« September 2004 | Main | November 2004 »
October 30, 2004
I Wonder if Anders Ever Bought That Submarine
Delkin (not to be confused with Belkin the maker of overpriced under-designed hardware) has released a CD-R that they claim will last for 300 years. It's coated in 24k gold, but other than that looks just like a normal CD-R.
This is just the thing to store the kernel for the logic engine on, but really, I need a stolen Nazi submarine to transport the disks between Norway and Peurto Rico... so that's where Anders comes in.
He had a plan to raise a scuttled Nazi sub that was sitting in the shallows off of, Gibraltor I think. If he's got the thing up and running we can be on our way to control the World's financial systems with the unstoppable power of logic.
Posted by ashusta at 08:30 AM | Comments (0)
October 27, 2004
Google's New Acquisition
Google announced today that they've purchased Keyhole.
So, who's Keyhole and what do they make? They're a small company over in Mountain View, run by some Haas School of Business MBAs, that make a globe exploration tool. You can repeatedly zoom in on a given target using satellite photos. I downloaded the trial version of their software a couple months ago, and it was fun to play around with. It looks like they've launched a new improved version since then as well.
There is however, competition in this space. NASA have released a free product, World Wind, that has basically the same functionality as the intro level version of Keyhole's software. Except NASA's version is free, and NASA's server's are constantly swamped. As to why Google is interested in this space, I've no idea. They claim they bought the company because "it's an incredibly powerful information tool." Perhaps Google is planning on sending some shock troops up to Redmond and they wanted to get the latest intel before shipping out?
Posted by ashusta at 08:24 AM
October 21, 2004
Targeted Search Engines :: Part 3
Snap - Lots of Metadata for the User
Using a combination of Javascript and PHP to generate page layout for results Snap attempts to differentiate itself by giving visitors a ton of information to work with.
So when your search is clearly for a specific product (an MP3 player or digital camera for example) Snap creates a set of PriceGrabber style product finder results that can be sorted by various criteria depending upon what you're searching for.
Interface: Okay, Snap has some work to do here. It's not that the page is horribly ugly or impossible to navigate, and I do realize this is all Beta, so they're probably looking to score some investors, but come on! The front page is a bunch of jibber jabber about Snap's statistics. The text field to enter search terms is clearly marked, but why have a bunch of irrelevant info on the front page?
Speed of results generation is also an issue. I don't know if they just need more processing power or if the technologies that are driving the page are inherently slow, but waiting 5-10 seconds for search results to be created, parsed, and displayed is really annoying when compared to Google's sub second response time.
Results: This is a mixed bag. For example a search for "C/C++ exec() call" on Google brings back only pages that talk about programming. On Snap the same search brings back some results for programming, and some results for softwood lumber from Canada (?) and rather amusingly the top result was a for a Geocities site about the band The Soviettes. On the upside, the product finder results are interesting and sort of useful, but your search has to very clearly be for a specific product to get Snap to generate those listings. A search for "MP3 player reviews" doesn't get the product finder listings even though someone who ran that search is most likely looking to make a purchase.
Finally, there is a "Refine results" bar that will interactively refine your search results. It works like "Find in this page" in Firefox. As you type the word or phrase you're refining onto, results that match are highlighted while results with no matches are removed from the screen. This is nifty.
Value Add: Meta data on results is supposed to allow users to make more informed decisions about which results are relevant, the interactive refine method allows users to filter results quickly, and the product finder creates a clear differentiation from other general purpose search engines.
Conclusions and Potential Business Case: Snap has some good technology here, especially the interactive refinement of results. However, the results in general just aren't up to snuff when compared to current gen. search engines. If Snap was framing Google results (which I believe is disallowed under the Google API license) this would be a killer page. The idea of offering a couple discreet sets of results (product finder and general results on the same page) is also interesting. Since Snap is still Beta, and in fact being funded by the tech incubator Idealab, they probably aren't looking to take over the search market just yet. Rather, I expect they're hoping to sell their technology to other sites that need local search engines for content management applications on an intranet. Given the cost of the Google search appliance I can see this being a great niche. 
According to the Snap weblog they've already started to make changes based on user feedback, which is great. Obviously metadata can be useful in searches, now Snap needs to figure exactly which types are most useful for many discreet search types and that's a daunting challenge. But if they can stay alive long enough to make those discoveries, and implement them on the site, Snap may become a truly useful general purpose search engine.
Posted by ashusta at 04:42 PM | Comments (0)
October 19, 2004
eBay on Campus
Daaaaamn, who'd a thunk it? eBay sent a team of 4 reps to campus to recruit from the ISM and CS programs today.
Of course, this had nothing to do with the student employment office and everything to do with the School of Engineering. Yay for Baskin!
Highlights of the presentation included a preview of eBay's upcoming ad campaign, lots of fun stats (6-8 Gigs per second of total bandwidth usage), and of course the chance to schmooze with the hiring managers and give them our resumes.
They're hiring fresh grads for a 9-12 month long Associate Software QA Engineering position/development program. The idea is to get new hires experienced across the entire enterprise, working with code from many different divisions so that at the end of the program when they make the jump to a career position both the individual and the company are fully educated about their opportunities and strengths.
Julian, my boss, wrote a superb letter of recommendation for me (THANKS JULIAN!), that I was able to include with my cover letter and resume. Leaving the meeting I'd say there's a pretty good chance of at least getting a call back.
Oh and guess what eBay is using to transact > 1 billion page loads per day.... Java. But not Sun's Java, they're using IBM's Websphere implementation, not because it's fast but because of its error handling capabilities and "some other stuff." But yeah, it was cool to get to talk with eBay guys and see what other graduation bound UCSC students are presenting to potential employers.
Posted by ashusta at 08:06 PM | Comments (0)
October 18, 2004
Stalin VS Hitler - The Comic
Yes it's old and most people have already seen this, however I just came across the link again so I thought I'd share.
Posted by ashusta at 04:28 PM | Comments (0)
October 15, 2004
One Day Ahead of Wired
Here's Wired's take on RSS and ad revenues, WIRED Mag. Choice quote from the article:
'Rich Skrenta, CEO of Topix, said that with the recent growth of RSS, advertising is likely to stay. "Folks understand that if there's not a way to monetize content, there's not going to be content," he said. '
Posted by ashusta at 09:11 AM | Comments (0)
October 14, 2004
Stephenson On Newspapers and Dare I Say, RSS?
"Now, most places did not have newspapers, and so, if Mrs. Arlanc had not brought him any, he would never have known that they were wanting. But London had eighteen of them. 'Twas as if the combination in one city of too many printing presses; a bloody and perptual atmoshpere of Party Malice; and an infinite supply of coffee; had combined, in some alchemical sense, to engender a monstrous prodigy, an unstanchable wound that bled Ink and would never heal. Daniel, who had grown to maturity in a London where printing presses had to be hidden in hay-wagons to preserve them from the sledgehammers of the Censor, could not quite believe this at first; but they kept coming, every day. Mrs. Arlanc brought these to him as if it were perfectly normal for a man to read about London's scandals, duels, catastrophes, and outrages every morning as he spooned up his porridge."
And I thought that I used too many commas, semi-colons and parenthetical asides. Ah, but that's what I meant about the RSS feeds. They make available information that would previously have taken time to investigate then locate then read. Stephenson is describing a paradigm shift in information flow in the previous paragraph, and that is exactly what RSS (or any pull oriented headline grabber) may engender.
Also, big thanks to Chyna for giving me The System of the World for my b-day.
Posted by ashusta at 10:35 PM | Comments (1)
The Magic of RSS
I've been using theSharpReader RSS reader for the past month or so to aggregate new stories from websites that I frequent. Every morning there are 50-100 headlines available for me to peruse from diverse sources. This has cut the time it takes to gather current news from about an hour to about 30 minutes.
Since most blogging software also offers RSS feeds I'm also able to see any updates that have been made to my friends' blogs without having to visit each site individually. This is especially useful since many people only update their personal blogs occasionally.
The custom newspaper is here, the only problem is that it cuts advertisers out. If everyone was to start using RSS readers instead of surfing, I suspect sites would stop offering feeds. Or, like Engadget, embedding small text advertisements in the feed itself thereby guaranteeing delivery to users. Of course, if sites are just pushing spam with their feeds, no one would use them. Eventually a balance between advertising and content will have to be struck, but for now RSS is a great way to quickly access content while avoiding ads.
Posted by ashusta at 09:02 AM | Comments (0)
October 13, 2004
Check out this great wrapping paper
Oh yeah, sexy pinup swimming. It seemed a shame to open the gift the wrapping was so nice.
But how does she keep her makeup from running?
Posted by ashusta at 05:15 PM | Comments (0)
October 12, 2004
Targeted Search Engines :: Part 2
Clusty - Using Semantic Clustering to Add Value to Results
Clusty's premise of grouping results by category dovetails nicely with some of Tim Berners-Lee's work on a semantic web. The idea that data is interesting or useful only in the proper context isn't revolutionary, but the ability to filter and sort the results of searches so as to preserve that context is certainly novel.
Inteface: Users have a choice of what type of search to run that is specified by tags along the top of the search form ala Google. Results are displayed under the search form, with a pane along the left of the screen that displays cluster topics. A pulldown menu allows users to re-sort how results are clustered. Overall, less emphasis is given to the clustering in results than one would expect. This basically looks and feels like current Google results with a sidebar.
Results: A search for "somethingawful forums" actually brought back a cluster named "Goons" so I gotta give props to Clusty for returning results that are appropriate and seem to be grouped pretty well. However, the presentation of these results is... not what it could be. If clustering results to give them a context is what this site is about, why isn't more space dedicated to this task?
Value Add: Clusty allows users to customize the site by creating their own category tabs using a set of pre-determined search engines. They also have a "Gossip" tab, which may be of interest to some people. Their news search service is almost exactly like Google's which calls into question its reason for existing.
Conclusion and Potential Business Case: Application of semantic analysis to results is an interesting idea, and I can see it being especially useful for academic topics. However, Clusty doesn't really leverage their technology as well as they could. If you're really looking for a specific bit of information, and refining your search over a couple iterations it would be more useful to see the clusters as the main results; thereby encouraging a user to drill down, generating new searches as they go, until they find what they're looking for. I'm sure that Clusty will revamp the interface (the site is still "beta") in the future, hopefully adding more emphasis to the clustering idea. What I'm not sure about is how Clusty plans to generate revenue. There are sponsored results from Overture, but unless Clusty is able to build a significant user base very quickly I don't see this creating the large cash flows a startup needs to grow.
Clusty has an interesting and useful technology, now they need to figure out how to highlight that and bring in some users and cash.
Posted by ashusta at 02:30 PM | Comments (0)
October 11, 2004
Targeted Search Engines Attempt to Add Value to Searching
Google has owned the search technology market for the past three years. They've managed to consistently generate the most relevant results to queries and built a successful business model around their AdWords services.
Now competitors are beginning to bring their own products to market in an attempt to gain some of the 625,000,000+ searches executed per day (per SearchEngineWatch.com).
For now, no one is claiming that they've got a better general purpose search engine than Google. Instead they're working to add value to the search by revamping the interface and making changes to how results are presented. Much of the information that's presented is meta-data or based upon the semantics of the search terms submitted.
I'll be looking at offerings from:
For each offering I'll evaluate the interface and quality of results presented, but more importantly (for me at least) I'll try to interpret the business case behind the offering. Clearly no one thinks that they can displace Google as the top dog of search, so each competing product must offer some innovation in approach or results that make it more attractive to a segment of users.
A9 - People are searching for things to buy, really.
Given the premise of A9 and the fact that they require you to have an Amazon account is troubling. I mean, these guys have your credit card number and now they want you to submit every search you make? Sounds like a perfect set-up for blackmail to me. Tinfoil hat theories aside, A9 keeps track of your searches over time and allows you to review or revisit previous results. They also link directly to Amazon's impressively wide catalog of stuff you might want to buy.
Interface: Results are displayed in vertical panes along the bottom 2/3's of the page, grouped by category. Users can choose from 8 generic categories, web, books, images, movies, reference, history, bookmarks, and diary. These panes are user resizable, so you can give more space to image results if you'd like. Extensive use of Java scripting results in a user experience that's more akin to a small program than a traditional web page.
Results: Information is culled from Alexa, Google, IMDB and Amazon's own bots, so the results were generally quite good. However, A9 doesn't support searching with regular expressions from the default interface. Users have to go to the "Advanced Search" page if they want to exclude results that have certain words, contain an exact phrase, etc.
Value Add: If you're shopping for media, A9 works well. The ability to look inside books alerted me to the fact that "Alexander And The Terrible, Horrible, No Good, Very Bad Day" was actually just one of a series. Results for THX 1138 brought in the expected hits as well. The best part of A9 is probably the ability to customize the result panes that are displayed. I found the IMDB and Reference panes to both offer relevant information that wouldn't necessarily be listed on other sites. The rest of the system (specifically requiring an Amazon account for full functionality) is nothing special and even felt intrusive.
Conclusion and Potential Business Case: Amazon has a decent search engine here, but they've decided to limit the potential number of users to people who already have an Amazon account or are willing to set one up for the express purpose of searching. This is quite a barrier to entry when compared to most other search engines. I assume they'll use search statistics to support strategic planning in light of enhanced information about customer interests. This may be a winning decision in the medium to long run, but I wonder if they'll be able to justify the ongoing expense without being able to show a direct impact on income.
Next: Clusty - Not so crusty, but they need a new logo.
Posted by ashusta at 11:20 AM | Comments (0)
October 06, 2004
*Sigh* He can't even speak.
The Bushism people have put out a DVD with clips and commentary of GW's little slips while speaking. There's a video preview in Quicktime format if you click the pic below.
The great thing about misspeaking so consistently is that one can plausibly deny saying anything.
Posted by ashusta at 08:36 AM | Comments (0)