E-mail should be like Electrical or Gas Service

I have had a horrible experience with my e-mail vendor. Yes, you might ask why I am not using Google or something but I am not and it is mostly a historical thing. Anyway, they decided to migrate to a new mail server architecture. This apparently justified essentially screwing up everyone's e-mail for a week or so: first flaky and then dead altogether and a slow recovery. If they were shooting for five-nines then they will need to be failure free into the next millennium in my opinion. Anyway, to make things worse, during this "planned outage" e-mails from the same provider were not being delivered about the fact that one of my domain names is about to expire. Now I know that I am certainly to blame for not renewing early and all of that but the consequences of my failure were dramatic: they shut down my site, tossed the domain forwarding information, and started sending undeliverable bounces back to whomever was trying to send me e-mail, meaning, of course, that I would not get that e-mail. It took me a bit of time to figure out that this was not another symptom of the "planned outage." To make things worse, they charged me a $25 reinstatement fee. I will be changing providers for that simple injustice.

What is the point of this rant? It seems to me that E-mail is as vital in our modern age as electricity and gas and the providers should view it as such in all of its ramifications.

Firstly, and most obviously, it should be utterly reliable and when it fails there should be a darn good reason. For example, our electricity occasionally goes out and it is usually essentially always an act of nature - a powerful windstorm or an avalanche or ice in the mountains where the power lines bring hydro power to Seattle. I have never in my life experienced a "planned outage."

Secondly, when a customer screws up they you don't just rip up their environment - you give them grace time. I have screwed up more then once and have failed to pay my gas bill and the gas company sends ominous warnings but acts slowly to turn off the gas. This is as it should be - people screw up and the consequences of a single screw up should be small, not catastrophic.

Five nines, Live Tennis, and Trust

I don't have a TV so when I want to watch some sports event, I usually go to my local pub to watch and have a few beers until I have satisfied my thurst for sports. Being somewhat of a tennis nut, Wimbleton is a definite event that I enjoy watching. Unfortunately, in Seattle, the final started at 6am - not really a time for a trip to the pub. So instead, I went online to NBCsports and was pleased that they were "live-casting" the game over the Internet. While the resolution was not great, it was adequate given the circumstances.

That is, until during the final set the browser starting showing an ominous "reconnecting" black rectangle in the middle of the image. It would then repeat the previous 10 seconds of the match, show "reconnecting", and then show the same 10 seconds again; kind of like the movie "Groundhog Day". Eventually, the browser went black with a notice of technical failure and a suggestion to try again later.

I think we as humans watch live sports because we want to imagine ourselves at the event itself and "liveness" gives an approximation of this. Even when the event is delayed by a small amount of time, as the browser seems to do to properly buffer the data, is still seems "live." But when the whole thing collapses we are annoyed and frustrated. The "technical difficulty" was eventually resolved but I missed 30 minutes of the longest fifth set in finals history. Worse, the sense of liveness was totally destroyed and I pretty much gave up, thinking to myself: "I'll watch the rest later."

Five nines is the way people like to talk about the level of reliably - five nines means that 99.999 pecent of the time a service is up. To recover "five nines of reliability" after a 30 minute technical failure would seem to imply that the service now needs to go uninterupted for the next 5+ years. Few services are five nines but the goal is there because the utility of a service is tied our degree of trust that it will reliably function. TV, for example, is much more reliable then "Internet TV."

Other items in our life are surprisingly reliable. Cars are a good example. Even in those cases, we often choose a car maker based on the level of reliability that the manufacturer has demonstrated. This is a good indication of how much we value reliability. The struggling US automakers did not seem to have taken seriously enough this expectation and when the Japanese attained a level of quality that sufficiently surpassed the US makers, many Americans switched and never went back.

In my opinion, until "live Internet broadcasts" attain a significantly higher level or reliability, they will not attract much of an audience. This problem plagues many "utility" aspects of Computer-based services. The company that takes five nines seriously might expect to reap significant benefits.

A pristine machine, remote desktop, and unity

I finally had to wipe clean my Windows XP 64-bit edition. I had no idea that Microsoft would orphan an OS to the extent that they have this one; so far as I can tell, the only way to "upgrade" is to do a complete install of a different OS. So that is what I did. I dreaded having to install all of the apps that I had before. Then I had an epipphany - why install anything other then a web browser? So I decided to do that and use Remote Desktop to access my e-mail from my main development machine in my office.

There is something quite lovely about this setup - e-mail is totally readable in this manner and I no longer have two Outlook in-boxes to cope with. There is a nice feeling about having a minimal machine - perhaps this feeling is really a response to the crufty state my machine had gotten into over a number of years. I'm sure that I will install more apps but for now, I am enjoying the simplicity.

I wish that Remote Desktop had a "unity mode" ala VMWare. This would allow a smooth integration of remote and local functions. Arguably, this should be easier then Unity if the local and remote desktops are the same OS - one can have a notion of local and remote apps and you simply launch them locally or remotely and that is that. You might want to merge the desktops.

The bliss of this situation will of course vanish as soon as I have to go out of town but perhaps I'll simply use web access to my e-mail. Alternately, I could screw around to make remote desktop accessible from the web - something that seems worriesome.

The main realization I have from this experience is that my laptop is for the most part an e-mail and browsing machine. And yet, it would not be adequate for it to be limited to these functions. Remote desktop gives me a window into a richer environment and that may be all I need - time will tell.

In contrast, my development machine, which has three screens that make it a much more "information expansive" machine then the laptop. The cheapness of LCD monitors has made this setup really very practical (I tend to by the monitor size that is just one or two steps below the "state of the art" and, as such, vastly cheaper). It is interesting to me that the many monitors environment really makes a laptop (even one with a large screen) seem ackward for any kind of serious development activity: I do mechanical, hardware, software, and some graphics design and it appears true for all of these scenarios.

I now have four major "computer-like" devices that are in different form factors:

iPhone
Kindle DX
Laptop
Desktop with many monitors

I can imagine the Kindle DX and the laptop merging in some way - particularly if remote desktop were sufficiently performant. The iPhone and desktop appear to hold useful niches - very compact and information rich. This makes me think that this "middle area" will be the most exciting one in the near future.

My silly shoes: design, spectacle, and embarrassment

Here are my new shoes. A friend (Kevin) got a pair and they seemed like an excellent idea. They have no support, they give your toes exceptional freedom, you can move surprisingly like you were in bare feet, and you are protected by, albeit thin, soles from pointy things that annoy your feet. I have worn these playing tennis (limited to an hour at a time out of caution) and they are great. I think that I would wear them every day were it not for the spectacle that they cause and the derision that they inspire in people. The laughs and surprise is, well, surprising to me. It is true that I have big feet (size 13) and so these shoes do look like the gorilla feet that you might purchase at a costume shop. I have had to spend quite a lot of time explaining these shoes to people and deflecting silly remarks.

At the end of the day, this is really a design issue. I can imagine one of two possible chains of thought by the designers: ambivalence or intentionality. I can totally imagine, if the designers are out-of-doors types trying to improve on Tevo's or something that they might completely miss the utter strangeness of these shoes. This ambivalence to the consequences of design (me being forced to explain my shoe-ware) will likely narrow the audience of these shoes to those who are willing to put up with the spectacle (I pick my battles). An alternate interpretation of intent is that the designers are fully aware of the weirdness of their design and are going with the "let your freak flag fly" approch to design. I think that is a mistake. My reason is simple: the shoes are very comfortable and it seems reasonable to offer a style that does not require one to become a walking advertisement.

To me, the solution is trivial: put a flexible shroud over the toes so that the most peculiar aspect of the shoes - the individuated toes - is not readily apparent. I'm not sure that this would entirely eliminate the spectacle but it would at least require a more observant bystander.

I am reminded of a sitter I had when I was a little boy. He had Birkenstock's when they were a totally new thing (probably the late 1960's). I remember thinking - wow, those are weird and his reply was very much what I find myself doing: explaining why they made sense. Perhaps, this is the path that excellent design ideas must travel - eventually, they will prevail and become the norm; perhaps they need to hold their ground to make their point. Perhaps I am too old to "let my freak flag fly" without complaining!

Looking for a Good Tennis Book

I'm looking for a book on tennis. I go to Amazon, the subject index gets me down to 1000+ books sorted by something called "relevance." The books don't look interesting and I'm not going to scan 1000+ books for one that is. What should I do? I switch the sorting method to "Avg. Customer Review" but this gives me a ton of five star ratings by a single person. So, if I am the same as this other person then maybe this is a good book for me. I don't want to try to find my tennis doppelganger. I'm wishing that Amazon had New Egg's "most customer reviews" - definitely dangerous because someone could game the system but it eliminates most of the one review wonders. I go to Google Books but it is hopeless. I use Google and search for the "best tennis books" but the results are not helpful. I try the new Microsoft search ... after I try www.ding.com, which is not really the correct name. What was it? Oh yea, Bing (actually, I used Google search and asked for "Microsoft Search"). I got pretty much the same junk that Google gave me.

Perhaps the web has no suggestions about good tennis books - I kind of doubt that. The real problem is that search really does not understand what I am asking - there is no semantic anything that might suggest that I am looking for a good tennis player's opinion of tennis books. It does not need to be the best tennis player, it does not need to be someone famous, etc.

I came up with an aphorism for this problem:

With context, Artificial Intelligence is easy

How to infer context, then, would seem to be the "fundamental problem of artificial intelligence." I have no particular insight as to how to crack this problem but until it is done I will be looking for good tennis books by linearly scanning all of the books Amazon has to offer.

Aesthetics, LED Technology, and Change

I'm sure you have seen a chandelier or sconce that attempts to look like a candle with a "flame shaped" bulb and a long stem that has faux wax drippings and maybe a little bowl shaped bottom to catch the faux wax. Similarly, there are lights that look like gas-lights. These are a kind of transitional objects that gets invented at the transition of technology generations. For some reason they remain as an aesthetic for much longer.

I am of the Frank Lloyd Wright school of materials: all materials are good but you should not try to turn one material into another (for example, plastic with wood grain). These transitional objects violate a similar principle: don't try to turn one technology into another. This is generally not possible when an older technology is deeply entrenched and you have to live with the form factor that it has created for itself. Still, it seems that true good design emerges when the old is abandoned completely and the new embraced for what it is - it's essence.

This transition is happening with LED lighting. The early uses of LED lighting will probably be in replacing incandescent and fluorescent lights while living within the constraints that these lights demand. In the case of incandescent, the form factor addresses the need to deal with the significant radiant heat generated by the bulbs; in the case of fluorescent lights, it is their space-filling nature. LED replacements provide few aesthetic benefits: an LED-based can occupies essentially the same footprint as a normal can - the selling point is the lifetime of the LED itself and the lower energy cost.

There are interesting and novel uses of high-intensity LED. One interesting use is when red, green, and blue LEDs are used to make the light and, as a result, it is possible to cover a wide spectrum of colors. I first saw this as a back light to a bar in Seattle - it was cool as the color would change periodically. This is a dramatic use so it is not particularly refined.

A more interesting, future use, will be when interior designers realize that they can tune the color of lights to match or contrast the colors of fabrics, carpets, etc. Interior designers have quite a sophisticated notion of color and the ability to control the color content of lighting should be an interesting and sublime development. This notion can be extended to changing the light with the change in illumination from sunlight. In our living room, the walls are green but the more sunlight there is, the more yellow the walls become. One can imagine adjusting the ambient lighting to either enhance or counteract this change. Achieving such effects will require photo-sensors that can monitor the light levels in parts of the room.

The dramatic power of lighting is particularly well demonstrated in movies. "The Red Shoes" is an excellent example or the lighting designer's art as are most of Vincente Minnelli's movies (he was a set designer before becoming a director). It is not clear that we would want to live in a "dramatic" environment, but who knows.

The most direct change that we should expect from LED lighting will be the change in lamps, ambient, and task lighting sources. High-intensity LEDs are very intense point sources of light and this produces a significant challenge if you want ambient light. Maintaining their efficiency will require clever diffusion technologies. Because of the low heat generated by the LED itself, one can imagine plastic lenses and reflectors. This is not something that is practical with incandescent bulbs because of the radiant heat. The form factor of an LED - even with it's heat sink - is very small: a common module is a single high-intensity LED mounted on a 3/4" radius PCB with an aluminum back that is screwed against a heat sink. CNC machining will allow heat sinks to be organic forms as opposed to the ordinary notion we have (rectangles with fins).

One comment for technical completeness is that high-intensity LEDs generally require heat sinks. This is not so much because they generate much heat but rather because an LED is a power semiconductor that has a very high heat density over a very small area. The semiconductor junction will not operate properly at temperatures above around 85 degrees centigrade. The heatsink is to keep the junction sufficiently cool.

It is an interesting thought experiment as to how LED lighting will alter our idea of the aesthetics of lamps, ambient, and task lighting. The possibility of "reactive lighting" that can alter color schemes and react to sunlight intensity levels will provide for a broad range of innovations.

A Speech Recognition Conundrum for Old Folks

I while back I purchased Dragon Dictate with the fantasy that I might be able to convey my thoughts in the same manner that I talk. Anyone who knows me knows that I like to talk so this seemed like a rational idea. Unfortunately for me, the experiment did not work: I would sit in front of my computer and my mind would be blank; if I forced it then what would come out would be trite drivel. Then I thought "this is happening because there is no spontaneity" and I got a hand dictation device that Dragon would apparently then translate into words for me. Again, blankness and now the dictation device and the software languish.

I conveyed this amusing failure to a friend (wrs for those of you who know him) and I vaguely recall him saying something to the effect that "I think with the keyboard and not with my voice."

This is an interesting aha if we consider that having trained our hands to do the work of transcription that we have perhaps diminished our ability to use our voices instead. For old folks like me this then would be a natural barrier to the use of speech recognition: for it to be useful I will need to reverse my bias toward using my fingers. This is presumably an old folks problem in the sense that our children (if you don't stick them in front of a keyboard too soon) may find speech recognition more natural.

I came to this thought from another chain of ideas that I track - diet related issues. A friend (Brent) sent me an interesting post whereby the observation was made that we look for genetic causes when there is no variation in behavior. For example, most of the developed world shares a similar diet and instead of looking at the flaws in the diet we all share, we look for genetic causes for some of its consequences (heart disease, high blood pressure, type two diabetes, etc.)

Speech recognition may be suffering a similar fate - we live in a world of keyboards and the failings of speech recognition is that we type.