Tim Oren's Due Diligence

Driving With Google's Bots

Back in the spring, driving home at dusk along 280, I noticed a Prius up ahead with a curious spinning truncated cone on its roof. From my experiences at the DARPA Urban Challenge I knew I was looking at a lidar range finder, often used on robotic vehicles. Intrigued, I pulled up alongside and could see a glowing LCD panel mounted in the passenger seat. I could tell that there was someone in the driver's seat (phew!) but not what he or she might be doing.

At the time I guessed this was a calibration run for another vehicle from the Stanford Center for Automotive Research. But it seems that what I saw was identical to a car video'ed by Robert Scoble (it even looks like 280 on his vid) and since identified as one of Google's self-driving cars. Less than three years from a set of rather twitchy military-funded demostration projects to something that was doing at least as well as the average Silicon Valley commuter (and was presumably signed off by Google's attorneys). A decent learning curve!

From a technologist's point of view, this is of course awesome and great fun. From an investor's POV, I'm left scratching my head a little. Some of Google's seeming digressions, such as Chrome and Android, actually make a lot of sense. They're about making sure that Google's end user touch points won't be compromised by closed networks or proprietary software. Jumping clear over telematics and into robotics is a bit of a stretch, though, along many dimensions: product cycle, capital requirements, channels and regulation, to name but a few.

The other possibility is that there is NO business plan. Perhaps what we're seeing is the result of the biggest cabal of Google engineers ever, using their 20% innovation time to pull off a state-of-the-art robotics project right under the eyes and flapping ears of Silicon Valley's pundits. Which would be doubly cool, and a sign that Google is assuming one of the social roles of companies with (near) monopoly margins over time, sponsoring open ended research. Shades of Bell Labs back in the day, minus the centralized planning?

The Roving Eye: Cool Technology Edition

Just in case you missed these:

25nm achieved. The Intel/Micro Flash memory JV announced the launch of a 25 nanometer fab, which will double storage densities. SSDs getting ever larger and, since many such are connected via USB 2.0, making that spec look ever more aged.

'Big Dog' robot procurement. DARPA lets a $25m contract to Boston Dynamics to create a deployable version of its Big Dog prototype walking robot. Lots of speculation on futures, some a bit out there, at the first link.

Spray-on glass? It's claimed that Germany's Nanopool has developed a spray-on glass coating material that creates a film a few dozen molecules thick, radically altering surface properties, including on living materials. Pretty cool if it pans out. Let's hope they've got a safe way to apply it.

The right turn in space. I spend enough time trashing Obama's economic incompetence that it's only fair to throw a bone when he does something right. Such is the decision to get NASA out of the rocketship building business. NASA was great back in the day, when it was new and had a clear charter, but like all bureaucracies it's turned into a haven for pork barrel projects and long term employment for functionaries. If we want space flight to be sustainable, not a taxpayer funded stunt, it's time to turn over the construction projects to private concerns that worry about things like cost learning curves, and minimizing the manpower and capital used for building and operations.

The Roving Eye: Ember the Mini-Bot, P. J. O'Rourke Nails It, Newspapers On Notice

How much? Thats my question about iRobot's new mini-warbot, developed for DARPA. Not only is it cute, but equip it with open hardware interfaces and/or programming environment and it should have serious potential for the hobbyist market, as well as repurposing to other applications.

The truth about suburbs. I've been a P.J. O'Rourke fan since his National Lampoon days. In a WSJ piece nominally about cars, he unloads this:

We’re way the heck out here in Valley Bottom Heights and Trout Antler Estates because we were at war with the cities. We fought rotten public schools, idiot municipal bureaucracies, corrupt political machines, rampant criminality and the pointy-headed busybodies. Cars gave us our dragoons and hussars, lent us speed and mobility, let us scout the terrain and probe the enemy’s lines. And thanks to our cars, when we lost the cities we weren’t forced to surrender, we were able to retreat.

Yes. He's defined part of the the battle-line of today's political cold war. Those 'futurists' who think that increasing fuel prices will bring suburbanites back to live in central city tenements are out in la-la-land. Instead the jobs may just follow the productive; Detroit gives a preview of that outcome.

A failed state. California's state expenditures have increased 35% since the 'tax cutting' Governator took office in 2003. As Matt Welch asks:

What, exactly, has been the return on this added investment? If spending under Gov. Arnold Schwarzenegger increased 6.75 percent a year during mostly good times, surely there must be, say, a 3 percent increase in the quantity or quality of...something? Crickets.

If we had back that 35% increase, the California budget would be in surplus. Instead of considering rolling back the spending, we're asked for more taxes and budget gimmicks. When the electorate wisely refuses them, the political class once again threatens to shoot their hostages - public safety and parks - while protecting the public employees' unions. What is there about layoffs and salary and benefit cuts that they don't understand? The private sector gets to deal with this in every downturn, while our betters in the political class skate through. Needless to say, this contributes greatly to the business climate in California.

A relic of the 20th century. Just what the newspapers need, more advertisers headed for the exits. Local fish-wrappers have long been the vehicles for required public notices of events such as mortgage foreclosures, public meetings and business name registrations, which creates an effective subsidy to their businesses. This requirement dates back to when the town rag was in fact read by most of those who would need notice of such events. With the readership shrinking, that's no longer a good bet, and some government entities are starting to take notice. It's pretty hard to argue that publication on dead trees is more useful than a combination of appropriate RSS feeds and a searchable archive would be. As the linked article points out, there are issues regarding reliable and trusted implementation, but it shouldn't be hard to beat the status quo. A business opportunity out there for someone - Google? Iron Mountain? Maybe even archive.org, if they'd like an actual revenue stream.

The Roving Eye: Best of the 'Bots, Flipping Newspapers, Dry California, Inaugural

Bots in review 2008: Here's a collection of vids of some of the most interesting robotic work of the past year. The Japanese are still on the humanoid path, but the diversity of forms from elsewhere is intriguing. If you haven't seen the Big Dog DARPA project, be sure to check it out - it's apparent mannerisms range from gigantic bug to strangely human. The USC reconfigurable components projects is also of interest; up to now robotics hasn't had much 'platformness' - they are all pretty much tailored to single functions.

How to become a newspaper millionaire. Start with a billion. (Rimshot.) The last week has brought us a Russian oligarch buying up the Evening Standard of London, for assumption of debts and one UK pound, and Mexican billionaire Carlos Slim putting $250m of senior debt into the New York Times, helping the Grey Lady keep the wolf from the door, and further diluting the Sulzbergers. Slim's track record when away from his telco and media monopolies is quite mixed. He managed to buy Prodigy just as the Internet was swamping the proprietary online services, and CompUSA as the computer specialty retail channel imploded. Update: More on Slim and the NYT.

It's official. La Niña is back. The reverse of the better-known El Niño Pacific Ocean climate pattern that brings floods to California, La Niña instead brings us drought. With last year already being short of rain, we could well be looking at water rationing and an even more intense fire season this year. This to go with California's economic woes, both micro and macro.

Good luck to President Obama. I didn't vote for him, but I wish him well. May he do better than I have feared. That doesn't mean the political bent here will change. I'll take my stand with Perry de Havilland:

So now is not the time for 'unity' and 'bipartisanship', which is just a genteel way of demanding surrender, it is the time for resistance and the renewal of purpose by those who see the liberty and prosperity that comes from constitutionally limited government as a prize worth any price to defend.

Hopefully, dissent is still patriotic post January 20th.

The Roving Eye: Arrrrr Is For Robot, Following the Genomics Learning Curve, More Public Finance Follies

Talk like a pirate day for bots? Just as plenty of alternative uses have been found for formerly mil-only spy sats, so the applications for robots keep expanding. Here a report that a US Navy underwater droid trial resulted in finding new artifacts near a Revolutionary era ship wreck. Development budget hurting? Try turning your AUV loose in certain Caribbean and Florida waters, maybe.

Tick. Reports of an inaugural meeting of would-be hobbyist biological engineers. Today Cambridge, MA - tomorrow, ?

Tick. Biologist Richard Lenski at my alma mater, Michigan State has a long-running experiment to follow evolutionary trends in thousands of generations of fast-reproducing E. coli. Seems he's now made the first observation of a two-step mutational change that gave his bacteria the ability to utilize a different nutrient. If you don't know enough to engineer what you want, get nature to do it for you.

GOOG + Open Office = LUV? Should Google embrace Open Office? It's hard to see what they'd have to lose by doing so. My own experiments with Google Docs showed it can hit scaling limits due to bandwidth and/or server scheduling pretty quickly. Sometimes you just need a locally-hosted interface or computational engine. Microsoft has long been engaged in gilding the --- something nastier than a lily --- on Office. Each generation has a little more feature creep, cuts off more backward compatibility, and fails to address long-standing usability problems. Apple's shown there's a need for a light and usable alternative. Google could simply exploit this opportunity in a more mainstream fashion than Jobs & Co. ever will, and could create a smooth transition from/back to their SaaS offering. I'd say go for it!

The land of negative ROI. In what world does every dollar sunk into capex guarantee you will lose more operationally? That would be government-run mass transit in the US. Long viewed as a combination of make-work spoils program and welfare for the riders, every buck 'invested' means a larger system that can't cover its opex, let along depreciation and additional capex. Out here in the Bay Area, we have a heavy rail system running at near capacity due to the gas price rise, and it still can't cover its costs. Something smells rotten.

Fleecing NYC of a half billion bucks. That's apparently the amount of unfunded public pension liability covered up by an actuary who just happened to be an employee of the city, and therefore a beneficiary of the programs he was supposed to be vetting. Someone who did this in private industry would be in jail for fraud. Why isn't he? And people are running for office based on the idea of government controlling more of the economy.

The Roving Eye: Someone Poke Fake Steve, Spybots Invade England!, Genetic Nannyism

Reality sets in at Facebook: Low entry barrier, low CPM. Even Fake Steve has noticed that most Facebook apps are fluff. More pointedly, developers are seeing very low effective CPM rates. CPM being a rather obsolete concept, I'd prefer to see data presented in revenue-per-user terms, as here. But it's clear that the open, easy, widely adopted Facebook is struggling to monetize, while the smaller, focused, and closed LinkedIn is claiming much higher effective rates (see interesting discussion here.) I'm still picking Facebook to be the Pointcast story of this decade.

Winning the War with Rhino Snot. Letting the troops (as versus the Pentagon) name your products can have interesting results. This company looks like it's self-supporting, but it would be fun to see its CEO try to get through a VC pitch without cracking a grin. It would definitely earn a place on the 'best name' honor roll.

Better see what's hiding in the closet. Following the precedent of DARPA's robotic Grand Challenge, Great Britain's Ministry of Defense (MoD) is mounting its own competition for automatons, with a twist. This time the robots are for surveillance purposes, meant to spy out snipers, IEDs and armed vehicles and soldiers in a village sized trial area. The task specific element of the challenge ups the ante from DARPA's trials, which were mostly about success in navigation. The MoD's competition also allows cooperating teams of bots, which could be of different types. One team anticipates coordinating a team of flying and earthborne bots. The MoD trial will take place in August.

The best things IBM ever made... were those clicky, battleship-weight PC keyboards. Dan's Data sings the praises of the 'buckling spring' design. (Via Derek Miller). Having spent many hours banging on an awesome converted Selectric used as the console on a IBM 1800 way back when, the PC keyboard was a welcome relief from the mushy action of the VT series and other 'glass TTYs' of the time. Almost worth having to put up with segment registers. The economics of the PC industry and weight requirements for portability put these things into the museum, but I still miss 'em.

What happened to 'Know Thyself'? Our self-appointed guardians in New York and California are on watch, making sure you can't get your genome analyzed at your own expense. Because it raises big concerns: 'What will patients do with this information? " Gee, would you suppose that might be their own business? if you wanted to freeze private investment into this area, you couldn't have picked a better way than this kind of statist fear-mongering. And maybe that's the point.

The Roving Eye: Eqyptian Unrest News Source, Howl's Moving Moonbase, Yahoo's Ad Platform

Unofficial Eqyptian Media. The Egyptian Sand Monkey is blogging up a storm about the revolt and riots there. There don't seem to be any Western journalists on the scene, and the police are trying to shut down the city and strikers, so Internet reports funneled out through blogs are the best information source.

Miyazaki On The Moon? Fans of a certain Japanese animated film might get a bit of déja vu from this NASA proposal for a mobile moon base for astronauts. Miyazaki always loved airships, but this is above and beyond.

Ad Vapor From Yahoo. Yahoo preannounces an ad management platform, that might help to enable the kind of site ad network that I blogged about before. The catch is that the rollout won't start until Q3, which won't have any impact on the current Microsoft takeover attempt. On the other hand, this could enable the type of destination site grand alliance that I mentioned in that post, combining the Microsoft and Yahoo properties and other big exposure ad venues, where the Google approach to ads hasn't worked so well.

Mistitled Robot Article. New Scientist has an interesting article titled "The Rise of Emotional Robots". The title, however, is pretty much backwards, as the content largely describes humans that get emotional about their bots, including milbots that get blown up in Iraq. For those who've followed the area of human computer interaction, finding humans projecting emotions - often unconsciously - onto machine displaying complex behaviors is not a surprise at all. More interesting is one cited study showing people when questioned about robot interactions don't believe the bots are 'thinking', perhaps due to their skepticism about current technical capabilities. Matching the likely human reaction to the task being accomplished by the robot is a very complex problem.

The Roving Eye: Tiny UAVs, So Long Brett, The Pachinko Effect

Itty-Bitty UAVs. Check out these pix from a recent unmanned vehicle conference in DC. The first one looks like one of those scale models you see displayed as a trophy in an aerospace exec or pilot's office. But no, it's a fully functioning drone - get it airborne with a bungee cord or using a grenade launcher attachment for a rifle. Have a fat wallet and an urge to build your own, or start a UAV venture for that matter? Check the next image for an assortment of controller components from tiny to chunky. If you're into things robotic, check out the main DIYDrones home page and feeds for everything from hobbyist to milbot news. Organized by Chris Anderson (yes, that one) - which is a good recommend given his record as a trend spotter.

The Roving Eye: Special Cluelessness Edition

Maybe it's Leap Year madness, maybe it's terminal election ennui, maybe it's the mind chilling effects of global warmening, but it sure seems like there's a spate of "what were they thinking" caliber cluelessness out there in the last week or so. Let me count the culprits:

The Associated Press sends its attorneys after a website that has highlighted their sad history of carrying 'fauxtography' - doctored images that amount to uncritical propaganda - often for terrorists. Yup, that'll make sure no one pays attention to the AP's issues.

The US Air Force tries to make sure our airmen can't see blogspot, or maybe even the blogosphere itself, 'cause something might happen. Never mind the blogosphere is a source of open source intelligence and a major IO theatre itself. La-la-la-la-la, I can't hear you! And these guys are supposed to be in charge of mounting the military's cyberdefense? I feel safer already.

Comcast, starring on this page for the nth time, can't just promise transparency on its broadband traffic policies, which seems to be all the FCC chairman really wants. No, they have to try to pack a public hearing at Harvard. Yeah, these guys look just like who I'd expect to see at a Cambridge Internet gathering. Heh. As far as I can tell, Comcast is where flaks and public affairs wonks end up when the USAF or AP finally fire them.

Publicity chasing AI professor Noel Sharkey says we're all in danger from kill-happy milbots with their virtual fingers on the trigger. (I don't usually link the Register, since much of their 'analysis' is unresearched snark, but ridicule is quite appropriate here.) While I've been following the military development thread in robotics for some time, it takes minimal research to find out that logistics and scouting are the designed uses and no-one's even close to having 'bots adjudicate rules of engagement. The history of academics forecasting commercial developments is a sad one, and it gets even worse when looking at military futures where their biases are engaged. Anyone else old enough to remember how we were told SDI was impossible by a bunch of CS profs? How'd that forecast work out?.

The Roving Eye: 2/21/08

Robot buoys to monitor the spread of oil spills. We could have used some of those in the Bay Area a few months back. Hopefully the Coast Guard will take notice.

What's "Volkswagen" in Hindi? For those who think we can solve the carbon emissions problem in the developed world alone, read 'em and weep. Like Detroit and Toyota before it, Tata knows that increasing prosperity of families and societies will cause demand for more elaborate vehicles to grow, and they are getting in on the ground floor.

Do your hiccups taste fishy? There's a reason for that. Another interesting result from reverse engineering of evolution. Hat tip to Ole Eichhorn.

There are some silly valuations out there right now ($550m for Slide? Gimme a break!), but here's the silliest funded venture I've seen in a while. Take a problem that's wicked for people with decades of experience, where your useful feedback happens in 2-10 years intervals, where many of the deal killer issues are inherently unknowable without executing the plan, and solve it all with AI! Which has been such a great success! (This reminds me of a post I've been meaning to write for a while now...)