Wednesday, November 18, 2009

BECC 2009: Day 3

The last day of the 2009 BECC Conference is just a half day, so I attended a session on energy competitions and a keynote regarding energy efficiency in the US economy.

The energy competition session kicked off with a presentation on Oregon's Home Energy Makeover competition.  Last year, 4 homes were picked out of 6,000 entries.    The presenters talked not just about the winners, but also how important it is to provide ongoing support to the losers, many of whom become motivated to go ahead and do energy improvements on their own dime.   For example, the sponsoring organization (Energy Trust) has created a site to evaluate the energy performance of their home.  I was reminded quite happily of the Blue Planet Foundation's Hawaii Energy Home Makeover.

The second presentation took it up a notch with the Cool School Challenge. This program is designed for high school students: they perform an energy audit of their school (involving electricity, solid waste, transportation, and heating/cooling), set goals for CO2 emission decreases, track progress, and report results.  The challenge website has curriculum materials, spreadsheets that do the carbon calculations, and reports on the various schools participating in the program.  It's won national and international awards (they showed a picture of high school students at the White House) and has been used in many states.  The program is designed to achieve four goals: (1) reduce carbon emissions; (2) encourage student leadership and empowerment; (3) foster a community of teachers and students; and (4) educate young people and their families.  What's more, they've provided a pathway for students to take the knowledge they gained in the classroom and put it into use in the community via a related program called EcoOffice, in which students do a similar energy audit of local businesses and help them determine changes.  Wow!

But wait: there's more.  The third presentation was on the Energy Smackdown.  The founder of this program began by asking the question: what if we could get people as excited about energy savings as they are about football and soccer?  Or American Idol?  The smackdown began as an energy competition between three households, and has more recently scaled up to a competition between three communities (I recall Medford, MA vs. Cambridge, MA; I forget the other one).  He showed some great previews of a reality television show based upon the competition to be released in early 2010.  In one scene, folks from Cambridge sneak into a hardware store and buy all the CFL bulbs, then leave chortling about how Medford is going to totally go down.  Kailua vs. Manoa?  Kaimuki vs. Aiea?  This could be so fun in Hawaii.

The day (and the conference) ended with a keynote by Hannah Choi Granade on a report by her organization (McKinsey and Associates) called "Unlocking Energy Efficiency in the US Economy".  The report indicates that we as a nation waste an unbelievable amount of energy, so much that even implementation of relatively conservative efficiency measures could reduce energy consumption by 9.1 quadrillion BTUs by 2020, roughly 23% of demand, abating approximately 1.1 gigatons of greenhouse gases. Indeed, we as a nation could actually decrease our aggregate energy use. 

That's the good news: the bad news is that there are "staggering" barriers to implementation. What's weird is that these barriers aren't typically "rationale": a straightforward economic analysis would lead most people and businesses to implement the measures.   The rather dense report goes into detail on the barriers and ways to approach them.  One thing I found interesting is that the authors could not apply the "80/20" rule to energy efficiency: in other words, there were not a small number of actions that would produce a disproportionate amount of benefit. Instead, we need to take a large number of small actions, and these actions cut across residential and industrial settings and all segments of society.

At the conclusion of the conference, one of the organizers took the microphone and asked us to take off our lanyards and leave them on the tables in front of us.  They collected them after we left and will reuse them at next year's BECC conference in Sacramento.  I'm going to do my best to be there.

Back to Day 2 or Day 1.

BECC 2009: Day 2

The second day of this year's Behavior, Energy, and Climate Change conference proved just as interesting as my experiences during the first day

I began the day by attending a talk by Barbara Farhar from the University of Colorado who presented some initial results from a qualitative study on Boulder residents and their views of the smart grid.  Of interest to me was the fact that she developed an energy literacy assessment instrument for the research, and that only (approximately) half of the respondents indicated that they would support demand response in their home (i.e. allow the utility to turn off their dryer or hot water heater).  While her sample size is small and not statistically generalizable, this is still an interesting datapoint.

I also saw a talk by Scott Pigg by the Energy Center of Wisconsin dealing with energy behaviors, and a very interesting talk by Hunt Allcott on ways to introduce randomization into energy/behavior research in order to obtain results with both internal and external validity.

The lunch speaker was Doug McKenzie-Mohr who gave a talk related to his book and website "Fostering Sustainable Behavior".   He views much of the current energy/behavior applications as "information intensive programs", in which the goal is to simply provide novel forms of information about energy and hope that behavioral change results.  He related a case study by Galler on an energy efficiency workshop that illustrates how this approach can fail in a big way.

In contrast, he advocates a more structured approach called "Community based social marketing" with a five step process that can be simplified as:  (1) Select behavior; (2) Uncover barriers and benefits; (3) Develop a strategy; (4) Pilot the strategy; and (5) Implement broadly and evaluate.   The key insight to me is (2), that without uncovering the barriers/benefits, it is hard to design a strategy (intervention) that is effective.

He illustrated the approach by describing a program in Canada to reduce auto idling time. An initial strategy was to erect signs requesting that drivers not idle their cars (for the environment or for the children).  It didn't work, since it did not address two barriers: (1) drivers think that keeping their car engine running is more efficient when the idle period is less than 4-5 minutes (the truth is 10 seconds) and (2) the increased number of car starts will burn out the starter (they won't).  So, the more effective strategy was to have a human talk to the car driver to disabuse them of these idling myths, along with a sticker they could put in their windshield to remind them (and others) not to idle.  A summary of the project provides more details.

After lunch I attended a talk by John Peterson from Oberlin College.  He spoke about his efforts to create "ecological feedback" systems for energy use in buildings, which he says involves the following key features: (1) near real-time; (2) socially comparative within monitored entities and among groups; (3) empathetically connects decision making to nature, community, and future generations; and (4) engaging, entertaining, etc.  He illustrated this with his work on the Campus Resource Monitoring System at Oberlin college, with the various dorm energy competitions they've done, and more recent work with the Great Lakes Protection Fund project.

The Campus Resource Monitoring System provides a nice user interface to energy information by the Lucid Design Group.   For a good article that overviews many of the current commercial energy dashboard offerings, I recommend the article "Visualizing Building Information" that appears in the Winter 2009 issue of Centerline, the newsletter of the Center for the Built Environment at UC Berkeley.

I then ran over to another session and caught the end of a talk by Jon Froehlich of the University of Washington, who described mechanisms developed by his research group to unobtrusively identify energy use at the appliance level by applying pattern/spectrum analysis techniques.   The basic idea is that turning appliances on and off produces "noise" in the power lines for a house which can be detected via some sensors and since different appliances have different signatures, you can disambiguate which appliance has been turned on/off and thus is responsible for the increment in energy consumption.  What's pretty cool is that they've gone beyond electrical use to apply this same general technique to water use and natural gas use.

After a few more talks which I won't summarize here, I attended a "Film Festival" session presented by Bill LeBlanc of E-Source.   He showed short clips (mostly ads, mostly humorous) regarding energy-related topics, but the best were a series of "PowerWalking" videos he has been doing for several years.   To get a taste, here's the PowerWalking 2007 video. We watched the world premier of the 2009 PowerWalking video, which he says will be on the E-Source website eventually.

The day (well, evening) concluded with a poster session.   There were lots of good ones, but one I found particularly relevant was by Sam Borgeson and Omar Khan on the Berkeley Campus Dashboard. (Here's a link to their 2008 poster session abstract.)  This is intended to be an open, web-based system for displaying energy data that has much in common philosophically with our WattDepot project.

On to Day 3 or back to Day 1.

Tuesday, November 17, 2009

BECC 2009: Day 1

This week I am attending the 2009 Behavior, Energy, and Climate Change conference in Washington, D.C.  This is the third year of the conference, and the first time I've attended it.   The conference has grown from around 250 people the first year to around 700 this year, and it shows:  some of the tracks yesterday were so popular that you had to stand outside the room and listen. 

Due to travel arrangements, I missed most of the first set of concurrent sessions, and so the first talk that I attended in its entirety was the lunch speaker: US Representative Brian Baird, who is the Chairman of the Subcommittee on Energy and Environment.  He is a former clinical psychologist and so is very open to the idea that behavior is an important issue when thinking about energy use and climate change. (This seems like a no brainer, though the political right is accusing him of mind control.) He had some interesting, low cost ideas on how to improve energy efficiency, such as to require that MLS listings of houses for sale should state the energy usage of the house (electricity and gas) over the past year.  This would incentivize home owners to improve the energy efficiency of their home (and use energy wisely), as it would make their house more attractive on the market, and enable home buyers to choose between two houses by picking the one with lower potential energy usage (all other things being equal.)

The afternoon had six concurrent sessions, of which two (Behavior Research and Policy Agenda; Technology Design) were of interest to me.  This appears to be a happy problem that I will struggle with for the remainder of the conference: there are always multiple places I want to be at once.  I chose the first session and listened to Ed Vine from the California Institute for Energy and the Environment give a review of nine papers sponsored by his organization for the California Public Utilities Commission.  They covered a lot of ground, from experimental design to process issues to market segmentation to behavioral issues.  You can get links to these and other related papers here.

The day concluded with a panel session on Smart Grid, with Matthew Trevithick from Venrock (a VC firm), Gregory Abowd from Georgia Tech, Omar Khan from Google's PowerMeter program,  and Carrie Armel from Stanford's Precourt Energy Efficiency Center.  (The Precourt Center website has an excellent section devoted to behavior with an 800+ citation database.)

Each panelist gave a very short presentation giving their perspective on smart grid and behavior, followed by lots of Q&A with the audience.  One of the many interested issues touched upon was an informal "stages of use" observed by the Google Powermeter team: users begin by learning about the system, then they engage with it to change their use, then they finally move into a maintenance phase.  The learning/engagement phases appear to last only a week or two, where users are actively manipulating the system and discovering issues with their usage and making concrete changes.  During this phase, they want to be able to drill down, do experiments, and so forth.  In the maintenance phase, they basically only want the system to tell them if something has changed in their energy usage.

On to Day Two or Day Three.

Monday, September 28, 2009

First experiences with the TED 5000

This past weekend I became the proud owner of a TED 5000 home energy monitor. I thought it might be useful to detail some initial experiences for the Renewable Energy and Island Sustainability Project and for our own subproject on From Smart Grids to Smart Consumers.

Overview of the TED 5000
The TED-5000 consists of four components.

MTU:  The "MTU" (for Measurement Transmitting Unit) goes inside your electric panel and monitors the power being consumed by your house.

Gateway: The "Gateway" is a small unit that plugs into a wall outlet and connects to your router (in my case, an Airport) using an ethernet cable. The Gateway receives current power consumption data from the MTU every few seconds via a proprietary powerline network protocol. 

Display Unit:  The "Display Unit" is a small handheld device (like an iPod) that communicates wirelessly with the Gateway using Zigbee and can show you (among other things) a near real-time display of your power usage.

Footprints:  Finally, "Footprints" is a web application that runs on the Gateway. You use Footprints to configure your TED, display various graphs and charts, and install Firmware updates.

Installing the TED 5000 involves: (1) opening the electric panel and connecting the MTU; (2) plugging in the Gateway and attaching an ethernet cable from it to my Airport; and (3) accessing the Footprints web application and setting some configuration values.

I found that connecting the MTU was pretty straightforward, basically because I had two electrically-akamai friends (Robert Brewer and Tony Querubin) helping and thus my personal involvement was limited to showing them where my electrical panel was located, holding a flashlight, and serving them Apple Huguenot torte when they finished.  The TED installation video provides a good approximation to what's involved, though there are some minor changes from the TED 1000 to the TED 5000.

Installation Glitch #1: No http://TED5000
The installation manual says that after hooking up the MTU and Gateway, you can access the Footprints web application using: http://TED5000.  That fails for non-Windows systems.  Fortunately, Robert was already aware of the problem and how to work around it. First, he looked up my Mac's IP address ( and invoked "ping" to find out all of the devices on the 10.0.1.* subnet. Three devices responded: my Airport (, my computer (, and a third "mystery" device at  Bingo: the URL to the footprints software on my computer is:

You might think that providing a URL that only works on Windows systems (and not even stating that in the installation guide) is pretty lame.  But wait, it gets worse:

Installation Glitch #2:  No HST.
The Display Unit provides the current time, and to do so, it needs to know the time zone.  Part of the configuration process in Footprints involves specifying the time zone, for which it provides a drop down menu with exactly four choices: EST, CST, MST, and PST.  No HST for those of us in Hawaii; no AKST for those of us in Alaska.   I am hoping the TED programmers get a clue in the near future and issue a firmware update that fixes this problem; in the interim, my Display Unit is laboring under the illusion that it lives in California.

Initial results
Having real-time display of power usage reminds me a great deal of when we first got our Prius: all of a sudden you're getting a whole new dimension of interesting data about your environment.  From a safety perspective, the TED has it all over the Prius, since it does not tempt you to monitor whether you're running exclusively on battery power while driving down the Pali Highway at rush hour.

Instead, from the sedentary position of my kitchen table, I can now inform you that our baseline power usage is around 400 Watts.  Each of our ceiling fans use about 50 W.  Turning on the microwave increases our power usage by about 1000 W.  I am too horrified to confess how much power our tiny, "Energy Star", 5000 BTU window air conditioner uses on startup, but it dwarfs every other appliance we own by a mile.

I am sure that we are similar to every other new TED user, in that we tend to run over to the Display Unit whenever we do anything (turn on/off the TV, turn on/off the stove, etc.) to see what effect it has on our power consumption.  Similarly, if we're in the vicinity of the Display Unit and our power usage is significantly different from 400W, we start wondering what's going on.

It's also cool to empirically test various things you've heard in the news.  For example, Joanne read some place that you should unplug your toaster oven when you're not using it because it uses up power.  That seemed suspect to me, so I took the Display Unit over to our toaster oven, unplugged it, and waited for a drop in power consumption.  Nada.

Behavioral change, or Is This Just a $250 toy?
The real question is whether the TED 5000 is anything more than an expensive toy, and the answer to that depends upon whether we actually change (i.e. reduce) our energy consumption now that we know what we're doing.

The jury is still out on that question, though of course the jury has only had about 36 hours to deliberate.  What I can say at this point is the following:

(1) Our espresso maker is an energy hog.  I was surprised to see that it uses over 1000W, and this is significant because we tend to turn it on first thing in the morning and leave it on for a couple of hours until we have finished making all our morning espressos.  Given its energy consumption, we should turn it off between uses even though that means we need to wait for it to heat up again.

(2) Our TV/DVR's standby power consumption is fairly low. We have heard horror stories about LCDs TVs that  consume many watts even when "off", and we were happy to discover that our TV uses only 1-2W when in standby.  The DVR uses about 20-22W in standby.  That's not too bad, although that 25W does account for about 6% of our "baseline" 400 W.

(3) We need to replace our remaining incandescent lightbulbs with CFLs without delay.  When your baseline is 400W, and you turn on the kitchen lights and it jumps by 25%, you just feel stupid.

I intend to do a bit more detective work on our baseline usage, and perhaps we can cut that down.  Come to think of it, we should not make any changes to our behavior for a week so that we can establish a decent baseline dataset of our current-but-soon-to-be-history energetically profligate lifestyle.

For another new TED 5000 user's experience, see here.

Monday, September 7, 2009

Pretty printing code with google-code-prettify

There are many ways to pretty print code in blogs, as a Google search will show you, and all the methods involve a certain amount of hassle. After some fooling around today, I have come across the following approach which is pretty simple and has the advantage that you don't have to preprocess your code to escape certain characters (such as angle brackets).

My currently preferred approach uses google-code-prettify, which I suspect is the package used to perform syntax highlighting in google project hosting. Integrating this package in Blogger requires three steps:

1. Update the <head> section of your Blogger template

Insert the following lines of code:
<link href=""
rel="stylesheet" type="text/css"/>
<script src=""
type="text/javascript" />
2. Update the <body> element tag of your Blogger template

Change it to:
<body onload="prettyPrint()">
Save these changes to your template.

3. Insert your code using <code class="prettyprint lang-java">

You must be in "Edit Html" mode, not "Compose" mode, when you do this. Here is an example of the results:

* Creates a new CD, provided its title, group, and song list.
* @param title The title.
* @param group The group.
* @param tracks The song titles.
public CompactDisc(String title, String group, String... tracks) {
this.title = title; = group;
this.tracks = Arrays.asList(tracks);
// Since CompactDiscs are immutable, the hash value will never change for an instance.
this.hashvalue = (new String(title + group)).hashCode();

OK, if you have followed me so far, and if your Blog template has a light background, then the code should be formatted nicely. However, if your Blogger template has a black background like mine, then the results are terrible, because the default colors in prettify.css are designed to look good against a white background, not a black background.

The reason why my code sample looks relatively reasonable is because I use a custom version of prettify.css where I have changed the colors to those that contrast well with a black background. This file is available to you at:

So, use this URL in the <link> element that you inserted into the <head> element of your Blogger template, and you should be fine.

If you don't like my choice of colors (or google's), you can create your own prettify.css, place it somewhere on the web, and reference it to get your own coloring scheme.

There is a bug in Blogger that you should be aware of:  if you switch back and forth between "Compose" and "Edit HTML" mode while doing this, at some point Blogger eliminates all indentation in your code.   You can avoid this by not using "Compose", or else by inserting your code as the last step in creating your posting.

Monday, July 13, 2009

Google Summer of Code Midterm Results for Hackystat

I am delighted to announce that all five of the GSoC students working on Hackystat have passed their midterm evaluation and will be continuing with us for the rest of the summer.

Myriam Leggieri is working on the use of RDF to facilitate integration of Hackystat data with other kinds of knowledge sources in the project hackystat-linked-sensor-data.

Rachel Shadoan is working on the integration of social networking with Hackystat data in order to create novel data mining opportunities in the project hackystat-analysis-socnet.

Harvey Weitz is investigating a case-based reasoning approach to management for service-oriented architecture, using Hackystat as his test case, in the project hackystat-service-manager.

Anthony Du is implementing a software engineering educational game in the project hackystat-ui-devcathlon, building upon the initial work done in ICS 414 last spring.

Last but not least, Shaoxuan Zhang is implementing support for Issue sensor data collection and analysis through the hackystat-sensor-ant project.

All of these students have made significant progress on their projects so far, and the summer is only half over. I look forward to see what they have working by September!

Saturday, March 14, 2009

Hackystat in a Nutshell

Another screencast, this time a short introduction to Hackystat:

Monday, March 2, 2009

Google Summer of Code 2009

Applications for Google Summer of Code 2009 are coming up soon. I recently gave a short talk on Google Summer of Code to some graduate students at the University of Hawaii. Here is a screencast of that lecture: