NabaTech

Google Subscriptions Someday?

I guess online businesses can make money three ways—transactions (buying/selling something), advertisements (pay per click), and subscriptions (monthly fees). What amazes me is that the pay-per-click advertising model works so well for search engines like Google. As for me, I ignore all the advertisements and only make use of the search results, just as I ignore all those Google ads embedded in most blogs and websites thanks to the Google AdSense program. When it comes to ad-clicking, I must be in the minority because Google is doing fine with the pay-per-click advertisements. Nevertheless, I was wondering if Google would someday introduce a subscription-based premium search service of some sort—not to replace its ad-based model, but to augment it. I don't mean the "ask a question, set your price" model of Google Answers, but a straightforward subscription-based access to some premium content. Sometimes, when I have trouble finding relevant material from Google searches, I almost wish for a subscription-based Google search. I think I'd go for it if the subscription fee is not prohibitive and if I could try out and see that the searches produce better results.

Tags: search advertisement Google google transaction subscription payperclick adsense

Blog Search Engine—Technorati Beta

Visitors often reach blogs through search engines and mine is no exception. The only surprise, at least to me, was that when it comes to searches that landed someone on my blog, it seems to be about 50-50 between Google and Technorati. I was surprised by this because I use Google extensively, but I had not realized so many people use Technorati as well. I had tried Technorati a few times, but gave up because it was too slow. Then I heard about the new Technorati beta and tried it out. The beta site looks better than the old site, but the response time is mixed—I had good response over the weekend, but the site seems sluggish during the workday.

Search for a keyword or a tag at Technorati beta and you get all the recent matching blogs along with supposedly related photos from Flickr. I found the tag search results more relevant than the keyword search, probably because blog posting tags are more relevant than whatever the keyword search is picking up. The Flickr photos don't typically relate to the search results, so I don't know what value they add. Displaying the most recent blogs seems like a good idea, but I wish Technorati would also display those blogs that are considered authoritative for a keyword—the problem is figuring out the "authoritativeness" of a blog; they have to figure out some kind of ranking system and number of links alone would skew this to the top few blogs. Maybe Technorati should add a "feedback/rate this" link to each search result so that users can give feedback and then factor that (along with number of links) into calculating what users find most relevant. Overall, I think I'll continue to try Technorati along with Google and also send them my feedback for the improvements.

Tags: tags tagsonomy Technorati technorati blog rank search flickr google feedback rating beta

Can MDA Play a Role in Global Sourcing of Software?

I like the idea of model driven architecture (MDA) as a software design and development tool because in involves modeling the software and then generating code with the help of tools. I don't know if MDA is widely used yet, but at least anecdotally many seem to consider MDA important. Software tool vendors seem to claim support for MDA although the support may not always cover the full OMG MDA specifications. One aspect of MDA that intrigues me is the potential for using it as a way to assure that code conforms to design when software development is outsourced, whether within the country or offshore. Here's the idea.

The in-house design team develops the platform independent model (PIM) for the software based on the requirements. This would be a creative step that requires experienced software designers and experts on MDA. Then the in-house team provides the PIM to the outsourced development team who prepares the platform specific model (PSM) and checks out the design by generating executable code from the PSM (I assume that an automated tool is available to generate code from the PSM). The PSM is then delivered back to the in-house team, who reviews the PSM and then uses the automated tool again to generate the executable code, run any static analysis tools to check the code, put it under change control, and give the code back to the outsourced team to do quality assurance. The idea is for the in-house teams to assure themselves that the design conforms to the requirements (after all, an in-house team designs the PIM) and that the PSM-to-code step is performed under in-house control. I think using MDA in this manner might be a good way to use outsourcing without relinquishing the key software design steps and mitigate any potential risks that code does not fully conform to the design.

Tags: mda uml model PIM PSM OMG softwaremodel outsourcing offshoring risk

Model Driven Architecture—What is it?

Although Model Driven Architecture (MDA) has architecture in its name, MDA is not at the same level of abstraction as Enterprise Architecture (EA), rather MDA is a software design methodology. MDA was developed by an industry consortium called Object Management Group (OMG) as a standard approach for modeling the functionality and behavior of distributed software systems, independent of the computer platform—operating system, programming language, and middleware—on which the software is eventually implemented. MDA is based on OMG’s Unified Modeling Language (UML), a diagramming notation for modeling parts of software and their interactions. Here's a figure illustrating the concept of MDA:

The primary output of MDA is a Platform Independent Model (PIM), which represents the software design using UML. The modular structure of the software as well as any computation logic can be modeled by using an extension of UML called executable UML. This platform independent model can then be converted to a Platform Specific Model (PSM) that’s also written in UML, but that expresses the dependence on the specific operating system, programming language, and middleware. The final step is to automatically generate the software’s code (for the target middleware environment) in a high-level programming language such as Java or C++. In short, MDA can go directly from the functional and behavioral specification of software to executable programs. If the hardware and software technology changes, the software can again be regenerated for the new hardware and software technology by starting with the PIM and going through the PSM to code. Thus, MDA is supposed to isolate the functional behavior of software from any changes in the underlying technology. At the OMG's website you can read more about MDA as well as check out the links to some MDA success stories. MDA is definitely a step in the right direction for software design and development, but it's not all automatic—someone still has to do the hard part of designing the PIM in the first place.

Tags: MDA mda UML uml model PIM PSM OMG executableuml softwaremodel EA

Another Ebook Benefit—No Returns!

Last Friday's Wall Street Journal had a front-page article on the insidious problem of "returns" in the print book publishing world. Because book retailers can return unsold books, even successful titles can end up with tons of returns if the publisher overestimates the demand or if the retailers order too many copies. The WSJ article says publishers end up having to raise prices to compensate for the lost revenue due to returns, which makes books harder to sell (too expensive) and leads to even more returns. You get the idea—"returns" are causing a big mess.

Even if print books are not going away any time soon, perhaps we can lessen the problem of returns for tech books by moving to ebooks. With ebooks, there's no need for returns; nor any need to waste all that money and paper upfront and ship boxes of books around the country in the first place. I'm sure you can think of many reasons why ebooks may not sell—no good reader device;  tried it, didn't work; and so on. Well, we don't need a ebook reader device for reading tech ebooks on computer and perhaps the current prices are too high to generate sales of ebooks. We can't expect people to shell out $30 or more for an ebook that's simply a PDF file version of a print book. I think an ebook has to be priced much lower. If this requires reducing the size of the ebook (and focusing on a narrower topic), so be it. And while we're at it, we should redesign the ebook to take advantage of the electronic format; for example, use a layout that's easier to read on computer monitors, and include multimedia that provides a richer experience than a print book. For a 75- to 100-page ebook packed with information, a price point of $5 or so sounds right. On the other hand, the same size ebook with  embedded multimedia could be priced $10 and up. What I don't know is whether such a pricing model for tech ebooks would work for mainstream publishers or are such ebooks more the domain of self-published authors or small niche publishers. I guess we'd find out as the market evolves!

Tags: books ebooks epublishing returns trends WSJ

Preview of Latest Microsoft Start

I had tried out the two versions of Microsoft's online RSS aggregator/reader—http://www.start.com/1/ and http://www.start.com/2/ a while back and now there's a third preview version of Microsoft Start that combines various features from the previous two versions. To try the preview version, go to http://www.start.com/3/ and answer the five questions to get the new version's URL (don't worry, if you don't know the answers, you can click the search link to get each answer from MSN Search). Here's a typical screen shot of the new Start:

When you first go to Start, the initial RSS feeds appear organized in blocks in a dashboard (you can toggle the dashboard on or off by clicking the link in the upper right hand corner). Click Show next to Start in the upper left corner to see the feed categories. Click the arrow to the left of My Feeds to open it up and then click [add feed] to add your own feeds. After typing in the feed URL and clicking Add in a dialog box, you can pin that feed to the dashboard. You can drag the blocks by their title bar and move around to rearrange the dashboard. When you click a link in a block, a few more lines of that item appears in a popup dialog box. The new version's user interface is a mixed bag for me. I wish that clicking an item expanded it within its block in the dashboard instead of showing another popup. Clicking more... in the expanded view of an item brings up that item's full content in a separate browser window. I don't know what a good solution is, but when I click a link, I would rather have the content appear in the dashboard area and then I can click Back to return to the dashboard again. I think they could divide the box on the left hand side into two scrolling vertical panes and make use of the lower pane as a placeholder for the current dashboard when that area is showing an item's full content. They could offer a link on that lower left pane to revert to the dashboard view again. Anyway, I noticed a feedback link at the bottom of the window—I guess I better click that link and give my feedback directly to the development team!

Tags: RSS aggregator Microsoft start trends MSN

Microsoft's New eHow-tos

I thought adding animations, if not full videos, would be a good idea for computer ebooks. As it so happens, Microsoft Developer Network (MSDN) has recently introduced what they call eHow-tos—multimedia documentation on various Windows software development topics. They are meant to be played in Windows Media Player, but the website says other JavaScript-capable media viewers would also work. I tried out several eHow-tos in Internet Explorer; for example, try the one on Creating a Bluetooth Application and you'll see how an ehow-to works. It's more of a video tutorial, but you see the code in the background screenshot with a smaller image of the speaker in the foreground, which seems quite effective. If you don't want to bother playing the eHow-to or don't have a suitable media player, here's a sample screen photo:

As the speaker explains the code, you can follow along in the screenshot in the background. The speaker can highlight parts of the code, point out specific items related to whatever is being discussed. When I was talking about interactive animations in computer ebooks, I was assuming simple animations, perhaps just the screen being captured with the mouse pointer highlighting code or pointing to specific parts as the ebook's text explained the concepts. I think full video with audio narration would be a bit much for computer ebooks, which may not be played in a full-blown media player; besides the file size would get bloated. Nevertheless, Microsoft's eHow-tos are an example of the new trend towards online multimedia tutorials and they're fun to watch—if you are into any kind of software development :-)

Tags: books ebooks epublishing multimedia tutorial ehowto microsoft msdn

Enterprise Architecture—Part #9: EA Integration

After you have considered an “enterprise” as a specific “business undertaking”—a set of related business activities aimed at some common goal, viewed the overall enterprise architecture (EA) of an organization as a collection of EAs, and developed the individual EAs, you can focus on EA integration. This step is aimed at ensuring that the separate EAs do not result in duplicate data or duplicate IT systems supporting different enterprises within the organization. As an example, consider the following diagram illustrating the EAs for several "enterprises" such as financial management, customer relationship management, human resource management, and the core enterprise (this is what the organization really knows how to do well):

The diagram shows the component architectures—business, data, application, and technology—for each enterprise. The two areas where the potential for EA integration should be most evident are the data and technology areas. Even though you might use commercial off-the-shelf (COTS) IT solutions to implement some of these EAs such as customer relationship management and human resource management, you would want to ensure that any data such as customer information or employee information is not duplicated. That would be an example of EA integration at the data level. Similarly, you would want to consolidate the hardware and software platforms and the network so that all EA implementations can share the infrastructure. The exact details of how you achieve EA integration depend on the details of the EAs and the strategy you want to employ. For example, you could have common data in a single database that multiple applications access or you might decide to define Web Services that manage the data and any application that needs the common data access it through specific services. No matter how you accomplish it, the key point is that after developing of the component EAs for an organization, you need to go through the EA integration step to ensure that you don't end up with what we call "stovepiped" systems and business practices.

Tags: enterprise architecture EA integration COTS stovepiped SOA service

Enterprise Architecture—Part #8: Technology Architecture

By now you have worked on the business, data, and application architectures. Next comes the technology architecture that describes the computing infrastructure—the computer hardware, software, and networking platforms—that would support the applications, that, in turn, enable the business activities. Undoubtedly, the technology architecture would be influenced by the latest IT trends, but the goal is not to focus solely on the technology, but to select technical solutions that can support secure and robust applications in a timely and cost-effective manner. In today's internetworked world, the enterprise network is at the heart of the technical infrastructure. Other elements of the technology architecture include hardware (mainframe, servers, storage systems, desktop PCs) and associated system software such as operating systems and databases. The technology architecture serves as the blueprint for the infrastructure and it puts forth the standards and specifications for its components.

At the highest level, you can document the technology architecture with diagrams and narrative descriptions. The diagrams invariably look like network diagrams. For example, here is a typical example:

This sample diagram shows a generic mainframe reservation system and some internal servers on a LAN that connects to the Internet and a WAN through a firewall. Users access the reservation system over the Internet. The internal systems access other partners' systems over a wide area network (WAN). When you draw the technology architecture diagrams for your organization, you should annotate the diagram with specific geographical locations of various systems as well as indicate the hardware and software details of different platforms.

Tags: enterprise architecture EA technology network LAN WAN IT

Lamenting the Size of Computer Books

I still have my copy of Kernighan and Ritchie's The C Programming Language that I bought around 1980 when I first learned C (transitioning from FORTRAN), on a VAX-11/780 running Berkeley Unix that we installed following instructions outlined by Bill Joy (Sun Microsystems wasn't incorporated yet). Kernighan and Ritchie's paperback book, published by Prentice-Hall, was all of 225 pages or so, yet it was packed with everything I needed to learn C. Kernighan and Ritchie subsequently had a second edition of the book in 1988, updated to ANSI C, and it's still only about 275 pages long. I continued from C to C++ and when Addison-Wesley published Bjarne Stroustrup's The C++ Programming Language in 1985, it too was only about 325 pages. Then something happened and computer book sizes jumped. In fact I was part of that trend—I wrote the 800-page The Waite Group's Microsoft C Bible in 1988. Of course, today the majority of computer books (at least the ones dealing with programming languages or operating systems) are typically close to a thousand pages in size!

I think I know the reasons why computer books end up so large. First, the subject matter has grown in complexity, so we need more book pages. For example, C relies on libraries to do all the hard work and once you start describing the libraries, it's hard to keep the book's size down. C++ also has standard libraries and when you describe them, guess what, the book gets larger. Even Stroustrup's third edition of The C++ Programming Language grew to 920 pages. The second reason is that it's not economical to sell smaller books at lower prices—I am not an expert on this, but it seems the numbers just don't work out. A third reason is the book buyer's expectation of getting a certain number of pages at a certain price. This is related to the reason that smaller print books are not economical. If there are a whole bunch of books on the same subject and all priced at $49.99, the feeling—right or wrong—is that the book with most pages would be perceived to be a better value. Even with all these reasons for selling the huge tomes, I still wish we could move to 300-page computer books that focus on narrow topics and offer good value to the book buyer. Perhaps computer ebooks will help us reduce the size and offer better value than print books.

Tags: books ebooks epublishing trends size C C++ Kernighan Ritchie Stroustrup Joy billjoy