Wednesday, February 15, 2017

A Hypothetical Analog vs. Digital Scenario

I have concocted a hypothetical situation in which analog beats out digital.  And it doesn't even require a zombie apocalypse.<p>
DPReview reports here on Nikon's significant financial problems.  They reported earlier that Sony's image sensor sales for smartphones are up but their camera sales are down.  So let's imagine a world in which the only cameras manufactured are smartphones---no large-sensor cameras.  We have a long way to go before we get to that point, but there is some small chance of that eventuality occurring.<p>
Smartphone cameras are quite good these days.  They now use high-quality sensors and optics coupled with sophisticated cameras running on the smartphone's high-performance embedded platform.  But small sensors have their limitations and for some applications, bigger is better.  We can expect some demand for large sensor cameras for the forseeable future.<p>
But how do you maintain these cameras?  Even ignoring firmware and driver updates, the hardware will fail eventually.  Both the image sensors and image processors are specific to the camera.  Once those chips go out of production, we are reduced to relying on a store of old chips stored in liquid nitrogen for replacement.<p>
Analog cameras, meanwhile are much easier to maintain.  Shutters are probably the hardest item to rebuild, but even those can be manufactured in a modest machine shop so long as the machinist understands how shutters work.  Otherwise, a camera is a box with a hole on one end. <p>
And although I can't make my own image sensor (even though I have access to a semiconductor fab line, it would be a very difficult and expensive process), I could make my own glass plates.  If Matthew Brady can make glass plates in a tent on a Civil War battlefield, I think that I can manage to make some for myself.
Digital systems have a lot of advantages.  Long-term maintenance is not one of them.  Specialized parts eventually fail.  The manufacturing systems used to build them also go out of service.  When we use computers to build long-lived systems, we need to think carefully about the future.

Thursday, February 9, 2017

Early Circuit Degradation in Atom

A post from Anandtech here describes what seems to be an early circuit degradation problem in Atom C2000.

Tuesday, February 7, 2017

Article on TVs that Spy on Viewing Habits

Extremetech reports here on the extensive spying on viewing habits and system configuration characteristics performed by Vizio TVs.

Friday, February 3, 2017

What Are You Buying?

A widely held principle in the law is that software is intangible; when you pay for software, you are purchasing a license to use that software. The law generally holds that software is fundamentally different from tangible goods.  Of course, we have discussed in an earlier blog post about the physical representation of software.  My book The Physics of Computing uses the computation of the mass of a bit as an example (Q = CV; m_bit = m_qQ).
But the line of reasoning of the legal profession has a more fundamental flaw.  When I buy a car, no one bothers to consider whether I am purchasing a license to make my own copies of the car.  Everyone agrees that I am buying a physical object, no questions asked.  In the 20th century, when building a car was a hard, long job, this scenario was a no-brainer.  But 3-D printing changes the equation.  We can assume that someone will, sooner or later, scan an entire car and 3D print a copy.  So where is that fundamental distinction between cars and software?