[Fri Nov 28 19:43:55 CST 2003]

Paul Thurrott writes a very interesting article about inductive user interfaces that manages to paint Microsoft in a very positive image, something we are not every used to. I suppose that, just as in the world of politics, image counts when it comes to technology too, and while we always assume that Apple is the greatest innovator in GUI design, perhaps the reality of their achievements is not so rosy. As Thurrott explains,

Apple has done very little to make its UI better per se, beyond simple enhancements to what is, again, a classic desktop OS. There's nothing wrong with that. Frankly, Apple's crowd is technical enough to deal with it. But saying that such an OS is "easier to use" or "more elegant" than Windows is wrong. Mac OS X is attractive, and arguably "better looking" than Windows XP, though that's a subjective declaration. But it is most certainly not "easier to use". And that's not "Apple bashing," it's just the way it is.

Believe me, Thurrott is not full of it. He makes a compelling argument to show us how Apple follows an applications-centered metaphor where one needs to know what application to run in order to perform a task (e.g., launch iPhoto in order to print a picture from our digital collection). Certainly, MacOS X is a very attractive and aesthetically pleasant implementation of this paradigm, but that still does not change its subyacent philosophy which is that of the classic GUI. Windows 95, 98 and XP already tried to come up with something new: a file-centered (or document-centered) approach to computing. Here, if one wants to print a picture the only thing that is needed is to open the My Pictures folder and right-click on any of the image files to select the Print option. Some might see this as a step in the right direction, but it still assumes that the user knows about the underlying filesystem and how files and folders are laid out there. Windows Explorer or a similar navigation tool is at the core of this model. However, where Microsoft is truly innovating is with Longhorn, which signals a clear switch to a new metaphor, that of a task-centered approach. Here is how Thurrott explains the new model:

You don't worry about the file system. There is a special shell folder/collection/Library (whatever they decide to call it) in Longhorn that aggregates all of the photos on your system automatically, and instantly. You can filter the view by various criteria (who took the picture, when it was taken, what it is a picture of; in other words, metadata), making it very easy to find. After that, there is a similar task-based approach to XP for actually printing the picture.

There is something to say for Microsoft's Longhorn. Sure, until they release, it is only vaporware. Yet, there is little doubt according to what we can read here and there that they seem to be coming up with something quite unique and creative almost for the first time since the company was founded. It is just fair to acknowledge that much.

[Mon Nov 24 15:41:49 CST 2003]

I read in Network Computing that the MIT has decided to reject all email messages that contain executable attachments. The measure may sound a little bit harsh, but I cannot tell how many times I had to ask to relatives and friends why they should never open these attachments. Needless to say, they also happen to ignore that .exe is not the only filename extension that Windows executables can use. Just in case, this announcement contains a nice list of all dangerous three letter filename extensions.

[Wed Nov 19 20:40:54 CST 2003]

Now, here is a surprise for all those who believe that all technological innovations come only from the USA. Chinese researchers are working on a new standard called Enhanced Versatile Disc (EVD) that promises five times the image quality of the DVD. The new standard, which may become available to consumers next year, will also allow a higher data storage capacity. It makes one wonder how long before the dragon wakes up and becomes a serious competitor. But, is there any chance that the new EVD standard will spread all over the world? Only time will tell, but I seriously doubt it. In the technology field, as in many others, money and marketing are just as necessary as quality.

[Wed Nov 19 20:33:38 CST 2003]

Not long after the company made the decision to discontinue its own low-end Linux distro, Red Hat has surprised everyone announcing that the renewal rate for support subscriptions of their high-end product reached more than 90%, when they had just planned it to be about 75%. It seems as if Red Hat is posed to become the first truly viable open source public company, although we should not forget that there are many private companies already making money in the field (MySQL, Progeny, Trolltech...). I still think their decision to turn away from the desktop market may leave their flank open to attacks from the competition, especially from the Novell/Ximian/SuSE conglomerate which certainly has the potential to develop an attractive business desktop product.

[Wed Nov 19 20:21:52 CST 2003]

SCO continues its lunatic legal race towards profitability. Darl McBride just announced that his company is comparing its own source code to that of the different BSD flavors, and will probably file a lawsuit against them "in the first half of next year". What SCO is doing may be legal, but there is something appallingly dishonest in a company that did not innovate, a company that has not given us a single thing worth mentioning, and whose main source of revenue consists in filing lawsuits against other people. Right when most BSDers thought they were safe from Mr. McBride's wrath thanks to the settlement between AT&T and the University of Berkeley back in 1994, here comes SCO to drag them back to the mud. It is disgraceful.

[Tue Nov 18 10:16:12 CST 2003]

It is not easy to find examples of company that managed to make a big comeback. Apple is, without a doubt, one of them. IBM is another one. It is not as if IBM was never in peril of extinction, but during the 1990s there was little innovation coming from Big Blue. However, things have changed quite a bit in the last three or four years. They have taken the lead in the Linux front, and now IBM has announced the prototype of some revolutionary storage system. Using what looks like ice cubes (see the article for a diagram of their design), their engineers can build a stack of Lego-style data-storage systems that provide not only better throughput but also better fault tolerance. When the system is finished, increasing the capacity will be as easy as adding more cubes.

Expectations are high for IBM's storage cubes. Users will be able to add bricks as needed to increase the storage they need, and software will automatically generate a single system image of data on the cube. Despite the highly dynamic and distributed nature of the physical storage, data will be replicated across disks as needed for fault tolerance and to allow it to be accessed within acceptable latencies.

The system will also automatically compensate for any hardware failures without user intervention. In fact, the philosophy behind the system is to allow failed modules -- which might represent 10 percent of a system's modules over five years -- to be simply left in place.

About half of all repair actions in a data center create other failures, which in turn require repairs. Under the brick approach, modules with hardware failures are simply left for dead in the cube. "A central tenet is you leave it alone," said Wilcke.

Of course, the main challenge is to write the software that will be able to interoperate with all this but few people doubt that IBM will manage to figure that one out too.

[Tue Nov 18 08:12:20 CST 2003]

Talk of an about face. Sun has announced an alliance with AMD to support Solaris and Linux on the Opteron processor. Of course, Sun being Sun, they also talked about an optimized Java platform that will run on the new products. AMD's Opteron is a 64- bit processor that can still run 32-bit code natively, and therefore poses an alternative to Intel's Itanium chips. It seems clear that Sun, unlike other companies like SGI, has seen the writing on the wall early enough to at least try some alternative strategies, but it is still not so clear whether they will manage to turn the ship around. One does not know to be too clever to realize that commodity processors are slowly taking over the supercomputing market too, rendering competing architectures nearly obsolete. Sure, MIPS, SPARC, PowerPC and other architectures still offer certain high-end functionality that is missing from Intel's or AMD's offerings, but it is just a matter of time before they catch up and the sooner we realize that the better.

[Mon Nov 17 11:29:44 CST 2003]

It seems as if the world of browser development is not dead after all, and we owe it mainly to the Mozilla project. I read somewhere that Microsoft's Internet Explorer will include a feature to kill popup windows in its next release (sorry, I could not find any link to the story now), and we all know where that idea came from. But this is not all, one is hearing again some level of excitement among web developers and end users about the potential of certain browser features such as tabbed browsing, support for RSS and collaborative tools that erase the distinction between viewing and editing online documents. It turned out that both the Mozilla project and Apple with its decision to launch Safari, an alternative browser based on KDE's KHTML rendering engine, have managed to stir things up a little bit. Now we only need that the same healthy competitive environment also expands to the other fields, such as the desktop systems or the office suites.

[Mon Nov 10 11:10:29 CST 2003]

I read that Google has introduced a desktop application to enable searches without using a browser. The idea is certainly not new, and it somehow reminds me of Sherlock, the application included with MacOS X that not only allows web searches, but also provides information about stocks, flights, eBay auctions and many other services. It seems as if web services are finally moving beyond the drawing board and into the real world, but one minor question remains in my mind: how come the open source community is not developing similar applications based on the Mozilla framework, for example? It strikes me as a nice way to develop an application that most end users will find useful regardless of their own technical expertise and, along the way, promote Linux on the desktop.

[Wed Nov 5 20:42:25 CST 2003]

Let us face it. Microsoft's Longhorn could pose a serious threat to the open source movement. Why? It is about the first time Microsoft is truly coming up with some innovative solutions. Sure, perhaps they did not invent it themselves. Still, they may be the first to implement many of these innovative ideas in real life. I am talking about some exciting stuff: a completele redesigned desktop GUI that is more modular, a filesystem based on an SQL database system, real web-like search functionality applied to the user's documents, an abstract runtime engine that allows on-the-fly compilation and interpretation of code written in multiple languages... It is time to be honest and acknowledge that Microsoft is taking competition seriously. Of all these ideas, perhaps the development of a database centered filesystem (WinFS) is the most exciting one. Built on top of the old NTFS filesystem, it will allow to store the files as items in the database, together with metadata associated to such files. The usual system and application files will still be stored the old way, but the user's documents will be stored using this new filesystem, which should allow for a much richer experience and more complex search criteria.

Now, the open source community is not standing still, and someone already started the GNOME Storage project which is supposed to accomplish the same as Microsoft's WinFS and then some. Yet, what should worry us is the fact that the open source model, while far more flexible when it comes to reacting to feedback, does not appear to allow for such quick moves and turnarounds as one can see in the commercial world. The reason should be apparent: there is no clear hierarchy, where the top of the pyramid gives an order and the whole organization automatically changes its strategy. Let us put it this way: it is like the difference between a democratic and an authoritarian societies. The former will be more open to feedback and will be more able to adapt, but the later will be able to move faster... as long as the head points in the right direction, of course. There is no reason to believe that Bill Gates has suddenly gone crazy, and he still appears to have the same nearly perfect graps of what is to come that has proven so successful in the past. On the contrary, I would dare to say that it is far more likely that both the GNOME VFS and the GNOME Storage project will simply lag behind and play catch up with Windows once again. One just hopes we do not simply sit on the side while a whole new paradigm is born.

[Wed Nov 5 20:23:53 CST 2003]

Steven Vaughan-Nichols writes an interesting article in eWeek about why Linux ended up being more commercially successful than any of the BSDs. I say interesting because what he finds as the main reason for Linux's success is precisely what so many BSDers deride about the little penguin that could: its development model.

... because of Linus Torvalds' combination of quiet leadership and a willingness to let anyone and everyone code for the project via mailing lists and Usenet. This more-open approach gathered up far more supporters than the much smaller BSD development communities. In Linux, anyone was welcome to lend a developing hand; in BSD, only the truly devoted were welcome. By being inclusive, rather than exclusive, Linus made sure that Linux avoided the problems that kept not the BSDs but all the other Unix operating systems from mass acceptance.

It is the dirty little secret about why BSD failed to gather the wide support Linux got in the past few years. Yes, the BSDs are every bit as capable and reliable (if not even more) than Linux, but (and this is something most geeks still fail to understand) technology is not all that matters. There is nothing as unwelcoming and "exclusive" as the BSD clubs. Their history is firmly rooted in academia, and it shows.

[Tue Nov 4 12:54:30 CST 2003]

Well, today appears to be a big day in the Linux news front. Novell has announced an agreement to acquire SuSE. Press releases tend to be all bright and sweet, but I have to agree with the company when they emphasize that after their acquisition of Ximian this helps Novell "become the first to offer comprehensive Linux solutions for the enterprise from the desktop to the server". They still have to prove the new strategy will indeed turn around the company, but it seems to me they have very clear ideas of where they want to go and are definitely putting all the bricks in place. It is a sound strategy, that is for sure. SuSE has some serious products both for the desktop and the server that expand even to the very high-end server market (up to 64 CPUs, which is something not even Red Hat is supporting yet), while Ximian offers a great desktop product as well as system management tools and the Ximian Connector to serve as the interface with Microsoft Exchange servers. On top of that, we have Novell's network expertise. Again, the gamble may or may not work, but there is little doubt in my mind that Novell's managers made the right move. In a matter of just a few months, the company has become one of the main Linux vendors out there and is generating some buzz again.

[Tue Nov 4 11:00:06 CST 2003]

Red Hat's Matthew Szulik has made some statements about Linux on the desktop that appear to be generating some controversy, but I simply fail to see what is so outrageous about his comments.

I would say that for the consumer market place, Windows probably continues to be the right product line. I would argue that from the device-driver standpoint and perhaps some of the other traditional functionality, for that classic consumer purchaser, it is my view that (Linux) technology needs to mature a little bit more.

Mind you, Szulik is not denying the clear improvements that Linux on the desktop has experienced as of lately. As a matter of fact, he seems to assume that it will get there sooner or later, which is something I would agree with. Yet, he is emphasizing that mainly due to the incompatibility of device drivers home users may still experience too many problems using Linux in their systems. The situation is completely different on the corporate desktop, where Linux can be deployed with little effort already and manages to provide a flexibility that other operating systems simply cannot match.

[Tue Nov 4 09:51:41 CST 2003]

eWeek published an article pitting Longhorn against the Linux 2.6 kernel. As usual, Microsoft is managing to create a lot of expectations and is even setting the tone of the debate with a product that admittedly will not be available to the public until 2005, if they meet their own deadlines only for this one time. Yes, what I hear about the product does sound interesting and even innovative. No, it does not represent a departure from Microsoft's typical tactics and lock-ins. As Jack Zahran, president of UnifiedAgent Inc. pointed out:

WinFS is actually NTFS [NT File System] with SQL Server replacing the current indexing method in Windows. XAML [XML Application Markup Language] is nothing more than another monopoly-styled move to pull the carpet from under Mozilla's open XUL [XML User Interface Language], which is already partially supported in Safari 1.1 [Panther's version].

In other words, nothing new under the sun when it comes to its usual tactics. Are there open standards Microsoft could be using to deploy in their new products? Yes. Will they do it? No. Why? Mainly because they get more from forcing proprietary technologies down the customer's throats, especially when they already own a significant portion of the market and user's only alternative is to migrate to something completely different. In spite of all this, Craig Mundie also makes a very good point when it comes to comparing Microsoft and open source:

We tell everybody what our vision is. I don't think [Linux kernel developers] tell the community what theirs is... There's also a difference between a community model that's sort of an evolutionary process and one that's actually designed and managed for effect in some specific way, which is what Microsoft is doing.

Mundie is talking about the differences between the open source and the proprietary models here, and he is right in pointing out that a lack of a managed roadmap is one of the characteristics of the open source world. The problem is that this could be considered an advantage as well as a drawback. After all, Longhorn itself represents a huge turnaround with respect to the older Microsoft products and we have barely been told about it a year and a half before it is supposedly released. I am not sure how the proprietary model has an advantage to offer in this sense, at least when it comes to the long term. I must admit though that the short and middle term are a different thing altogether, since a tight development model tends to guarantee more stability in that sense. On the other hand, the open source model offers an evolutionary approach that is far more flexible and open to the customer's feedback. Either way, more and more it looks as if the world of technology boils down to a choice between these two models with their respective advantages and disadvantages.

[Tue Nov 4 08:57:37 CST 2003]

Just a few days ago I wrote about Debian in the enterprise and the hurdles the community-based distribution might face there. Today, I came across a short piece in Ian Murdock's blog where he puts forward Progeny as the solution:

The IT industry wants to break the chains of single-vendor reliance and proprietary lock-in, whether the lock-in is based on proprietary technology or some clever new scheme. But is the only way to break these chains to bring everything in-house?

Actually, no. Progeny provides a third option: We're an outsource provider of Linux distribution maintenance that allows companies to essentially have their own Linux distributions, with a feature set, roadmap, and support model tailored to their needs rather than the vendor's, all without having to bring the distribution management function in-house.

It's a non-intuitive business model for an OS company, at least at first, because we're so used to being at the whims of our vendors. We're so used to it that the same model has carried forward from the proprietary OS world to the Linux distribution world. Now that Red Hat is turning the screws on the thumbs that have been so carefully positioned over the last several years, people are starting to realize they moved to open-source operating systems precisely to get away from this kind of screw-turning.

And now, a Cluetrain-inspired question:

Progeny's primary focus is on building distributions for what we like to call "Linux-powered products", products that are Linux-based but where Linux is just one layer of the overall stack, and possibly an invisible one at that. Lately, though, we've had a lot of folks ask if we might consider applying our customer- rather than vendor-centric approach to Linux to the more general-purpose deployment/enterprise space. Could we provide support for enterprise Debian deployments? Could we keep the updates coming for Red Hat 7.x after they are end-of-lifed at the end of the year?

This certainly sounds like a sound business plan, and it is something I have been considering myself for quite some time now. Since Fedora does not offer the same level or support that the old Red Hat Linux did, would it be possible for resellers to come in and fill the void? They could perfectly offer customized versions of Fedora Core as well as guarantee updated RPMs for a given period of time. It sounds like the perfect combination of open source technology and realistic business plan.

[Sat Nov 1 21:12:44 CST 2003]

The news that Red Hat would launch the Fedora Project and stop selling its own low-end distribution gave the shivers to many system administrators, and for a good reason. Up to that point, many companies were running a top notch Linux distribution that also enjoyed very wide recognition and had a known name behind. Sure, if one downloaded the ISO from any Red Hat mirror or bought a box and installed it on multiple machines there was no support contract... and what? Red Hat still provided free technical support in the form of a wonderful mailing list that usually answered one's questions within hours (they still maintain an open mailing list, of course) and, in any case, the distribution was so popular that one could bring the issue to pretty much any Linux-related bulletin board, list or newsgroup. Needless to say, this is all great for the end user and those companies that want to run their servers without spending much on the operating system, but its effects on Red Hat's bottom line were not so good. After all, Red Hat is a public company, and it needs to make money somehow. Simply selling support contracts did not seem to make it, in spite of having some big names on their payroll. On the other hand, customizing their low-end distribution, adding a few enterprise features and selling it as a high-end product did not appear to be the solution either, although it did point in the right direction financially speaking. So, it should not surprise anyone that the company finally decided to pull the plug on a product that did not return any significant amount of money and was not needed any more in order to gain respectability in the market or a name among the hobbyists. Let us be real and acknowledge that, under the circumstances, spinning off the low-end Red Hat Linux distribution into the Fedora Project is not such a bad outcome after all, at least when it comes to end-users. Of course, things are quite different among those who must make a decision on whether to stick to the Red Hat enterprise products in spite of a hefty price tag or bet their future on a community-based solution that does not and cannot guarantee any clear cut roadmap. I would dare to say that the issue is not so much one of stability as one of roadmaps, guaranteed length of support and the nice and comfy feeling (to managers, at least) that there is a corporation behind a product. In this sense, the comments I hear lately suggesting that Debian could offer an attractive alternative simply do not cut it. Far from what is suggested, the problems with a Debian implementation are not only limited to whether or not a given application is certified against the distribution (something that, by the way, those system administrators who intend to deploy Fedora will also have to deal with), but are mainly related to the other issues I laid out above: the community is in charge, which offers a chance for anyone to intervene and provide feedback, but it also makes it far more unreliable to make assumptions about the future strategy of the product. Nevertheless, it is also true that Red Hat's decision to switch to a more community-based approach pretty much erases all the advantages it had before when compared to projects such as Debian or Gentoo, at least when it comes to these issues that are not centered around the technology itself. Even worse, according to what I am reading, Fedora promises the end of life every 6 months or so, which is far from ideal for most system administrators out there, while Debian's stable release is well known for lagging behind but been actively developed and supported for a long period of time. Once we take those advantages out though, companies are still left with the same uneasy feeling about what to expect in the future, and that is definitely not something management feels very attracted to.

[Sat Nov 1 17:25:43 CST 2003]

It turns out that Lindows has been sponsoring the development of a web authoring package for Linux that should rival the likes of FrontPage and DreamWeaver, or at least that is the promise. The screenshots certainly look like FrontPage, but in these cases the devil is in the details. For the time being, Nvu appears to already include support for real WYSIWYG HTML development, templates, tabbed editing and integrated file management via FTP, all features that were so far missing from similar Linux applications and therefore making it quite difficult to recommend the OS to even the most sporadic editor of web pages. Believe me, I had precisely that problem with my wife. Sure, Linux already had plenty of other nice GUI editors but one had to be familiar enough with HTML to code the pages directly, line by line. Mozilla Composer (and Netscape Composer prior to that) was the only realistic alternative for those who needed to put together a quick HTML document in the Linux or UNIX world, and it lacked quite a bit compared to similar Windows applications. There is little doubt in my mind that if Nvu reaches its goals, it will do nothing but to bring the reality of the Linux desktop one step closer and that is always a good thing. Incidentally, I also found it interesting that the whole project is based on code from Mozilla Composer and, according to the project's FAQ it will be merged into the Mozilla project.