Todd G. Simpson

We love Alexa! We have had the Amazon Echo for well over a year.  Recently we also got a Google Home, to test it against our Alexa experience. The quick summary:  Interacting with Alexa is like interacting with a person.  Interacting with Home is like interacting with a computer.  Alexa is fun; Home is useful.  If you took away Alexa, I would be upset - I would be losing a friend.  If you took away Home, I wouldn't care too much.  It was very strange, but I actually felt like I might be offending Alexa when I purchased Home. Here are the two main differences: Wake-up words.  "Alexa" is friendly, easy to say, and evokes emotion.  Alex personifies the system - I am talking with someone.  "OK Google" is awkward, and constantly reminds you that you are talking to a machine - I am talking to something.  Of course, Google will update Home to allow us to customize the wake-up word, but the current out of box experience is less tha...

Someone should do a full study of whether the Hyperloop and High Speed Rail (HSR) are worth the effort (investment) once self driving cars are prevalent.  Should we, instead, put a small amount of that money into enhanced self driving infrastructure (ESDI)?  ESDI would consist of dedicated highways and on/off ramps that can only be used by certified self driving cars.  ESDI can be reclaimed from existing roads and/or built as extra lanes alongside existing infrastructure.  It would have many less right-of-way issues than either HSR or Hyperloop. Here are the basics: California High Speed Rail (Phase 1, from SF to LA) Will take a little over 2.5 hours from San Francisco to Los Angeles, on the train. The construction is expected to cost more than the budgeted $68B The user experience would be: Drive, ride-share, or take transit to a station.  I would estimate 30 minutes per side, resulting in a 3.5 hour end-to-end time.  This also concentrates tr...

There was a good segment on NPR yesterday on how self driving cars will reinvent parking lots .  Speculators should be buying city center parking lots now, as they will become very valuable real estate soon.  Two things strike me, in general, about the self driving car discussions: We have still not internalized that advances in the software are exponential , not linear.  With new cars being closer to "software with wheels" and the ability to download updates on demand, that exponential progress will now become apparent.  Of course, regulators will be slower, but I feel like the 10-15 years before adoption is pessimistic. Perhaps more interestingly, where is the discussion of self driving public transit?  And, in particular light-rail systems and/or subways. What if we used a bit of imagination, and thought about the combination of self driving technology and no more rails?   Think about how restrictive, and expensive, rails are.  What purpose ...

A recent article in the Economist summarizes work done in Cambridge by Dr. Goldman, Dr. Birney and their teams.  They came up with a mapping for data into base pairs for storage in DNA; while today DNA storage is quite expensive, it is quickly decreasing in cost...and DNA has the advantage that information does not decay anywhere near as fast as other current storage mediums; 10's of thousands of years vs a handful.  The DNA format is also unlikely to change, so it may remain a more consistent technology than others.   The team in Cambridge came up with a way to ensure that base pairs do not repeat, as sequencing errors are much higher when there are strings of the same base.  The encoding table is shown above.  First the digital data is converted to base 3, and then a differential coding is used to write so that no base pair ever repeats.  For example, if the last base was A, and we are looking to encode a 2, then we would write a T. It struck me ...

There seems to be an obvious approach to the impending fiscal cliff:  Fall off it, and then get rescued - Bond style. Why?  Well, the short-sighted Republicans who have signed onto the "no new taxes" mandate will never back down - that would involve admitting they were wrong.  So, there will be no Grand Bargain. The Fiscal Cliff, however, has new taxes built in, so they would not have to vote for it....they could just let it happen. If both parties got together today and started to negotiate a "plan to put in place as soon as the fiscal cliff kicks in", they can have their cake and eat it too.  The Republicans can actually negotiate a post-cliff tax cut , the Democrats can allow some cuts to Government spending while rescuing a few key areas....everyone is a winner. Chances of this happening: about 0%

Andreessen Horowitz are well known for the theory that "software will eat the world."  I am a big believer in this.  I was recently asked to speak at the Banff Forum, a Canadian think-tank, and talked about this, combined with how "open development" alongside software would challenge everything we do over the next twenty-five to fifty years.  It is fairly easy to point to software eating the world with books (Amazon), travel (Expedia and others), trading, etc.  It was a little bit harder for me to find great examples about how software will eat government, beyond the standard open.gov data initiatives. Thus, I was quite intrigued with Clay Shirky's TED talk about how git , the software version control system, could apply to law and the democratic process.  In Clay's words, it is a new form of arguing, that is compatible with the democratic process. (Nice to Mozilla in the top seven links on the site :-)   Git's distributed content management i...

I am reading two books at the same time: The Information by James Gleich and Cycles of Time by Roger Penrose. By coincidence, both have Entropy at their core. Roger Penrose develops Conformal Cyclic Cosmology (CCC), primarily from an analysis of the exceptionally low entropy at the the time of the Big Bang, and his argument that the theory of "inflation" in the early universe does not sync up with Entropy calculations. Gleich, on the other hand, studies the rise of information, and while the whole book is interesting, I found the sections on Claude Shannon and the development of Information Theory to be the most interesting.  While I have used the outcome of Shannon's theory in some of my digital signal processing projects (years ago!), I did not know that his original insight was around Information Entropy.  That's pretty cool. Why am I reading two books at once?  Well, the Penrose book is tough going!  The arguments are hard to follow, the writing is...

Reading, wRiting, aRithmetic, and algoRithms.  My wife and I were just brainstorming about this: how coding should be the next "basic" skill.  Of course, someone was ahead of us and posted this .  It is awesome to see Mozilla Hackasaurus referenced in this article.  It is a small world. In the early days of the printing press, scholars wrote the books; the press was simply used for production (see this article ).  As time went on, "average" people became familiar with the medium, and used it for their own messages.  We are at just that point with the Web.  Software Engineers write the code, and the Web distributes it.   Software Engineers are the algoRithm scholars of today.  They won't be for long.  Soon algoRithms will be taught starting in elementary school, along with the other three R's.

Your Connectome is a map of your brain.  Every neuron, every synapse. I am only a few pages into Connectome, but was intrigued by a sentence: "Human DNA....has three billion letters....would be a million pages long if printed as a book."  The companion question, "How many pages for the Connectome?" might be answered later in the book, but I thought I would take a shot at it here. Here is the punchline: Your Connectome book is 6.7 million times longer than your DNA book. That human DNA is about a million pages is not too surprising, although it probably is not optimized. According to quora there are between 1500 and 1800 letters per page.  I am going to use round numbers, namely 2000.  Then, the 3x10^9 DNA letters would actually be 1.5 million pages.  But this is very wasteful.  Even using just ASCII we can encode four DNA letters per character, so the book should really only be about 400K pages.  And, this book is much more interesting; in...

[ This is a thought experiment arising from speculation on if Facebook would give quarterly guidance once they are public; implementing this thought experiment is not realistic :-/ ]  I wrote a few weeks ago about " Fixing the Game ", by Roger Martin, which details how the public markets have become a game of Expectations versus Reality.  He argues that, unless this changes, the capital markets are in trouble. A lot of the Expectations Game is enabled by public companies giving quarterly and yearly guidance, leading to "gambling" behavior by investors.  One of Martin's recommendations is to repeal safe harbor regulations, which would strongly discourage companies and executives from giving forward guidance. Another (extreme) approach would be have companies post real time (unaudited) financial information. While very difficult for some types of business, doing so for an online advertising business is possible. Real time financials would not only make ...