From the Mind of Trevor Stone

The Atlantic has an interesting in-depth article about what we've learned in the six years since Malaysian Airlines flight 370. It's worth a read, but in case you don't have the time, a summary is that

• Pieces of the plane have been washing up on the western shores of the southern Indian Ocean for a few years
• The known portion of the flight path can only be explained by a human at the controls
• The person at the controls appears to have turned the electrical and communication systems off and then back on again
• It's likely that this depressurized the cabin, leading to the relatively quick and painless death of all the passengers and flight attendants, but not the pilot
• Folks who know the captain say he showed signs of social trouble and mental health challenges
• The captain ran through the fatal flight path on Microsoft Flight Simulator
• The Malaysian government, not known for transparency and good governance, has covered up a lot of evidence and wasn't forthcoming with accurate information during the initial emergency response

I found the article fascinating in part by how much can be learned from a few small pieces of data and a lot of collaborative investigative work. If you're of an appetite for more air travel mystery, the story of the hijacking and parachuting by D. B. Cooper is a fantastic rathole.

It's not simple to look for a job with narrow domain constraints (my main interests are a pretty small subset of the total programming work) but wide geographic flexibility (I'd enjoy most places on either coast plus much of the interior west). Keyword searches on major job sites can find some good matches, but some of my interests are broad enough that words or phrases might not be shared. For instance, I'm interested in geography, but that word might not appear in a job posting for a company making location-based mobile apps. And even if a word with "geo" as a root is present, can I trust Monster's search algorithm to match "geography" to "geographic?" It certainly won't infer GIS from geography or vice versa. Search result pages on these sites usually don't give much context, so there's a lot of forward-and-back browser action. Further, most sites don't hide jobs you've already looked at and rejected, so search variations may give you the same lame company you've already decided against.

The most usable approach I found was to subscribe to the RSS feed for the appropriate job category in every city I thought was interesting. This included a lot of postings for positions I wasn't interested in, but with Google Reader I could quickly skip those (j/k for down/up, vi uber alles!) and not see them again. After a few passes with this approach, I got pretty good at identifying postings that don't apply by the title or a quick scan of the first paragraph. The process was still time consuming -- the SF Bay software jobs feed can generate a few hundred postings in a week -- but I gathered a sense of the general software talent market in addition to finding companies of interest in what felt like less time than weeding through a narrowed set of results on traditional sites. Note that this approach can work for any industry and can be a low time commitment if you aren't following feeds for a dozen cities. If you're looking for work, go to http://yourcity.craigslist.org/ and click on the job category you want. Copy that URL into your feed reader and let job postings come to you. You can also subscribe to a more specific keyword search.

Of course, this process didn't lead me to the job I actually landed, but I did find some companies I was interested if I couldn't reach my top target and some products I think are interesting, even if I don't want to work on them. Some of these are mature, others show promise. I haven't tried most of them, but they look cool.

• Simply Hired job site aggregator which can apply multiple filters (location, position, skill, etc.) simultaneously. Still feels like it needs some work; I didn't find it as helpful as my Craigslist approach above. Indeed provides a more Google search-like job site aggregation.
• RoundPegg presented at this month's Boulder New Tech Meetup. They're working on an eHarmony approach to job search. Gather data on companies and teams, then have potential hires take a personality test to see how well they would fit interpersonally. The second phase will be letting job searchers put their info in and find employment matches. OkCupid's approach to dating is how I'd really like to approach job searching, so I think (and hope) this could be really successful.
• TouchGraph Navigator Automated data visualization.
• Alelo combines 3D gaming with language learning. With the Department of Defense as a major client, their program focuses on cultural norms and nonverbal communication in addition to phonetics, vocabulary and grammar.
• Outside.in provides hyperlocal content feeds. Pick a location (e.g., your house) and you can get news, blog posts, business listings, etc. They work on both natural language and geography problems and use some exciting technologies, so I was pretty interested in working with them. If I hadn't gotten a job at Google, these guys would be high on my list of interests. They use Mallet as their NLP toolkit; I'd like to check that out some time.
• SkyGrid provides real-time personalized(?) feeds of business and financial news across many reputable sources. This would be pretty cool if the service worked on any subject the user was interested in.
• Rosetta Stone has several personal and enterprise language learning packages. Their Boulder office has some folks from CU (including one of my teachers) and is working on cool stuff including online games where you can interact with native speakers.
• Geodelic is a mobile app that shows places based on proximity and interest. It can figure out if you don't like Starbucks and recommend coffee shops that are further away, but not part of the famed bean empire. They seem kind of focused on providing value to advertisers, so I'm not sure I'd rely on them for unbiased information.
• SimpleGeo is a Boulder startup that will provide an API and data cloud to make it easy for folks to write apps like the above without the expense of gathering and storing all the data first.
• Plectix treats microbiology as a programming language. If I was into bioinformatics instead of NLP, I'd be way excited about these guys.
• Allen Institute for Brain Science is developing atlases of the brain. Founded by Paul Allen, Bill Gates's lesser known other half.
• Acuitus makes some pretty grand claims about their digital tutor's results. Done right, it could be a huge boon to education.
• Language Weaver provides automatic machine translation for enterprise. I don't know what the state of the art in commercial translation is, but it's likely better than free translators like Google's. How much better, I don't know.
• imo.im provides web-based multi-protocol instant messaging. This is a good site for chat if you're often on public computers or just borrowing a friend's. Long-term, the company wants to do a lot of cool things with communication.
• The Experience Project a social networking site based not on who you know but who you are, with groups like "I survived cancer," "I am bisexual," and "I love bacon."
• Topsy a search engine based on links from twitter.
• The Game Crafter custom production of game components for homemade games.
• Urban Mapping provides geographic data like neighborhood boundaries.
• Open Layers is an open source platform like Google Maps. Thematic Mapping provides a JavaScript API to generate KML for Google Earth et al. FusionMaps is an open source project for creating interactive maps in Flash.