1. 26
  1.  

  2. 15

    Though the article doesn’t talk about the specifics, I don’t actually think the solution here is to push “sustainable computing”. Energy is complicated and availability will only become more complicated as renewables make up a larger part of the mix. Rather than try to create complex philosophies on what sustainable computing is, governments need to tax energy usage based on emissions. Once that happens, hardware, and consequently software, will adjust.

    1. 6

      No guarantee that there will be sufficient private sector investment to develop the new technology. It could be done faster with public planning and investment, as happened with the core computing technologies of today, and we don’t have time to spare.

      1. 6

        Indeed it may, but I still believe the first step is to properly price externalities. If you don’t, you’ll have folks optimizing things they don’t need to or leaving things unoptimized that they felt were “free”. After that public and private initiatives can take over adjusting to costs.

        1. 1

          I don’t know how you make that judgement without an idea of what changes need to happen and the overall scale of the effort. If we’re looking for marginal improvements carbon pricing might be fine, but if we are set on reducing energy usage by orders of magnitude then everything is wasted effort until we start coordinating on a global scale to completely reorganize our infrastructure.

          1. 3

            I don’t know how you make that judgement without an idea of what changes need to happen and the overall scale of the effort.

            Hm? Are you implying that if you price in externalities that folks will continue to pollute? How?

            If we’re looking for marginal improvements carbon pricing might be fine, but if we are set on reducing energy usage by orders of magnitude then everything is wasted effort until we start coordinating on a global scale to completely reorganize our infrastructure.

            Are you talking about creating worldwide central planning on production? I’m not sure that’s at all feasible. No developed or developing nation right now uses central planning, let alone coordinating across nations for planning.

            1. 1

              Hm? Are you implying that if you price in externalities that folks will continue to pollute? How?

              No, I’m saying I don’t know how you can be confident that carbon pricing should be the first step, at the cost of de-prioritizing large scale planning.

              If we’re looking for marginal improvements carbon pricing might be fine, but if we are set on reducing energy usage by orders of magnitude then everything is wasted effort until we start coordinating on a global scale to completely reorganize our infrastructure.

              Are you talking about creating worldwide central planning on production? I’m not sure that’s at all feasible. No developed or developing nation right now uses central planning, let alone coordinating across nations for planning.

              Do you consider the development of the Internet by DARPA to be “worldwide central planning”? Whatever you call it, it formed basis for all Internet use, including the market in which tech firms compete. Microsoft’s development of Windows also had global effects, defining the mode of computer use for the vast majority of users for decades. Now we have international organizations that define standards for cell networks, and two companies in charge of the software that runs on all consumer devices on those networks. This is all highly centralized, or at least globally coordinated, as compared with small firms competing in a market.

              I also wouldn’t want to give the impression that we should be limited by precedents. The world has never faced anything like the climate crisis and the response will necessarily be unprecedented.

              1. 2

                No, I’m saying I don’t know how you can be confident that carbon pricing should be the first step, at the cost of de-prioritizing large scale planning.

                How else do you know how to optimize energy usage? You’re going to have more or less energy depending on the natural resources available (wind, solar, etc). If you don’t quantify costs then what’s the “goal”, even of a public initiative? Just as efficient as possible? That seems like a mission statement ripe to have no end in sight.

                Do you consider the development of the Internet by DARPA to be “worldwide central planning”? Whatever you call it, it formed basis for all Internet use, including the market in which tech firms compete.

                ARPANet was only the germ of the internet we know today, and the internet created something new where nothing existed, which American universities evangelized outward. The work to operationalize and roll-out packet-switched networks to the public happened through a combined effort of the US DoD, AT&T, and several European organizations during the Protocol Wars 1 period. This time around, we aren’t creating something new, we’re taking existing processes with existing dependencies and trying to make them more efficient. Note that original process took almost 20 years and was created through lots of US and EU collaboration, only because of America’s heavy involvement with Western Europe at the time. A similar solution now would need to include collaboration with Asia, the Middle East, and Eastern Europe which is not nearly as politically aligned as the EU and the US.

                I also wouldn’t want to give the impression that we should be limited by precedents. The world has never faced anything like the climate crisis and the response will necessarily be unprecedented.

                Indeed but unless you can see a path by which our current geopolitical situation can give birth to such unprecedented solutions, then it’s probably not feasible. Humans generally don’t wish for large amounts of wholesale change and different nations have very different ideas about the urgency of climate change.

                1. 3

                  How else do you know how to optimize energy usage? You’re going to have more or less energy depending on the natural resources available (wind, solar, etc). If you don’t quantify costs then what’s the “goal”, even of a public initiative? Just as efficient as possible? That seems like a mission statement ripe to have no end in sight.

                  This begs the question that taxing carbon (by how much?) will tell you how to optimize energy usage. No, firms acting in isolation don’t magically come up with the optimal solution to anything but how to maximize short-term profits (and then only sometimes).

                  We can define the goals of a public initiative by looking at aggregate data as the author of this article is doing.

                  ARPANet was only the germ of the internet we know today, and the internet created something new where nothing existed, which American universities evangelized outward. The work to operationalize and roll-out packet-switched networks to the public happened through a combined effort of the US DoD, AT&T, and several European organizations during the Protocol Wars 1 period. This time around, we aren’t creating something new, we’re taking existing processes with existing dependencies and trying to make them more efficient.

                  The amount of existing dependencies that we accommodate is up for debate. The Internet eventually outgrew the telephone network, and any new infrastructure could outgrow the current one in more or less time depending on the resources and political will.

                  Note that original process took almost 20 years and was created through lots of US and EU collaboration, only because of America’s heavy involvement with Western Europe at the time. A similar solution now would need to include collaboration with Asia, the Middle East, and Eastern Europe which is not nearly as politically aligned as the EU and the US.

                  I never said it would be easy!

                  Indeed but unless you can see a path by which our current geopolitical situation can give birth to such unprecedented solutions, then it’s probably not feasible. Humans generally don’t wish for large amounts of wholesale change and different nations have very different ideas about the urgency of climate change.

                  I could say the same thing about the impossibility of passing a carbon tax, which activists have been trying to do for decades. We are getting wholesale change whether or not we wish for it, and as climate change progresses ideas about its urgency will change. It’s not a question of whether people will respond in a significant coordinated way, but when.

                  1. 1

                    This begs the question that taxing carbon (by how much?) will tell you how to optimize energy usage.

                    This is an open question, but there is already prior art here. Lots of energy providers throughout the world already charge tiered rates for energy usage. We can do the same thing for emissions. We can levy tiered taxes based on government emissions goals.

                    No, firms acting in isolation don’t magically come up with the optimal solution to anything but how to maximize short-term profits (and then only sometimes).

                    I never said anything about firms. I think you’re reading a political philosophy here that I didn’t ascribe (why?). Unless you wish to see private use of energy abolished (in which case, this is just going further and further from the current state of affairs), emissions will have to be accounted for one way or another, so firms, the government, and private-public partnerships can all be relevant. Moreover, this doesn’t really answer the question I posed. I asked why carbon taxes wouldn’t incentivize behavior. Your answer is that firms maximize short-term profits. What bad behaviors firms enghage in is immaterial to the question. Firms are just institutions, like the government, just not under some form of voter (direct or indirect) control. There are reasons that firms maximize short-term profits and it’s worth discussing these reasons, so that even if the solution is implemented only through the government, that the government doesn’t fall into these same traps as well. But again bad firm behavior is orthogonal to the question of energy accounting.

                    I could say the same thing about the impossibility of passing a carbon tax, which activists have been trying to do for decades. We are getting wholesale change whether or not we wish for it, and as climate change progresses ideas about its urgency will change. It’s not a question of whether people will respond in a significant coordinated way, but when.

                    I guess as someone active in environmental politics and organizing in my own area, I just find this to be unactionable, which is why I don’t like to think this way. Moreover once we divorce ourselves from the current state of affairs, it’s trivial to construct a magical society in which all of our problems are solved, and I can think of many of them. The solution space is constrained by the path taken from “here” to “there”.

                    1. 2

                      Great discussion, thank you both. I’m left with the question of why it has to be either-or and why we can’t just agree to pursue both. Developing new infrastructure will take time. Taxing externalities seems like a reasonable thing to kick off in the mean time. If we have two levers that only gives us greater ability to steer. Right?

                      1. 1

                        right

                      2. 1

                        I never said anything about firms. I think you’re reading a political philosophy here that I didn’t ascribe (why?). Unless you wish to see private use of energy abolished (in which case, this is just going further and further from the current state of affairs), emissions will have to be accounted for one way or another, so firms, the government, and private-public partnerships can all be relevant. Moreover, this doesn’t really answer the question I posed. I asked why carbon taxes wouldn’t incentivize behavior. Your answer is that firms maximize short-term profits. What bad behaviors firms enghage in is immaterial to the question. Firms are just institutions, like the government, just not under some form of voter (direct or indirect) control. There are reasons that firms maximize short-term profits and it’s worth discussing these reasons, so that even if the solution is implemented only through the government, that the government doesn’t fall into these same traps as well. But again bad firm behavior is orthogonal to the question of energy accounting.

                        You implied that carbon pricing tells you how to optimize energy usage. I inferred that you were envisioning the market finding an optimal solution in response to carbon pricing, which is why I brought up firms. If you didn’t mean that, I don’t know what you meant by “how else do you know how to optimize energy usage?” No one said carbon pricing wouldn’t incentivize behavior.

                        I guess as someone active in environmental politics and organizing in my own area, I just find this to be unactionable, which is why I don’t like to think this way. Moreover once we divorce ourselves from the current state of affairs, it’s trivial to construct a magical society in which all of our problems are solved, and I can think of many of them. The solution space is constrained by the path taken from “here” to “there”.

                        Carbon pricing is appealing because it applies to many sectors of the economy, but it’s not the only thing you can reasonably push for. It may seem unactionable when speaking in generalities, but for each issue you can come up with a specific plan. For transportation the solutions are relatively obvious: more trains. The solutions are less obvious in computing because the existing infrastructure is so complex and there are so few constraints on what we could do differently, but that doesn’t mean it’s not worth pursuing.

                        1. 1

                          I actually think we mostly see eye-to-eye here (though feel free to disagree). I can post a more detailed reply if you’d like, but I pretty much agree.

                          1. 2

                            cool, have a nice day

        2. 2

          tax energy usage based on emissions

          please, please mean “tax [corporate] energy usage based on emissions”..

          1. 3

            This is tough. I certainly think corporate energy use should be taxed more, but what about situations where folks are using residential or personal energy use for running an illegal business? Or simply just as a disincentive against profligate use as an individual, especially a wealthy individual.

        3. 7

          The current emissions from computing are about 2% of the world total

          This article is pretty ridiculous. How about we focus on the things that generate vastly more than 2% of emissions, like the automotive industry and coal-burning power plants? This article is like the campaigns telling California city dwellers to take shorter showers during a drought, when it’s huge factory farms using 90% of the state’s water (at 1/10 the price consumers pay.)

          And let us not forget that computers don’t emit carbon. They consume electricity. So by moving to carbon-neutral power generation, which is of course hugely important, we also fix the emissions by computers.

          We are not fundamentally energy-limited. The amount of solar energy available to harvest is vast, and solar cell efficiency is growing a lot faster than anyone expected. There are very credible, mainstream analyst reports predicting that coal and gas power generation will be priced out of existence soon if enough renewable capacity is built.

          1. 5

            I also mostly agree with this, but there’s one point to add:

            And let us not forget that computers don’t emit carbon. They consume electricity. So by moving to carbon-neutral power generation, which is of course hugely important, we also fix the emissions by computers.

            The article argues that the production, distribution, and disposal of computers is where a large portion of the carbon emission lies. While they don’t emit carbon directly during operation and I agree with your point about moving to carbon neutral power generation, it’s also important to address the carbon that comes from our appetite for more computing devices. Building a culture of reuse and repair is going to take some time though, it’s a radical departure from the kind of consumption culture that exists in many countries today.

            1. 2

              Mostly second this (“Things like reducing the size of font files on your web properties is, frankly, like polishing the silverware during a house fire.” is how I put it here: https://climate.davis-hansson.com/p/big-picture-2020/). But some qualifiers:

              1. Computers can aid transitioning to a clean grid by reducing the need for battery storage and transmission, by moving heavy workloads off-peak; there is a significant impact if you can shift just 1% of load a few hours

              2. While the globally optimal thing to do is to move to a clean grid and electrifying transport and industry, you may as an individual be situated such that your time would be best spent optimizing some resource intensive software your employer runs, maybe.

            2. 6

              I don’t need incentives like “carbon neutrality” or “global warming” to know that we must reduce our resource consumption, both energy and materials.

              We need a culture of repair rather than buying something new, and people make it too easy for themselves in blaming the big corporations. Indeed, there should be legislation for producers and vendors to publish schematics and sell spare parts to allow repair, but it must also be within our spirit to nurture a culture of teaching repairing and keeping it in mind when faced with a defect.

              Regarding computing, a big step would be to de-bloat the web (which would also make it faster). Lots of energy is wasted with that, and it would have a great overall benefit. I don’t think you can push it with legislation, but maybe as a new counterculture?

              Streaming is another thing that consumes a lot of energy (because it’s not multicast and each stream-connection must be served individually). A very ambitious solution here are peer-to-peer-networks where the distance between nodes gets shorter, improving efficiency (and latency). If you consider personal devices and the fact that DRAM consumes energy if filled or not, I can imagine any handheld or desktop to become a data-node without much added energy consumption in each device.

              1. 7

                Regarding computing, a big step would be to de-bloat the web (which would also make it faster). Lots of energy is wasted with that, and it would have a great overall benefit. I don’t think you can push it with legislation, but maybe as a new counterculture?

                This will do almost nothing. The entire internet is just 0.6% of the world’s emissions, and the vast vast majority of the internet is streaming content. Netflix by itself made up 15% of the total internet bandwidth in 2018. Website bloat doesn’t matter.

                (You mention streaming after this, but it’s not the individual connections that matter, it’s the transfer of data that’s energy intensive.)

                1. 2

                  Phones and tablets (which make up the majority of computing devices) really aren’t very repairable, beyond trivial things like cracked screens. Have you ever seen the insides of one?

                  DRAM consumes energy, but that’s an absolutely tiny quantity. Much more is used by the CPU, and it’s the display backlight that’s the great majority. Phones and tablets are very energy efficient compared to desktop computers with their GPUs glowing cherry-red in their helium-cooled enclosures.

                  1. 4

                    Current phones are already able to be repaired, the current political debate is in response to manufacturers artificially constraining supply of spare parts by tweaking them just barely enough to break compatibility, then demanding exclusivity of the custom part from the manufacturers. That way, they can demand people pay for a $300 new phone instead of a $10 new chip plus $50 in labour costs.

                    That said, making phones easier to repair and especially easier to replace the battery of, is something that isn’t done because it’s simply not a priority, not because it’s necessarily hard.

                2. 2

                  The discussion about server lifecycles raises an interesting point: how to introduce business incentives to reuse hardware rather than replacing it after some fixed window. The cost reductions from new hardware make a compelling argument, even if it’s short sighted.

                  Also, how would this interact with regulations that require the destruction of a computing device after it is used? Like the article points out, some amount of this is going to require regulatory action.

                  1. 1

                    As developers it’s so wasteful how much we recompile code at ever phase in many setups: every local dev rebuilds, testing machines rebuild, pushing to staging and production cause rebuilds. I hope things like Cachix for Nix’s reproducible builds becomes the norm.