Every 10 years or so, a conspicuous bubble bursts, and in doing so it resets the expectations of the next generation of young adults.
- 2008 financial collapse
- Now Theranos
Reading this article, I’m astonished at how little substance the adulation of Elizabeth Holmes was based on. And how much secrecy her investors allowed her. Given that she was claiming that her system would be ~100x better than established technologies, why didn’t they demand evidence? Why was it left to a reporter to figure out that the emperor had no clothes? And, was she nothing more than a successful con-artist with no genuine scientific expertise?
“In a searing investigation into the once lauded biotech start-up Theranos, Nick Bilton discovers that its precocious founder defied medical experts—even her own chief scientist—about the veracity of its now discredited blood-testing technology.”
Source: Exclusive: How Elizabeth Holmes’s House of Cards Came Tumbling Down | Vanity Fair
#firstsevenjobs is an interesting example of crowdsourcing research
- Dishwasher (^3) (Exeter, Harvard, and a summer job)
- Lifeguard (Local swimming hole)
- Library assistant (^2) (Harvard. One was work and one was a sinecure. I’m still really fast at putting things in alphabetical order.)
- Sci. programmer (Smithsonian Astrophysical Observatory)
- IT salesman (IT startup company)
- Business programmer (Xerox. My first taste of a really big company, and I hated it.)
- Energy consultant (DC consulting firm)
from Twitter https://twitter.com/RogerBohn
August 10, 2016 at 10:11PM
Secretive Alphabet division aims to fix public transit in US by shifting control to Google (from The Guardian)
Documents reveal Sidewalk Labs is offering a system it calls Flow to Columbus, Ohio, to upgrade bus and parking services – and bring them under Google’s management.
The emails and documents show that Flow applies Google’s expertise in mapping, machine learning and big data to thorny urban problems such as public parking. Numerous studies have found that 30% of traffic in cities is due to drivers seeking parking.
Sidewalk said in documents that Flow would use camera-equipped vehicles,…. It would then combine data from drivers using GoogleMaps with live information from city parking meters to estimate which spaces were still free. Arriving drivers would be directed to empty spots.
Source: Secretive Alphabet division aims to fix public transit in US by shifting control to Google
Notice that this gives Google/Alphabet a legitimate reason to track every car in the downtown area. Flow can be even more helpful if they know the destination of every car AND every traveler for the next hour.
The next logical step, a few years from now, will be to track the plans of every person in the city. For example Mary Smith normally leaves her house in the suburbs at 8:15AM to drive to her office in downtown Columbus. Today, however, she has to drop off daughter Emily (born Dec 1, 2008, social security number 043-xx-xxxx) at school, so she will leave a little early. This perturbation in normal traffic can be used to help other drivers choose the most efficient route. Add together thousands of these, and we can add real-time re-routing of buses/ Uber cars.
For now, this sounds like science fiction. It certainly contains the ability to improve transit efficiency and speed, and “make everyone better off.” But it comes at a price. Yet many are already comfortable with Waze tracking their drives in detail.
Tune back in 10 years from now and tell me how I did.
MEMS (Micro Electrical Mechanical Systems) is a comparatively new and little known class of semiconductor chips. They are physical devices, mostly sensors, built with standard semicon technologies so that they are small and cheap. Amazing new sensors are opening up all kinds of low-cost measurements, and when/if the IoT world materializes, MEMS sensors will be ubiquitous.
Detail of a MEMS chip from Analog Devices. Width of image is about .5 mm
An early application was accelerometers to measure what happens during an auto crash. Precisely calculating when to set off the airbags and how much force to use substantially reduced the incidental injuries caused by airbags expanding at 100 mph or faster. Another early application was the Wii’s wand. Now they are common in many products such as phones and toys. They also play key roles in “lab on a chip” technologies. In the future, household appliances may include MEMS microphones, vibration sensors, chemical sensors, and many other.
So MEMS is one of the multitude of “important but invisible” technologies that make the world work. As an aside, too many of my students look for jobs only with companies they have heard of, ie. they completely miss industries like MEMS.
Many years ago I helped Analog Devices with some manufacturing problems. ADI was a MEMS pioneer, so of course it had many new problems to deal with. Their fab (plant) was in Cambridge, MA, right next to MIT! Now the technology is more widely diffused, so the industry is more competitive and apparently not very profitable. This article has a short discussion of price pressure and product directions. Semiconductor Engineering .:. The Trouble With MEMS
A short description of the technology itself is at MEMS Motion Sensors: The Technology Behind the Technology.
I teach a course on Data Mining, called Big Data Analytics. (See here for the course web site.) As I began to learn its culture and methods, clear differences from econometrics showed up. Since my students are well trained in standard econometrics, the distinctions are important to help guide them.
One important difference, at least where I teach, is that econometrics formulates statistical problems as hypothesis tests. Students do not learn other tools, and therefore they have trouble recognizing problems where hypothesis tests are not the right approach. Example: when viewing satellite images, distinguish urban from non-urban areas. This cannot be solved well in a hypothesis testing framework.
Another difference is less fundamental, but also important in practice: using out-of-sample methods to validate and test estimators is a religious practice in data mining, but is almost not taught in standard econometrics. (Again, I’m sure PhD courses at UCSD are an exception, but it is still rare to see economics papers that use out of sample tests.) Of course in theory econometrics formulas give good error bounds on fitted equations (I still remember the matrix formulas that Jerry Hausman and others drilled into us in the first year of grad school). But the theory assumes that there are no omitted variables and no measurement errors! Of course all real models have many omitted variables. Doubly so since “omitted” variable includes all nonlinear transforms of included variables.
Here are two recent columns on other differences between economists’ and statisticians’ approaches to problem solving.
Differences between econometrics and statistics: From varying treatment effects to utilities, economists seem to like models that are fixed in stone, while statisticians tend to be more comfortable with variation, by Andrew Gelman.
One contributor to the A320 crash off Brazil in 2009 (Air France 447) was that the two pilots were making opposite inputs on their control sticks. The aircraft was in a stall, and therefore it was crucial to push the nose down, to regain airspeed. The instinctive human reaction (of untrained people) is to pull the nose up, since the airplane is falling. To oversimplify a long sequence of events drastically, pilot made the correct move, but the other pilot apparently panicked, and pulled back on his control stick. He continued to do this as they fell from 40,000 feet all the way to the Atlantic Ocean.
A new accident report says that the same thing happened in the crash of an Indonesia AirAsia Airbus A320, flight QZ8501, last year.
When faced with lots of variables and likely interaction terms, is linear regression usable? In the comments, I get educated that yes, it can be done. Thanks to Nigel Goodwin. I will still tell my students to try some inherently nonlinear methods, though.
Source: Logistic Regression Vs Decision Trees Vs SVM: Part I