out 3G (and perhaps, 4G) on their networks. Considering that voice revenues are likely to flatten (and even fall, with VoIP), the mobile phone companies in fact need to think seriously about getting into data networks.

So, even as the short-term outlook for broadband remains bleak in India, I am optimistic in the long-term (2006 and beyond). The broadband wireless revolution in India will happen. The government will, however, need to be forward looking on making spectrum available for some of the technologies. The Indian telecom companies and ISPs do have the financial strength (with assistance from international investors and partners) to build out an always-on, high-speed and ubiquitous network. The New India has to built, literally, out of thin air.

Tomorrow: Infrastructure