Unwinding the STEM - why a career in tech is not my dream for my children

As Infocomm time grows nearer (for my non-professional connections,   Infocomm is the annual tradeshow for professional audiovisual industry) I think back towards a conversation I had at the Women in AV dinner with Kristin Rector of ListenTech. We'd been talking about family and, at the moment, about my then-seven year old daughter (in a development which is both predictable and a complete wonder to me, she's now eight. Parents will understand). Recker asked, as was fitting for the occasion, if I'd encourage her to go into the audiovisual or other technical field.  It's n obvious question with an obvious answer - STEM (Science, Technology, Engineering, Mathematics) fields are, after all, the engine of the modern technology-centric economy. In an increasingly connected and tech-dependent world, we need people who understand science and tech just as much as future generations need to understand the technology which fuels their lives. So my answer was, of course, a completely equivocal "maybe. It depends."

Why? Three thoughts: reflections on what I value, the limits of technology, and the potential of the arts.

The Limits of Technology
I studied computer programming in high school and my early college years. After taking off the training-wheels of BASIC, we learned  FORTRAN and Pascal because those, of course, were the programming tools one needed in a science-based curriculum. I learned to navigate through directory trees in DOS. On the hardware side, I built some terrific High School Science projects utilizing individual logic gates on purpose-built integrated circuits - a toy car that redirects itself after collisions and a digital speech recorder with variable sample rates and playback speeds (the latter of which increased or decreased speed without changing frequency by repeating or cutting some data). Even if I hadn't forgotten more of my early programming than I have of my High School French lessons, these are specific skills which are, at this time,  quite obsolete.  

On a professional level, I had a terrific discussion about this very issue with one of my colleagues at SMW. His contention is that, while it's important for us to understand the capabilities of current technology, our greatest service is in "big picture" conceptual planning. It's nice to know exactly which product goes where, but what we excel at as consultants is creating an overall vision - and that is a much broader less narrowly technical skill.  My early AV education included learning the difference between component, composite, RGBHV, and S-video encoding. Two years ago we all learned about HDBaseT, another technology which is rapidly fading from the heard of AV system design to a side-note.  What is today indispensable will rapidly become legacy knowledge, little of which will be of any use.

Does that mean technology education is useless? Far from it. A solid education with an emphasis on theory and process can be portable, but I find it very easy to lose the forest for the trees. This is especially true for those who see the purpose of education - and this is more prevalent in STEM education - as a road to a better job. We trade the intrinsic reward of self-improvement for the extrinsic reward of a better living through application of new skill-sets. I fear - especially in the "T" and "E" facets of "STEM" - that this puts too much weight on learning to do "stuff", and not enough on understanding ideas and process. If my children are to pursue science or technology careers, I want it to be out of passion for learning how things work, not as some kind of glorified tradeschool.

Yes, we need to learn about technology. But more than that, we need to know how to think - and why.

The Power of the Arts
Two years ago I referenced a discussion I had with SMW's Rob Badenoch about IP-based transport systems. It's an interesting thought worth repeating here:

I'd said that an IP-based system allows us to create a "virtual" matrix to replace a physical one. His answer:
"no. It's not a matrix. It's making video sources available as a tool for a PC or other application. I'm sorry, that might be mere semantics."

A notebook is more valuable than a sliderule

No, it wasn't mere semantics. It was modifying the language used to describe a thing in order to better think about it in different ways. That is one practical value of a liberal arts education - we learn how to think, how to use words to shape our ideas and how to use ideas to shape the world. The ability to think about something in a different way is not only technology-agnostic, but portable to areas beyond technology.

Then, of course, there is the power and beauty in literature and philosophy. The ability to understand other cultures. The context to understand our own culture. Formal structures around which to base ethical discussions on the new quandaries technology brings us. The ability to answer "Why should we...?" as well as "How should we....?"

I've said before that, while it is the rocket scientist who gets us to the moon, it is the poet who gives us the dream of getting there.

My daughter has, as I said above, shown an aptitude for math. She's also shown an aptitude for poetry. Is one MORE worth nurturing than the other? Do we want to live in a world which is all function and no form, all practicality and no beauty? I do not. And I see the creation of beauty, even if for nothing more than its own sake, to be of great value.

What do we Value?
Another online discussion comes to mind. In the wake of baseball's annual celebration of Jackie Robinson, a commenter on New York Mets-related blog Amazin Avenue said this:

(by Ford it was later confirmed that he meant Henry, not Gerald or Betty). Bending AA's "no politics" rule until it nearly snapped, I pointed out that this is an odd, narrow, and parochial view of "most important". One major point is that it included nobody who contributed to culture; no composers, no writers, no thinkers. It also raises the question of what "important" means. Does one have to be nice to be important? One could argue that Mao has deeper footprints in history than many others listed. Yes, Ford changed the world of manufacturing. Did not people like, say, Marcel Duchamp change the direction of culture by making us question what art is? Was Kurt Vonnegut "important"? If you're a certain type of horror fan, can you escape the influence of HP Lovecraft?

In technology fields in general and the AV field in particular we create tools. Tools that can bring people together, that can make the world smaller, that can broadcast numerous visions and ideas to new audiences. Without content and without vision, all of those tools amount to deserted highways and empty containers.

We need more. We need thinkers and dreamers to guide us and to bring value to that which we do.

Would I encourage my children to enter a technology field? Again, perhaps. If for the right reasons.

But if they do, I'd encourage them to study and learn more than technology. I'd encourage them to be human first, and technologists second.

The Pixel and Ink-Stained Wretch vs. Residential Tech - My experiences with Nest and Lyric

Now that I own a home out in suburbia I have cause to sample the residential side of the tech market. As winter approaches, I'm making a two-pronged attack on the elements - low tech (weatherstripping and better attic insulation) and high (a new fancy thermostat). As this is a half-technology blog we'll focus on the latter, specifically my trials and tribulations with the Nest learning thermostat. Was this a good experience? Read on!

What to look for?
There are a few elements I look at when evaluating technology. At first look, the Nest passed most of them:

• It is physically attractive and solidly constructed.
  • This is a matter of taste, but the Nest feels solid and has a clean minimalist look.
• It has an intuitive and pleasant UI
  • Again a matter of taste; I think that the Apple-inspired minimalist aesthetic in UI has gone perhaps too far, but the Nest is not only simple to use, but gives clear feedback as to what is happening; when the system is heating, for example, the face lights up red and says "heating". The mobile and web UI - an image of the actual unit - is clear and easy to use.
• It has interesting and useful features.
  • Another win. It learns not only your schedule, but the time it takes your system to reach temperature. It connects to a web portal. It gives monthly reports on energy usage. It would be nice to have geofencing, but one can even workaround that using the IFTTT service.

There is, of course, something very important and basic missing from this list. More on that later.

Unboxing the unit was a great start; the thermostat display, base, and even a little screwdriver are packaged with the kind of care and attention to detail that says you're getting something special. It mounted to the wall easily enough and, after running some wires (I was replacing an old high-voltage thermostat which is, for the nonce, retired in place). It mounted easily to the wall, and I had an easy enough time wiring it to my system. And then the trouble started.

The first issue is that it wouldn't turn the heat on. Some troubleshooting with Nest support (available 24/7, albeit with fairly long wait times) revealed a defective base unit. Back to the store, swapped the unit for a new one, reinstalled. Success! It lit up, heated when it said it was heating. And also heated when it said that it wasn't.

Trials and Tribulations
Some quick Googling revealed this as a known and fairly common issue, albeit one which Nest does NOT disclose in any of their documentation; sometimes two-wire setups as the unit inadvertently calls for heat when charging its battery. No problem; I'm a technology professional. It's easy enough for me to add another wire, add a low-voltage transformer and... voila. No more power issues. It only called for heat when it said it was calling for heat, battery voltage remained consistent, but there was a new problem: the temperature on the display had no relationship to the temperature in the room. It read anywhere from dead-on to 7 degrees warmer.

After more calls to tech support then offered to send a new base which they assured me is a different model than the ones stores might have. While I appreciate this, the fact that the units in stores have a known issue but have not been recalled and replaced is another warning bell. If you know that the units in the field don't work, it's a bad idea to let people get their hands on them and sour their image of your product. The good news for Nest is that prior versions souring the experience ended up being a non-issue for me, as the new unit did exactly the same thing; cool in the house, just under 70 on the spirit thermometer built into the old thermostat, 78 degrees on the Nest. It was not 78 degrees.

This brings up the qualification I missed above:

• Performs its primary function.

One of my last trouble-shooting calls with Nest support went something like this:

They: Sometimes the WiFi makes the temperature read higher. I want you to turn off your WiFi on the thermostat for about five hours, see if the temperature normalizes.
me: What is the long-term solution here? Without the WiFi the thing isn't really all that useful.
They: You might have to change something on your router.
Me:  What would I have to change on it?
They: (pause) -- uh... settings.
me: (through gritted teeth): Which settings?
They: Maybe the channel?

It's much warmer six inches to the right.

Everyone at Nest tech support was professional and courteous, but they clearly didn't have a solution. A Google for "Nest Reading..." autofills for "wrong temperature" "high" and "too high" as four of the top five results.

Hardware is Hard

The barriers to creating and marketing hardware have never been lower, with "internet of things" as the latest buzzword, crowdfunding for sexy ideas and the possible reward of a buyout from a big player (Nest was purchased by Google for a billion dollars). The problem is that creating and testing good hardware isn't always easy. Is is a subtle design flaw in the Nest that makes it heat up sometimes and not others? Perhaps. What's clear is that features and even UI can be easier to design and implement than solid, reliable performance.

How does this story end? I took the last Nest back to the big box store from whence it came, returned home with a Honeywell Lyric. It doesn't have the fancy learning algorithm like Nest, nor is the UI as nice or slick. What it does do is accurately control the temperature, taking advantage of Honeywell's decades of experience in the field. 

I'm sure that with the money available, companies such as Nest will eventually catch up and produce hardware worth the premium they charge. That time, alas, does not yet seem to be here. Hopefully Lyric will release an API in time for my next big project, to be discussed sometime next year.

SVSI N1000 series - Uncompressed video over network switches

Teaser post showing new devices,
not yet tested or even wired neatly.

Two weeks ago, before a much-needed week-long vacation, I left a teaser post showing some shiny new hardware on the Audiovisual Professionals group at the Google+ social network. For those not following me over on G+, this is the N1000 series from SVSI - a video and audio over standard network switches. It's a technology in which I've been interested for a while, but true uncompressed video has always been a big missing item; if you have a source and display in the same room, you want to eliminate the latency that comes with compression as much as possible. Does this solution fit the bill? Time to find out!

My test rig. HDCP protected content from the tablet,
non-HDCP from a laptop. And I'd not be the pixel-and-ink
stained wretch without an actual bottle of ink!

Physically, each encoder or decoder unit is about 7.875"x5.125"x 1" (according to my tape measure here). This fits them comfortably behind displays, in wall-boxes sized for digital media receivers, and other locations you'd want to stow one. There's also a 1RU rack-mount kit for two units, and a 2RU card-cage which will hold six card-versions of the same units. Encoder units have a DVI-I input for video with embedded audio, a single network jack, as well as captive-screw connectors for IR, RS-232, Audio, and DC power. The units can also accept power over ethernet, which is how I tested them. The decoder units are similar, with the exception of slightly different placement on the DVI connector.

I was able to get started and have a functional matrix within probably less than a half hour, using SVSI's Conductor Netlite software. It auto-dected all four units (plus a controller - more on that later) without a hitch, and easily populated them into a 2x2 matrix with reasonably intuitive mouseclicks to select crosspoints and then "take" to transfer. Separate matrices are created for N2000 and the forthcoming N3000 series devices. Sadly, I didn't have any of those to evaluate as of yet, although I'm working at acquiring an N2000 kit.

Front and rear views of encoder with network switch

So how was the switching? As fast and close to seamless as I could detect. Running tasks through the system (drawing on a tablet, using a mouse, etc) felt as natural as they would via a direct connection. Conductor will also let you rename units, give you their IP addresses, and send you directly to their page on a web-browser to adjust various settings, see extended EDID information, set up local play, etc. This is also where you'll find the ability to allow HDCP protected content, which is a quick and painless process. Once you do so, switching between protected and unprotected sources is as fast as any other switch. Compare this to the Extron XTP matrix which took well over a second for the same task.

Decoder and control processor

Along with the encoder and decoder units I got my hands on an N8001 controller. This appears to be nothing more than a small webserver, allowing control of a system via a web browser. IN addition to the familiar auto-discover and matrix controls, there are script editors and a "panel builder" utility for creation of a control interface. Scripts can direct video streams (either by IP address or the unique stream number the assigned to each encoder), direct audio, operate transport controls for SVSI's DVR appliances (not available for the N1000 series), or windowing processor (ditto) as well as embed other scripts, send RS232 commands, switch from "live play" to "local play" (calling up static images stored on the device) and give a delay between commands. Custom button graphics can be uploaded, or a standard squarish-button with rounded off corners labelled as desired. For relatively simple systems, it does the job quite well and is easily loadable to an Android or iOS device via an app. IT doesn't seem to have an intuitive enough design interface or enough options (conditionals, page-overlays, etc) to build really complex systems, but it seems very workable for simple interfaces. I was able to build a simple panel to route either source to one or the other destination very quickly and easily.

Concerns? There are a few. While the devices switch very quickly, they boot up very slowly. From a cold reboot of one of the encoders or decoders, it took a solid two minutes between plugging in the cable and having the unit recognized on the network. What's worse, it didn't always get the video stream back or sync with the display without re-sending the control command. There's also an odd delay and, on some displays a loss of sync, when switching from "live play" to "local play". Sync was re-established, but it took several seconds and some odd color-artifacts and vertical roll. This was odd.

Secondly, and not unexpectedly, is the issue of bandwidth. Each stream is 880Mbps. I have no idea how this would be able to scale up to 4K; it doesn't seem that there's be enough bandwidth available. Will we need to start deploying 10 Gigabit switches? Will this spell the end of uncompressed video over the network? Will we need to send multiple streams and stitch them together? The drawbacks to all of these solutions is obvious.

All told, it's an intriguing set of options. Pricing - especially for a large-scale system - would be far less than a similar HDBaseT solution with fewer proprietary parts, less rack space, less power draw.