A mighty Fortress...

At JavaOne last week, Sun updated the world on Fortress, its Java-based programming language for massively parallel processing. It’s consistent with Sun’s view of being an innovator, rather than a low cost commodity supplier.

The program from Sun’s Programming Language Research Group was originally developed using DARPA HPCS (nee High Performance Computing) money, and then with its own dollars.

For most of the past decade, Sun missed numerous opportunities by trying to control Solaris and Java rather than win adoption through open source. Given Sun’s origins as an “open systems” supplier, its failure to embrace openness was perplexing, in the face Linux at one extreme and Microsoft at the other. However, this time the prototype Fortress interpreter is already released under a combination of BSD and GPL licenses.

About 20 years ago, I was a programming language geek, working on compilers and programming language design, including some ideas about language design for one of the earliest massively parallel computer systems, the JPL hypercube.

At the time, the most radical programming language effort was Occam, intended for the Transputer processors. It was interesting to note that Sun Fellow Guy Steele acknowledges Occam in an interview about the Fortress efforts.

Having the smartest people in the field in house (like Steele, co-inventor of MIT’s Scheme language) is the path to differentiation through innovation, and having your own people implement the language is the surest path to make sure it works well on Sun hardware and software.

The problem is adoption. Historically, network effects have been the death of specialized languages (coming from the inventor of a specialized language). People would rather work with libraries wrapped around standard languages (like Fortran, C, C++ or Java) than learn a new language. So major breakthroughs like APL, Simula, Simscript, Smalltalk, Scheme and Occam were used to teach programming, but were rarely used to solve real program.

So the challenge will be to make Fortress the standard language for massively parallel systems — first at DoD and DoE research labs (like LLNL and Oak Ridge), then for government studies and analysis, and finally for computing-intensive industry problems like biotech and oil exploration).

To become the standard, Sun needs the cooperation of both the user and vendors of massively parallel computers. It doesn’t have any computers on the Top 10 (of the Top 500) supercomputers, which is dominated by IBM with other systems from Cray, HP and SGI. (Sun only has one machine in the top 50). So, as with any other standards-based competition in the industry, good technology is only valuable if it leads to adoption, tipping and network effects — and much of the adoption is driven by politics and alliances rather than the quality of the technology.

Upgrading to Office 2004

Last night, I upgraded from Office 2008 to Office 2004. Office 2008 is native for my Intel-based MacBook Air, but it turns out the four year-old Office 2004 (which I’d never used on an Intel Mac) actually runs faster. Plus is more reliable and has some more features. (Apparently I’m not the only one to notice).

I had hoped the SP1 update (Office 2008 12.1.0) released Tuesday would make Office better, but it’s about the same. For example, PowerPoint still wipes out the last changed date of any file you open (when or not you change anything). If not for the dreaded DOCX disease (a nonfatal virus spread by casual contact), I’d rip out of Office 2008 entirely.

I have not yet bought Windows Vista so I can upgrade to XP, but does anyone see a pattern here? Windows XP shipped in 2001 and its successor did not ship until six years later. Microsoft corporate is spending $7 billion a year overall on R&D — somewhere between 10% and 1/3 of that on Windows — so after six years that amounts to several billion dollars of R&D on Windows Vista. The cost of Office 2008 was probably only a few hundred million.

I am tempted to beat on Microsoft for poor management or lack of motivation, but I think the problem is more serious than that. Microsoft’s legacy code base is so large (if not bloated) that it’s very hard to add new features while keeping the old features, performance and reliability.

Rather than trying to be all things to all people, Linux (or FreeBSD or NetBSD) has the potential of allowing groups to customize the operating system to their own needs, serving a range of niches through decentralized community innovation. Desktop Linux seems dead for now, but such decentralized approaches seem one of the few options to overcome the inherent limits of coordinating such monolithic OS releases.

Wi-Fi: can’t even give it away

Not only is Earthlink getting out of municipal Wi-Fi, but it can’t even give it away:

The Atlanta-based Internet service provider said Tuesday that it could not find a buyer for the $17 million network and that talks to give it to either the city or a nonprofit organization had failed.

City officials have said it would cost taxpayers millions each year to operate the network.

“It's been an unfortunate situation,” Chief Executive Officer Rolla Huff told The Associated Press. “It was a great idea a few years ago, ... but it's an idea that simply didn't make it.”

EarthLink, which will give current customers until June 12 to switch to another provider, said it even offered to donate the Wi-Fi equipment to someone and give them an additional $1 million.

Earthlink’s announcement implies that their other business are sound, but, frankly, if they don’t grow in wireless access it seems that they are doomed to be a declining seller of dialup access and reseller of broadband that competes with their cable and DSL suppliers.

Still, Earthlink will probably outlive the imminent death of the municipal Wi-Fi movement, which has been facing imminent death for more than a year. With Earthlink abandoning systems already built, other cities have decided not to even start.

It must be a pretty lousy business that you can’t even make a business with a zero capital cost? At least the new Iridium was able to turn a profit after acquiring its assets for a half cent on the dollar. At least satellite phones have a small niche of customers who have no alternative, as opposed to connectivity in a big city where there are cable, DSL, dialup, local hotspots, libraries and increasingly 3G mobile phones.

However, the local nonprofit, Wireless Philadelphia, seems to be in denial:

Wireless Philadelphia and the City of Philadelphia continue to work together to ensure a positive future for Philadelphia's municipal wireless network and nationally-recognized Digital Inclusion program, the vision of which is to provide all citizens with access to essential technological resources for education, employment, and other life opportunities.

Good intentions and noble causes don’t obviate economic reality. It’s good to see such “frill” services held to some measure of economic accountability, even if the core functions of government are not.

HP buying EDS

To increase its services revenues, HP confirmed it is buying EDS for about $14 billion. They’ve been down this road before.

In the fall of 2000, CEO Carly Fiorina was going to spend $18 billion to buy PWC, but then gave up a few months because it was too expensive. After the end of the bubble, IBM bought PWC for $3.5 billion, and was almost immediately successful in making things work

Increased exposure to services was a major factor behind the 2002 Compaq acquisition that nearly destroyed HP and cost Fiorina her job. More recent analysis claims that the merger was a success, some of that with insiders arguing their case after Fiorina’s departure. (Whether the merger made sense or not, the process for making it work set a new standard for the rest of the industry to follow.)

As Business Week points out, EDS (like Compaq) has fallen far from its heyday. The company has been in serious trouble for nearly six years. In September 2002, it announced an earnings shortfall of 80%, prompting a strike price lawsuit after its stock lost two-thirds of its value and its credit rating approached junk bond status.

Now, the hometown paper writes:

For EDS, the deal represents a chance to cash in after years of cost cutting and reorganization failed to give the company's shares much of a lift.

For H-P, an acquisition would boost the company's ability to compete in the services area with rival IBM Corp.

And for EDS employees, the purchase almost certainly means at least some job cuts.
...
Peter Bendor-Samuel, founder of Everest Group, a Dallas consulting firm that helps companies do business with outsourcers such as EDS, said EDS is a mature company.

While the company is widely recognized for its expertise, chairman, president and chief executive Ron Rittenmeyer hasn't been able to boost the stock price, Mr. Bendor-Samuel said.

"Ron has been running the game plan to take costs out of EDS, but you can't cut your way to greatness, and quite frankly the stock has languished," he said.

"There's really nowhere for EDS to go," he added. "They can do minor acquisitions, but it's hard to see how that drives stock prices."

Mergers to increase scale are common in mature industries, although Fiorina (a modern-day Ahab) was obsessed with surpassing IBM’s scale in hardware and services. Mark Hurd seems to share those goals.

Under Fiorina and now Hurd, HP increasingly looks like a slow-growth commodity company that grows via acquisition rather than organic market creation. In that light (so to speak), the photonics summit that HP hosted Monday would be more about gaining licensing revenues for HP’s patents than enabling HP to differentiate its products through the use of photonics.

Loving the summer of code

When I was growing up, the big news out of the Bay Area was the “Summer of Love.” Now it’s the “Summer of Code.”

Google has announced its “Google Summer of Code™ 2008,” with 1125 students (out of 7,100 applicants) working on 175 open source projects. The announcement that it would spend $5 million funding summer projects got some blogger coverage, but the program (which began in 2005) is highly visible here in the Silicon Valley and among the open source community.

About a third of the applicants and 40% of the accepted participants are from the US. Eyeballing the pie chart, India, China, France and Sri Lanka had a lower acceptance rate than average, while the US, Canada, Germany, Poland and UK had a higher acceptance rate. (Brazil looked about average).

In looking over the projects, the sponsoring open source organizations have become more democratic (i.e., less elite). I browsed through the list of projects of the 10 or 15 most recognizable projects. A few of these looked interesting:

• DSpace: Moving DSpace into the age of the Semantic Web
• DragonFly: Port DragonFly to the AMD64 architecture (being done by a Russian student)
• Free BSD: TCP/IP regression test suite
• Internet Archive: Book Viewer for the OLPC

Beyond these, most of the projects are small incremental improvements to existing subsystems (and many of them obscure subsystems of obscure OSS projects). Very few are ones that I would have been excited about at age 21 (when I’d been a paid programmer for 5 years).

So my guess was that the attraction will be mainly for students who don’t have the option of getting a paid summer job with a real software company for real money. The high rate of application from India, China and Sri Lanka is consistent with this. (The high rate of Germans could be the generous public support for German students, who don’t need to receive worthless dollars to pay their rent).

Sure enough, if you look at the top universities for applicants and participants (covering only about 100 of the 1125 participants), conspicuously absent are schools with strong summer programs and strong local IT markets like MIT, Stanford and Berkeley (Georgia Tech being a notable outlier, apparently due to strong word of mouth). This would imply the participants are either highly motivated free software supporters, those without good professional opportunities (seeking to “flip bits not burgers”), financially comfortable, or conversely places where $4500 is a lot of money.

Google is spending this money without much direct benefits, for which they should be commended. However, I found the FAQ claims of altruism to be misleading at best. Either Google’s geeks aren’t very precise writers, or they’re being disingenuous:

3. Is Google Summer of Code a recruiting program?
Not really. To be clear, Google will use the results of the program to help identify potential recruits. But that's not the focus of the program. Take a look at the organizations we've worked with in the past, and you'll see the vast majority are engaged in work that's not directly applicable to Google's business. That said, the more code out there, the more everyone benefits.

Additionally, we've heard from several of our past student participants that their participation in GSoC made them more attractive to potential employers, and most participants who have gained employment as a result of their GSoC work are not currently employed by Google.

This is pretty easy to parse:

• "not related to Google technologies" might be true, but when screening applicants it’s clear that working on NetBSD provides a better evaluation of programming ability than a class project
• "most participants" could mean "we have 49%" and “not currently employed” could mean “we hired them initially.”

Google’s legendary secretiveness means that we don’t know what the true benefits are of the program to Google: it might be altruistic, but the evasiveness of this wording seems like they are shooting themselves in the foot. I could imagine that after accepting 15% of the applicants, they could hire 10% of that 15% to work at Google — a net cost of about $45,000 per recruit. (Not cheap, but not outrageous). Since Google does everything by the numbers, this calculation has certainly been done — perhaps why they don’t disclose their hiring rate.

Still, this is more altruistic than a typical summer internship program. Thanks to the Internet (and the open source development tools), there are relatively low coordination costs for such virtual distributed work. It’s clear that the sponsoring organizations (such as the Perl Foundation) have put a lot of thought into how they’ll organize the efforts.

It’s not something that would have been possible when I was in college, when long distance calls were $1/minute and a high bandwidth data channel was sending a 9-track tape (holding at most about 150 mb) via Federal Express. The upper limit was “station wagon bandwidth,” attributed to Andrew Tannenbaum of Minix fame.

Who's the Mac user?

A friend forwarded an Internet joke that has been circulating since at least February 2004 but has also been posted in 2005, 2006, 2007 and in 2008. Here it is:

Abbot: Super Duper computer store. Can I help you? ..
Costello: Thanks. I'm setting up an office in my den, and I'm thinking about buying a computer.

Abbot: Mac?
Costello: No, the name's Lou

Abbot: Your computer?
Costello: I don't own a computer. I want to buy one.

Abbot: Mac?
Costello: I told you, my name's Lou

Abbot: What about Windows?
Costello: Why? Will it get stuffy in here?

Abbot: Do you want a computer with windows?
Costello: I don't know. What will I see when I look in the windows?

Abbot: Wallpaper.
Costello: Never mind the windows. I need a computer and software.

Abbot: Software for windows
Costello: No. On the computer! I need something I can use to write proposals, track expenses and run my business. What have you got?

Abbot: Office
Costello: Yeah, for my office. Can you recommend anything?

Abbot: I just did.
Costello: You just did what?

Abbot: Recommend something?
Costello: You recommended something?

Abbot: Yes.
Costello: For my office?

Abbot: Yes
Costello: OK, what did you recommend for my office?

Abbot: Office.
Costello: Yes, for my office!

Abbot: I recommend Office with Windows.
Costello: I already have an office and it has windows! OK, let's just say, I'm sitting at my computer and I want to type a proposal. What do I need?

Abbot: Word. Costello: what word?
Abbot: Word in Office.

Costello: the only word in office is office.
Abbot: the Word in Office for Windows.

Costello: which word in office for windows?
Abbot: the word you get when you click the blue W

Costello: I'm going to click your blue W if you don't start with some straight answers. What about financial bookkeeping, you have anything I can track my money with?
Abbot: Sure, Money.

Costello: that's right. What do you have?
Abbot: Money.

Costello: I need money to track my money?
Abbot: it comes bundled with your computer.

Costello: what's bundled with my computer?
Abbot: Money

Costello: money comes with my computer?
Abbot: yes. No extra charge.

Costello: I get a bundle of money with my computer? How much?
Abbot: one copy

Costello: isn't it illegal to copy money?
Abbot: Microsoft gave us a license to copy Money.

Costello: they can give you a license to copy money?
Abbot: why not, they own it.

Of course, the parody is a tribute to one of the best vaudeville skits of all time.