Motion theory, markets and the speed of information

• By Chris Maresca

One of my side hobbies is building small CNC machines.    Some part of this requires a pretty deep understanding of motion theory and, in particular, the interaction between physical hardware and computer systems.  The major problem in computer controlled motion systems is that, while computer commands to move are virtually instantaneous, motions of physical systems take some time.

One would think that the solution would be to measure this lag, but it’s actually much more complicated than that.   Part of the motion problem is that the lag is highly dependent on various environmental factors, such as load or amount of wear in the system.   In order to deal with all these problems, motion control engineers have developed something called PID feedback. Read more…

Deficit Spending for Profit

• By Chris Maresca

An interesting thing happened last Monday.   The yield on 3-month Treasury bonds went negative.  For most people, that statement is basically meaningless as bonds are pretty much black art.   The simple explanation is:

People are paying the US government to take their money.

Usually, it’s the reverse.   The US and other governments need to pay interest on money they borrow from people, usually at a rate higher than inflation.   Rates are usually priced so that they are sure to sell, but, in this economic climate, they are essentially priced at zero, less than zero if you take inflation into account.   Having the yield go negative, however, means a lot more than being priced at zero. Read more…

Open Source – The Walmart Effect?

• By Chris Maresca

There’s been a really negative meme going around in the last week or so about Open Source destroying the ability to make money from selling software.   It’s patently nonsense and quite the contrary, Open Source saved the software business.  Here’s why.

1. Software development as an act of alchemy

Making software has always been a complicated thing, somewhere between engineering and art.  Software developers, while outwardly pretending to be engineers, are much closer to music composers and what they produce is, much like music, rather nebulous and intangible in quality.   Steve Jobs famously said that great software was “indistinguishable from magic”, ergo something much akin to alchemy or some other opaque creative venture (like music). Read more…

Bad times good for Open Source?

• By Chris Maresca

There is a certain meme floating around about bad times being good for Open Source.   I’ve heard it from a wide variety of people, from project maintainers to execs at large companies to VCs.   On the face it, it would seem to be an obvious statement, that Open Source does well when people are squeezed economically.

I’ve seen this first hand in work I did for the US Navy.   As budgets were increasingly re-directed to warfighting, critical but non-combat systems were increasingly developed with Open Source technology as a way to alleviate the funding gap.  In one case I know of, a portable system budgeted at 20 units doubled it’s deployed units by switching to an Open Source architecture.   That and other successes allowed more widespread acceptance of Open Source, with resulting wider adoption.  That said, in the last downturn, I also saw a lot of executives use the threat of Open Source use as leverage in contract negotiations with proprietary vendors.  In actual fact, they probably would have never deployed any Open Source…. Read more…

Precision vs mass manufacturing

• By Chris Maresca

I’ve recently gone on a watch buying spree*.   For years, I used my cell phone as a watch, but I’ve recently come to realize how much I actually hate doing that.   It’s hard to be discreet about it and it just screams ‘I’m bored’ in a meeting.   Plus, watches have experienced a renaissance of sorts in the last 10 years and there are some really interesting new ones.

Of course, since I’m a technologist, I just had to understand how these things work.  The quartz ones are pretty obvious, although there is still a surprising amount of analogue, mechanical technology in even the most modern quartz watches.   Electrically driven watches were actually pioneered by two American companies, Hamilton and Bulova.   Pure digital watches (with numeric displays) were developed by Hamilton in the early 1970′s and lead to the digital quartz revolution (which is a fascinating story in itself, but for another time).  Up until this time, watches were expensive jewelery that one bought perhaps a 1/2 dozen times in your life.   But, as a result of mass manufacturing techniques pioneered by the semiconductor industry, companies like TI and Casio were able to crank out millions of cheap LCD watches that anyone could buy. Read more…

When electric cars were the future

• By Chris Maresca

I was reading Car and Driver’s commentary on electric vehicles recently, which was as follows:

“The media are making all kinds of noise lately to the effect that electric cars are coming, that they’re going to help us kick our imported-oil habit, and that you’ll be able to drive them for pennies a day.

A company that can develop a non-petroleum-fueled car palatable to the masses stands to make a pretty good buck.  That’s why GM will shell out some undisclosed number of billions on electric vehicle development during the next five or so years.”

Sounds about right and probably true too, nothing all that revolutionary about any of this.  This is precisely what, in addition to GM, Tesla and Fisker are trying to do.  Let’s read on:

“It’s a tall order, but GM is already well on the way to pulling it off.   Whether the buyers will be there, however, is a question GM is still struggling to answer.

A study commissioned by Gulf & Western predicts that we’ll have something like 34 million EVs – about one quarter of the national fleet – on the road by the year 2000. GM … has publicly committed itself to mass-producing electric cars by the mid-to-late 1980s – probably 100,000 per year or more.”

Wait.  What?  Actually, the Car and Driver issue I was reading was dated June 1981.

Fascinating. 

Fast forward 27 years later, more than a quarter century, and the meme is once again in the air:  electric cars are the future.

Are they really?  Even in 1981, just after a period where the US almost ground to a halt due to lack of oil, there was some skepticism about relying on electricity:

“In fact, compared to gasoline, electricity is a downright ludicrous power source for cars.  In terms of volume, it takes 350 liters of lead-acid batteries to match the energy in one liter of gasoline.”

While we no longer rely on lead-acid batteries for portable power (can you imagine one of those in your laptop?), the fact remains that gasoline and diesel are some of the densest forms of power ever discovered.  Electricity is not nearly as portable or dense.

And this may be at the core of the problem for all electric vehicles for some time to come, even if it is just a talking point for most people.  This is because the average person drives daily average distances well within the range of a Tesla or any of the proposed EVs.

Still, Americans like their freedom, and part of that is freedom to roam, without a tether (e.g. limited range, 4 hour recharge).  In that sense, nothing has changed in a quarter century.

In fact, not much has changed in the last 18 years.  You see, 18 years ago, VW showcased their new hybrid technology at the 1990 Geneva Motor Show.  It was to be available in every VW model in the next 2 years.

There are a lot of things that can be learned by reading the state of the early ‘80s electric vehicle market and from VW’s hybrid introductions from the early ‘90s, but one important potential lesson is that oil prices are volatile and cyclical.

Alternative energy gets great boosts when energy markets go haywire and mostly falls down spectacularly when markets calm down, often taking billions in R&D with it.

The companies and technologies that do survive are the ones that are sustained through the troughs, either by prescient companies or by smart government support.  In the United States, both of these are in rare supply.  This goes a long way to explaining why GM is not a significant player in the ‘green’ vehicle market, and why 20% of our electricity is from hydrocarbons and 46% from coal.

Perhaps it’s time to think about energy technologies on a slightly longer timeline than quarterly results and presidential election cycles.   If we don’t, in 2030, we’ll be looking back on 2008 like we are now looking back on 1981.

NB: Just more of a poke in the eye from 1981… Smack in the middle of the electric vehicle special is an ad for the 1981 Mitsubishi/Dodge Colt, EPA est. fuel mileage 37 city, 50 highway.  30 years of technology improvement and we’ve gone backwards.  Even a Prius is not that fuel efficient.

Free WiFi harder than expected

• By Chris Maresca

I can’t say this was really a surprise, but MetroFi just folded.   They sent an email out to subscribers that the service was going offline June 20th, 2008.   Chances are, if you spent any time in the South Bay and have a laptop with WiFi, you ran across MetroFi’s service.   

MetroFi’s plan was to subsidize free WiFi with ads and offer a faster, premium ad-free service for a fee.   Apparently, they were not able to make this work economically, despite the fact that a whole lot of places were looking to roll out muni WiFi.   One wonders if these wide-area, ad supported networks will actually work since there have been a bunch of failures with this business model in the last few months.   Even here in San Francisco, which is attempting to roll out a muni WiFi network with the help of Earthlink and Google, not much has happened.

And, even if it eventually did, it seems that there is little utility in this.  After all, most cell companies provide blanket wireless coverage at increasingly fast speeds for a nominal fee.   By nominal, I mean that if you are using this for business or it’s otherwise important to you, an additional $10/month should not be much of an issue.   And the price for this has been dropping for years.  I can remember paying $130/month of less than modem speed not too long ago.   Chances are that trend will continue and data will just be bundled with your phone.   Plug it into your computer and, voila, instant internet access.   My expensive Blackberry plan already operates this way, but it’s twice the price of a generic cell phone plan.

So, given all this, who is going to use free WiFi?  There is an argument to be made around universal access and the value that brings to communities, but this might be better served by different technologies.   San Bruno, California, for example, provides low cost internet access through it’s cable system.   In Japan, Korea and Sweden, Internet access provide by telecos is far cheaper than in the US, so perhaps there is a fundamental problem with the overall pricing model that might be solved through other creative means, such as re-purposing parts of the universal access fees, or using the $12/month I pay in San Francisco cell phone tax to subsidize DSL/cable access for people who can’t afford it.

In the end, however, it seems that ad-supported WiFi may be a non-starter.   People who really need on-the-go Internet access will pay a little extra on their cell phone bills, people who use it casually will rely on places like coffeshops and hotels for free access.   In that view, it doesn’t seem like there is much room for ad supported free WiFi and, even if there is a demand, it might not be the demographic that advertisers are really looking to reach.

High power in small things

• By Chris Maresca

In my spare time, I’ve been experimenting with a new generation of microcontrollers.   Microcontrollers are at the core of modern electronics, and are basically full computers the size of a postage stamp.   Following Moore’s law, these things have been getting faster and faster.   In the last five years, they’ve become almost full computers and the cost has fallen through the floor.  It’s now possible to get a chip that is the functional equivalent to a early 90′s computer for less than $1.

The downside of many of these chips is that they are typically complex to program, requiring rare combinations of deep hardware and software skills.   Traditionally, the way around this has been to buy a development environment (aka, software for specific hardware) which costs thousands of dollars and usually has associated royalty costs.  Not much of a solution for the hobbiest or quick, cheap solutions.  There has been some light at the end of the tunnel in the form of a new generation of open source embedded operating systems such as FreeRTOS, EtherNut and eCos.  While these remove a lot of the software costs, they still require a lot of skill to make functional, as developing for these systems is largely a C-language, bare metal affair.

So, the hardware is cheap, the software is open source, but the learning curve is still huge as you need to be an expert in two disciplines.  This puts most microcontroller development out of reach of most of the developer community.   Once again, Moore’s law is coming to the rescue.  I recently discovered that microcontrollers capable of running mainstream operating systems such as Linux have dropped to a price point just around a traditional high-end microcontroller.   This is particularly true for ARM-based microcontrollers.   Offerings from ST, NXP and ATMEL are currently priced at between $3 to $8 (in quantity) depending on features.  

Why should you care?  Well, microcontrollers run everything.   From your microwave to your car to the elevator, pretty much any place automation is needed, microcontrollers apply and run logic in systems. Linux-based microcontrollers open up this embedded world to a whole new field of developers, one’s who’s skills are largely software-based rather than hardware.    Part of the reason that Google is able to build and deliver a software-based phone platform is that a lot of phones now use ARM9 microcontrollers on which Linux can easily be run.   Another interesting example is Wiring, an open platform developed by  Hernando Barragán at the Universidad de Los Andes in Colombia.   Wiring is designed to be an easy to use bridge between software and the real world and is targeted at artists, although you can use it for anything.  These are just the tip of the iceberg. 

Who knows what people will do cheap, embedded Linux, but whatever it is, it’ll be very interesting.

Patent rulings since 2000 invalid?

• By Chris Maresca

In a very strange twist, a law professor John Duffy discovered last year (PDF) that, since March 2000, the appointments to the Patent Board may have been unconstitutional.   The net effect of this is that nearly 2/3rds of the Patent Board judges are not legally appointed and their decisions could be invalidated. 

While the details are very much deep in constitutional law, it basically stems from the fact that because members of the Patent Board are judges, they have to be appointed by the President or one of his officers.    In March 2000, Congress changed the law to allow the Director of the USPTO to appoint Patent Board judges.  Somehow or another, they missed the fact that the Director of the USPTO is not a constitutional officer and could not appoint any judges….

Why is this important?   Well the Patent Board is responsible for reviewing and judging appeals to the patent process, such as evidence of prior art (which sometimes invalidates patents) and other disputes.   These happen all the time, so there are a lot of rulings in jeopardy as a result of this.  And the only way this can be fixed is that Congress amends the law retroactively to make the head of the patent office a constitutional officer or legalizes all the judges.   Either way, it’s a complete mess and lawyers are poised to take advantage of it.  According to Professor Duffy, there is every chance they will succeed as the government has never refuted his analysis:

"You shouldn’t take such a position — saying the whole board is unconstitutionally structured — unless you’re pretty sure. I thought this was incredible. I checked it every single way I could before I went out on a limb to say this. … The government has never argued I’m wrong. "

So, could a court invalidate a large chunk of the US intellectual property infrastructure?   Well, maybe, but probably not.   Chances are lawmakers will step in before large amounts of damage is done, but it may take some time.   In the meantime, it sure looks like a lot of patent rulings made between 2000 and now could be invalidated, although the lack of any cases since the discovery shows a certain conservatism on the part of patent litigators.   Perhaps this is too much of a nuclear option.  That said, if Translogic Technology v. Dudas (see previous link) succeeds, then it will open the floodgate to more litigation.

Note: I am not a lawyer, I don’t even play one on TV, so you should really take this whole post with a grain of salt.