News
Tesla announced that all new cars it sells will have a camera, ultrasonic and radar package that it thinks will enable full autonomous driving, if and when the software techniques for doing this arrive, at which point it will sell you a software upgrade. It also predicted that it will be able to demo a car driving with full autonomy from coast to coast across the USA in 2017. Vey few people in the field think that autonomy will be possible without a LIDAR sensor any time soon (LIDAR makes building a 3D model of your surroundings much easier), and very few think it will be possible even with LIDAR within 5 years. Many think 10 years (or longer) is more realistic. So, Tesla says it will do the harder approach sooner than anyone else does the easier approach. If that is possible, then having all the cars it sells from today collecting training data (even without LIDAR) will be very valuable. Somewhat buried in the announcement, the existing 'Autopilot' assisted cruise control product is not in these new cars, for a while, until Tesla's new tech (not just the full-autonomy part) is working, which is probably because Tesla fell out with Mobileye, which was providing the underlying technology. Link
A chunk of the internet went down this week, effectively, because someone did a massive distributed denial-of-service attack using a botnet of millions of hacked IoT devices - mostly, it seems, IP webcams from one Chinese company that don't have decent security. This is an interesting structural problem - the devices once sold are either impossible or unlikely to be patched, the users probably don't even know that their device is hacked, and the manufacturer has no motivation and probably few of the necessary skills to do anything about it. A network designed to withstand nuclear attack, brought down by toasters. More interesting/worrying - who is doing this, why, and what will they do next? Link
AT&T is buying Time Warner, for $84bn. (Verizon already got AOL, so get the other half.) People who own commodity pipes want more control over what goes over them, especially as it looks like the US pay TV system might start to unlock. But this is about consolidation in old-world industries, not tech anymore. If it gets past the regulator, of course. Link
Apple has, apparently scaled back or restructured its car project. One can read this as a mess, and one can also read it as the kind of reconfiguring that often happens in these kinds of projects a few years before they ship (and happened to the iPhone). My reading: Apple could make a nice EV, but so can BMW - Apple has no advantage in making doors that sound nice when they close. It can also make a nice autonomous pod that's centrally routed and optimized, but that's where Google's lead is. So how can it use software, and its skill in the way we interact with technology, to do something that only Apple can do? I suspect the middle-ground is fruitful - the car that won't let you crash but still needs a driver (that is, L2/3 autonomy). But if Apple hasn't yet worked out what it could do that would be unique, there's no point having lots of people speccing out lighting clusters and suspension systems. Link
Apple hired a director of AI research. Link
LeEco, a Chinese tech conglomerate doing everything from electric cars to phones to streaming TV, wants to enter the USA and build a big research centre in the SF Bay Area. This may end well. Link
Blog posts
"Someone is learning how to take down the Internet". Link
Pittsburgh is experimenting with using machine learning to optimize traffic signals. Link
Uber is trying to crack Japan by starting in small villages. Link
Sebastian Thrun on automation and changing jobs. Link
Your brilliant Kickstarter idea is already being copied in China. Link
REDEF presentation on the future of video. Link
Book shops and Amazon's last mile. Link
McKinsey report on monetizing car data. Link
Statistics
Uber has 40m MAUs spending an average of $50/month. Link
IBM: Macs have lower TCO than Windows PCs. Link
Slack is now at 4m users. Link
|
|
|
|
|
|
|
|
