Google personalized search in beta

Google is beta testing personalized search.

It's a profile-based approach (not what I would use) and it doesn't appear to be all that effective. But it's an impressive step forward.

Personalized search is inevitable. With only one general relevance rank, it is increasingly difficult to improve search quality because not everyone agrees on how relevant a particular page is to a particular search. At some point, to get further improvements, relevance rank will have to be customized to each person's definition of relevance. When that happens, you have personalized search.

Electricity from Sewage

Impressive innovation in fuel cell research. Could be quite interesting if it can be done at scale.

Freedom of Information Act strikes again

See the campaign contributions of all your neighbors. An incredibly clever use of publicly available data.

More news at Findory News

There's a new news feed at Findory News with thousands of additional sources, better categorization, and improved summaries.

Anything into Oil

This is an old article, but it's important enough to bring up again. Discover Magazine published an incredible article (a free version of the article is here) on turning waste into oil. The basic process is like oil refining, but with a first step that can effectively take extremely low grade oil in the form of ground up waste. While the US doesn't have a lot of crude oil these days, we have plenty of waste, so this technology could be a fairly big deal.

Update: A research team is trying to use the same technique to turn pig manure into oil.

Update: The first of these plants went online today (May 19, 2004), producing a few hundred barrels of oil a day from agriculture waste.

Outsourcing and labor costs

Outsourcing skilled work like software development has been in the news quite a bit lately. I've been surprised by how simplistic the thinking seems to be of some business managers when deciding whether to outsource. The argument seems to be, "The wages are less abroad, so we save money doing it." In a talk a few months ago, I heard a CEO from a major US company claim that, with Chinese wages at 25 cents an hour, you can't afford to not be in China.

But labor costs are not determined by wages alone. Labor costs are determined by two factors, wages and productivity. High productive but high cost workers can be competitive with less productive but cheaper workers. To analyze your labor costs, you need to consider productivity.

There are some reasons to believe that productivity for outsourced software development and other skilled labor will be much lower even with substantial training and education. Infrastructure in many countries is very poor, with bad roads and unreliable electricity and telecommunications. Communication challenges, cultural differences, and management difficulties across thousands of miles and several time zones can reduce output.

This isn't to say that outsourcing isn't justified in many cases. But the analysis isn't as simple as using the lowest paid workforce.

Update: The April 28, 2004 New York Times ran a front page article, "Companies Finding Some Computer Jobs Best Done in U.S." that supports my claims, quoting one executive as saying, "The cost savings in India were three to one, but the difference in productivity was six to one," and another predicting, "the hype cycle about Indian outsourcing runs its course, [and] projects will come back to the United States when people find that their productivity goals have not been met." "Geographic distance and the differences in business contexts" are the cause of this productivity drop. The article also cites distance from the customer and "the customer's needs" as a cause of lower productivity and performance.

Another executive points out that, "Whenever the pace of innovation is very rapid is when the work should be done closer to the client. What cannot be sent to India is the invention of new business processes and technologies. Innovative business processes result from an understanding of the business that happens when people get into a room and talk to each other. That is very difficult to outsource."

Update: One year later, CNet reports on a Gartner Research study:

80% of organizations that outsource their customer management operations purely to cut costs will fail to do so, while 60% of those who outsource parts of the customer-facing process will have to deal with customer defections and hidden costs that outweigh any potential savings offered by outsourcing.

And InfoWorld reports on a Deloitte & Touche research study:

70% of participants have had negative experiences with outsourcing, [25%] realized that they could handle certain functions better in-house, and yanked those back inside the corporate walls, [and] 44% did not see cost-savings from outsourcing. 73% are [now] working to reduce outsourcing vendor dependancy.

Update: Two years later, I came across this excellent lecture by Paul Strassmann, "Is Outsourcing Profitable?" (PDF). Paul's research shows that there is no correlation between outsourcing and profitability, outsourcing increases overhead costs, and outsourcing damages the knowledge assets (and therefore the long-term potential) of a firm.

Styrofoam recycling

Disposing of large amounts of styrofoam is a real pain. While there's some packaging stores that will take clean styrofoam peanuts, it's a challenge to dispose of large styrofoam blocks.

Most attempts to recycle styrofoam don't work very well. Styrofoam is cheap to produce. The standard method of recycling -- grinding up the styrofoam, heating it, and reforming it into blocks -- makes low quality styrofoam. But I was impressed when I came across this paper (pdf) from Sony. It's a clever method of melting the styrofoam, storing it efficiently, then extracting the raw material. All with non-toxic solvents. Sounds like a great solution.

Update: Seven years later, a styrofoam recycling facility has opened near Seattle.

Antec Sonata case

I recently built a couple systems using the Antec Sonata case. I've got to say, it's one nice, quiet case. The power supply and case fan are nearly silent, quieter than the laptop I have in my office. A good review is available on Ars Technica.

Build your own HTPC

I just finishing building a home theatre PC (HTPC). Thought I'd share the specs in case anyone is interested in mooching off all the research I did to pick these components.

It's an Antec Overture case (quiet case) with a Athlon XP 3200 (AMD wins on cost/benefit every time), EPOX 8RDA3+ motherboard (nice motherboard with great features, comes with lots of high quality cables), ATI RADEON 9800 PRO video card (top of the line card, comes with Tivo-like software), Audigy2 ZS sound card (allows up to 7.1 speakers), Seagate 160 SATA Barracuta drive (quiet, high speed drive), Samsung combo drive (black, a quiet drive), and 512MB of Corsair DRAM in two sticks (allows dual channel for increased speed).

Should run beautifully. Entire system was just a bit over $1000. I did need a receiver -- I picked the ConsumerReports highly rated Panasonic SA-HE100 -- to power the speakers, but you might be able to get away without that if you have powered speakers. Connecting the sound card to the receiver is a bit of a pain -- you need special cables to split the signal and convert 3.5mm stereo jacks to RCA male jacks -- so I'd recommend powered speakers if that option is available to you.

Update: Two months later, I'm still working on getting this system configured with the Sony VPL-HS20 projector. The primary problem is that the 9800 PRO All-in-Wonder card seems to barely support component video output, working only in a very few resolutions and often with bad artifacts. In addition, all DVD software players apparently also contain restrictions on using resolutions higher than 540p, but only enforces that restriction when using TV outputs (like S-Video, composite, or component video), not with DVI output. Bizarre. So, I ended up switching the system to use DVI output, which works wonderfully, but requires pulling a DVI cable. In tests, using DVI does resolve all the issues, allowing a wide range of resolutions and providing crisp output.

Update: In early July, I have the system finished and configured. It does work well using the DVI cable. I'm able to use every pixel of the projector (1386 x 788), watch DVDs, watch TV, play music, browse the web, do PowerPoint slideshows, and play games. Unfortunately, the TV recording software bundled with the ATI Wonder card is pretty weak, definitely not as easy to use as Tivo; I may have to replace it with SnapStream. And, although DVDs play beautifully, the bundled software doesn't provide full support for 5.1 speakers, so I may have to upgrade the DVD software as well. But, generally, it's a great system with incredible flexibility.