Technology and the Environment DD's Eco Notes

When I recently blogged (here and here) the question of how much energy it takes to build a car, the obvious question is how about a server?

We're starting to do a deeper dive on the complete environmental life of our servers. The big picture is pretty straightforward, and isn't really specific to servers. In short, we're thinking of six distinct phases of a a product's life where energy and natural resources are used:

1. Design the product
2. Manufacturing of the product
3. Package and deliver the product
4. Operate and maintain the product
5. Collect the product at end-of-life
6. Recycle the product

I'll be coming back to this list over and over in the coming months, but at this point I'll just make some high level observations. First, the energy and natural resources to design the product are probably not significant for a high volume product, but for low volume or one-off products, it could be very high. By the time you spend server time on design, build prototypes, do lab tests, etc, this may be higher than the cost of the product itself in many cases. I bet this is true for the space shuttle, for example.

Three and five are interesting, as they include issues such as product packaging, shipping, and a number of supply chain activities. Number 3 may also include documentation and other things that are delivered with the product, but which aren't the product itself. More on them in future posts.

Finally, there's recycling, which is one of the most important parts of this cycle. If you're involved in products and you haven't read Cradle to Cradle by McDonough and Braungart, make sure to pick it up. Recycling starts with design - if you didn't design a product to be recycled, its very unlikely it can be effectively. And its not just whether you can recycle, but how good the materials are that you end up with, and how much energy it takes to do the recycling.

I suspect that this will end up being one of the big differences between today's product engineers and those fifty years from now. Between laws and social responsibility expectations, future engineers will be required to do a complete accounting for their products at design time. They'll be far more attuned to the various material and chemical options for their products (and more, better options will be available, see here), and they'll be able to account for the impact of their choices.

I'll post more as we go through this process ourselves.

One random thought on Earth Day (US version): I think we need a second one that falls on work day. I believe that over the next few years we're going to find that, for many people, the biggest environmental impact they can have is related what happens at work, not at home. It's just a hunch, but in this consumer and service driven economy, I bet I'll be proven right.

One of the things I'm increasingly interested in is the role of gaming technology in the office. As Sun, with its OpenWork program and increasingly global workforce, gets more and more distributed, we need integrated core services such as secure identity and presence, audio conferencing, multi-party chat, etc etc etc. Today's massive, multi-player online games are as close as we have to examples of systems that integrate all of these things in a coherent way. Unfortunately they're almost always closed systems, and lack some of the basic features that an enterprise would require (yes, I need to be able to have an internal meeting without worrying about pig grenades). We're working on two separate threads of this right now. The first is Project Darkstar, our Java-based infrastructure for multi-player on-line games. We've been getting tons of attention since our announcements and demos at GDC, and there's more to come at JavaOne. The second is a SunLabs project called MPK20. The name is a play on the building naming scheme in Menlo Park, where our campus currently goes up to MPK18. MPK20 is the next building, but it will only exist in virtual space, so that we can finally get some reasonably sized auditoriums, window offices for everyone, etc. Lots of cool work going on with this - check out the demos at JavaOne for MPK20 as well. We're not sure where this is going yet, but the more time we spend on it, the more we're convinced that there's something there. If you're interested in this space from a research or product perspective, give us a shout - there's tons to work on. Also, we've got some developer positions open - check out this, this and this. Game on!

Uh oh...the competition is heating up. Fake Steve is getting things rev'ed up at (fake) Apple:

"After seeing Arnold on the cover of Newsweek touting his big commitment to the global warming cause, I said you know what? That's it. We're going green. We're gonna be greener than everyone else in the world put together. No more losing out to Lenovo and Dell on this one. No more fighting Greenpeace. To hell with it. We're gonna pay them off. Name your price, Greentards.

Yesterday I called together my best people and told them to go out and figure out some great ideas and bring them back to me, and no matter what it costs, let's get going on this initiative. By the end of this year I want Apple to be known as the greenest company in the world -- not just in tech but in everything. If we've got to make hydrogen-powered computers and iPods that run on solar energy, so be it. Let's get this done. And readers, if you have ideas, send them along. We want your input."

(Via The Secret Diary of Steve Jobs.)

On Friday I asked about relative environmental impact of buying a new, more efficient car versus driving the old one. I got some good comments and emails, so thought I'd respond to a few here.

• A couple of people commented that they didn't believe the idea that a new car wouldn't be built if someone opted not to buy one. In the case of single person, that may be true (although I'd disagree with that as well), but in the case of a more widespread change in behavior, its patently not true. If many people extended the life of their car by a year (i.e. waited a year longer than normal to buy a new one), it would have a huge impact on the auto industry and result in less cars being built.
• I wholeheartedly agree with the "drive less" comment. That's more important than any purchase decision. How much you're driving also factors into the answer to my question. If you only drive a few thousand miles a year, its very unlikely that a more efficient car is a net win. If you drive 50K miles per year, its much more likely that a more efficient car is a net win.
• I'm intrigued by the comment about the usable lifetime of the car. That hadn't occurred to me at all, and it definitely should be a factor of some kind.
• One commenter questioned whether the energy to build a car was significant. Josh Simons sent me this website which includes this study, which says that it is. Now this may have been written by the Hummer Owners of America or it may be totally accurate, I don't know. But its worth looking at how they calculate that an H3 has a lower total energy impact than a Prius (yes, you read that right - read the study).

A couple of reasons that this kind of intellectual experimentation is useful for me. First, its a good reminder that none of these seemingly simple questions is really as simple as they look. These energy issues are deep, and its good practice to understand where the subtleties are and get in the habit of looking beyond the initial answer. I'm getting better at thinking these through, but I'm still surprised by some of the points every time, like the comment about lifetime above. Seccond, we're embarking on a similar study on the complete energy cycle of computers, so its interesting for me to look at another complex product. More to come on this over the next few quarters.

Here's a question I've been wondering about:

If I upgrade my car to a brand new one that's more efficient, I save energy and emissions based on the miles I drive with the new vehicle. However, a lot of energy went into making and delivering the new vehicle, and, in theory, that energy wouldn't have been spent if I didn't buy a new car.

So how big of a MPG increase do I need to get before it's at least environmentally break even?