<html>
  <head>

    <meta http-equiv="content-type" content="text/html; charset=UTF-8">
  </head>
  <body>
    <div class="css-ov1ktg">
      <div class=" css-1fkm2o5">
        <div class="rail-wrapper css-so4veu">
          <div class=" css-ac4z6z">(Zelfs de Guardian omarmt de
            luddieten)<br>
          </div>
        </div>
      </div>
    </div>
    <div id="root">
      <div class="css-vexkl">
        <div class="css-g13mkd">
          <div class="css-ov1ktg">
            <div width="718" class="css-s84953">
              <header class="css-d92687">
                <h1 class="css-twhgrd">To decarbonize we must
                  decomputerize: why we need a Luddite revolution</h1>
                <p>Wed 18 Sep 2019 06.30 BST</p>
                <div class="css-1v1wi0p">
                  <div class="css-7kp13n">By</div>
                  <div class="css-7ol5x1"><span class="css-fgeroe">Ben
                      Tarnoff</span></div>
                  <div class="css-8rl9b7">theguardian.com</div>
                  <div class="css-zskk6u">7 min</div>
                </div>
                <div class="css-1890bmp"><a
href="https://getpocket.com/redirect?url=https%3A%2F%2Fwww.theguardian.com%2Ftechnology%2F2019%2Fsep%2F17%2Ftech-climate-change-luddites-data"
                    target="_blank" class="css-19cw8zk">View Original</a></div>
              </header>
              <div class="css-429vn2">
                <div role="main" class="css-1ciuztk">
                  <div id="RIL_container">
                    <div id="RIL_body">
                      <div id="RIL_less">
                        <div lang="en">
                          <div class="RIL_IMG" id="RIL_IMG_1">
                            <figure> <img
src="https://pocket-image-cache.com//filters:no_upscale()/https%3A%2F%2Fi.guim.co.uk%2Fimg%2Fmedia%2Fb7c2b87218ff0a7f5f11016e2d868d1c150083fc%2F0_38_3900_2340%2Fmaster%2F3900.jpg%3Fwidth%3D300%26quality%3D85%26auto%3Dformat%26fit%3Dmax%26s%3Dcef83874a454b67815ff71df006e1a2e"
                                alt="The data server hall at Facebook’s
                                storage center near in Lulea, Sweden.
                                Data centers currently consume 200
                                terawatt hours per year – roughly the
                                same amount as South Africa.
                                Photograph: Simon Dawson/Bloomberg via
                                Getty Images"> <figcaption>The data
                                server hall at Facebook’s storage center
                                near in Lulea, Sweden. Data centers
                                currently consume 200 terawatt hours per
                                year – roughly the same amount as South
                                Africa.
                                Photograph: Simon Dawson/Bloomberg via
                                Getty Images</figcaption> </figure>
                          </div>
                          <div>
                            <p><span><span>O</span></span>ur built
                              environment is becoming one big computer.
                              “Smartness” is coming to saturate our
                              stores, workplaces, homes, cities. As we
                              go about our daily lives, data is made,
                              stored, analyzed and used to make
                              algorithmic inferences about us that in
                              turn structure our experience of the
                              world. Computation encircles us as a
                              layer, dense and interconnected. If our
                              parents and our grandparents lived <em>with</em>
                              computers, we live <em>inside</em> them.</p>
                            <p>A growing chorus of activists,
                              journalists and scholars are calling
                              attention to the dangers of digital
                              enclosure. Employers are <a
href="https://www.wsj.com/articles/the-new-ways-your-boss-is-spying-on-you-11563528604">using</a>
                              algorithmic tools to surveil and control
                              workers. Cops are <a
                                href="https://stoplapdspying.org/resources/architecture/">using</a>
                              algorithmic tools to surveil and control
                              communities of color. And there is no
                              shortage of dystopian possibilities on the
                              horizon: landlords evicting tenants with
                              “smart locks”, health insurers charging
                              higher premiums because your Fitbit says
                              you don’t exercise enough.</p>
                            <p>Digitization doesn’t just pose a risk to
                              people, however. It also poses a risk to
                              the planet. July was the hottest month on
                              record. Large chunks of the Arctic are
                              melting. In India, more than half a
                              billion people face water shortages.
                              Putting computation everywhere directly
                              contributes to this crisis. Digitization
                              is a climate disaster: if corporations and
                              governments succeed in making vastly more
                              of our world into data, there will be less
                              of a world left for us to live in.</p>
                            <p>To understand the relationship between
                              data and climate, the best place to start
                              is machine learning (ML). Billions of
                              dollars are being spent on researching,
                              developing, and deploying ML because major
                              breakthroughs in the past decade have made
                              it a powerful tool for pattern
                              recognition, whether analyzing faces or
                              predicting consumer preferences. ML
                              “learns” by training on large quantities
                              of data. Computers are stupid: babies know
                              what a face is within the first few months
                              of being alive. For a computer to know
                              what a face is, it must learn by looking
                              at millions of pictures of faces.</p>
                            <p>This is a demanding process. It takes
                              place inside the data centers we call the
                              cloud, and much of the electricity that
                              powers the cloud is generated by burning
                              fossil fuels. As a result, ML has a large
                              carbon footprint. In a recent <a
                                href="https://arxiv.org/pdf/1906.02243.pdf">paper</a>
                              that made waves in the ML community, a
                              team at the University of Massachusetts,
                              Amherst, found that training a model for
                              natural-language processing – the field
                              that helps “virtual assistants” like Alexa
                              understand what you’re saying – can emit
                              as much as 626,155lb of carbon dioxide.
                              That’s about the same amount <a
href="https://www.theguardian.com/environment/ng-interactive/2019/jul/19/carbon-calculator-how-taking-one-flight-emits-as-much-as-many-people-do-in-a-year">produced</a>
                              by flying roundtrip between New York and
                              Beijing 125 times.</p>
                            <p>Training models isn’t the only way ML
                              contributes to the cooking of our planet.
                              It has also stimulated a hunger for data
                              that is probably the single biggest driver
                              of the digitization of everything.
                              Corporations and governments now have an
                              incentive to acquire as much data as
                              possible, because that data, with the help
                              of ML, might yield valuable patterns. It
                              might tell them who to fire, who to
                              arrest, when to perform maintenance on a
                              machine or how to promote a new product.</p>
                            <div class="RIL_IMG" id="RIL_IMG_2">
                              <figure> <img
src="https://pocket-image-cache.com//filters:no_upscale()/https%3A%2F%2Fi.guim.co.uk%2Fimg%2Fmedia%2Fc438344138315871faa44fe46288a3e7f4872077%2F0_126_5472_3283%2Fmaster%2F5472.jpg%3Fwidth%3D300%26quality%3D85%26auto%3Dformat%26fit%3Dmax%26s%3D7548ef595ae677826b28578ead0ee484"
                                  alt="‘Digitization doesn’t just pose a
                                  risk to people. It also poses a risk
                                  to the planet. In India, more than
                                  half a billion people face water
                                  shortages.’ Photograph: R
                                  Parthibhan/AP"> <figcaption>‘Digitization
                                  doesn’t just pose a risk to people. It
                                  also poses a risk to the planet. In
                                  India, more than half a billion people
                                  face water shortages.’ Photograph: R
                                  Parthibhan/AP</figcaption> </figure>
                            </div>
                            <p>One of the best ways to make more data is
                              to put small connected computers
                              everywhere: Cisco predicts there will be
                              28.5bn networked devices by 2022. Aside
                              from the energy required to manufacture
                              and maintain those devices, the data they
                              produce will live in the carbon-intensive
                              cloud. Data centers currently consume 200
                              terawatt hours per year – roughly the same
                              amount as South Africa. Anders Andrae, a
                              widely cited researcher at Huawei, tells
                              me that number is likely to grow 4-5 times
                              by 2030. This would put the cloud on par
                              with Japan, the fourth-biggest energy
                              consumer on the planet.</p>
                            <p>What can be done to curb the carbon costs
                              of data? Greenpeace has long pushed cloud
                              providers to switch to renewable energy
                              sources and improve efficiency. These
                              efforts have seen some success: the use of
                              renewables by data centers has <a
                                href="http://www.clickclean.org/downloads/ClickClean2016%20HiRes.pdf">grown</a>
                              substantially. Meanwhile, efficiency gains
                              from better techniques and bigger
                              economies of scale have <a
                                href="https://iopscience.iop.org/article/10.1088/1748-9326/aaec9c">moderated</a>
                              the cloud’s power consumption in recent
                              years. When it comes to ML, a group of
                              researchers are <a
                                href="https://arxiv.org/pdf/1907.10597.pdf">calling</a>
                              for a more energy-conscious approach,
                              which they call “Green AI”. These are
                              encouraging trends, and tech workers
                              themselves are likely to play a key role
                              in advancing them: Amazon employees have
                              been <a
href="https://www.theguardian.com/technology/2019/may/22/amazon-workers-climate-crisis-board-jeff-bezos">organizing</a>
                              for a climate plan since late last year,
                              and they recently <a
href="https://medium.com/@amazonemployeesclimatejustice/amazon-employees-are-joining-the-global-climate-walkout-9-20-9bfa4cbb1ce3">announced</a>
                              a global walkout for 20 September. Among
                              their demands is for the company to commit
                              to zero emissions by 2030 and to stop
                              selling cloud services to fossil fuel
                              companies.</p>
                            <p>But it’s clear that confronting the
                              climate crisis will require something more
                              radical than just making data greener.
                              That’s why we should put another tactic on
                              the table: making less data. We should
                              reject the assumption that our built
                              environment <em>must</em> become one big
                              computer. We should erect barriers against
                              the spread of “smartness” into all of the
                              spaces of our lives.</p>
                            <p>To decarbonize, we need to decomputerize.</p>
                            <p>This proposal will no doubt be met with
                              charges of Luddism. Good: Luddism is a
                              label to embrace. The Luddites were heroic
                              figures and acute technological thinkers.
                              They smashed textile machinery in
                              19th-century England because they had the
                              capacity to perceive technology “in the
                              present tense”, in the <a
                                href="https://books.google.com/books?id=csndAGqJlk0C">words</a>
                              of the historian David F Noble. They
                              didn’t wait patiently for the glorious
                              future promised by the gospel of progress.
                              They saw what certain machines were doing
                              to them in the present tense – endangering
                              their livelihoods – and dismantled them.</p>
                            <p>We are often sold a similar bill of
                              goods: big tech companies talk incessantly
                              about how “AI” and digitization will bring
                              a better future. In the present tense,
                              however, putting computers everywhere is
                              bad for most people. It enables
                              advertisers, employers and cops to
                              exercise more control over us – in
                              addition to helping heat the planet.</p>
                            <p>Fortunately, there are latter-day
                              Luddites working to stem the tide.
                              Community groups like the <a
                                href="https://stoplapdspying.org/">Stop
                                LAPD Spying Coalition</a> are organizing
                              to shut down algorithmic policing
                              programs. A growing campaign to ban the
                              government use of facial recognition
                              software has won important victories in <a
href="https://www.nytimes.com/2019/05/14/us/facial-recognition-ban-san-francisco.html">San
                                Francisco</a> and <a
href="https://www.bostonglobe.com/metro/2019/06/27/somerville-city-council-passes-facial-recognition-ban/SfaqQ7mG3DGulXonBHSCYK/story.html">Somerville</a>,
                              Massachusetts, while workers at Amazon are
                              <a
href="https://gizmodo.com/amazon-workers-demand-jeff-bezos-cancel-face-recognitio-1827037509">calling</a>
                              for the company to stop selling such
                              software to law enforcement. And in the
                              streets of Hong Kong, protesters are
                              developing techniques for evading the
                              algorithmic gaze, using lasers to confuse
                              facial recognition cameras and <a
                                href="https://twitter.com/Jordan_Sather_/status/1165327628825284610">cutting</a>
                              down “smart” lamp-posts equipped with
                              monitoring devices.</p>
                            <div class="RIL_IMG" id="RIL_IMG_3">
                              <figure> <img
src="https://pocket-image-cache.com//filters:no_upscale()/https%3A%2F%2Fi.guim.co.uk%2Fimg%2Fmedia%2Ff84c6ea8fbd3f984274cf30086871b58f68e6e26%2F0_263_5568_3341%2Fmaster%2F5568.jpg%3Fwidth%3D300%26quality%3D85%26auto%3Dformat%26fit%3Dmax%26s%3Df9e321ab096d5e52f06ef3dc4a6ac4e9"
                                  alt="Teenagers and students take part
                                  in a climate protest outside the White
                                  House in Washington on 13 September
                                  2019. Photograph: Nicholas
                                  Kamm/AFP/Getty Images"> <figcaption>Teenagers
                                  and students take part in a climate
                                  protest outside the White House in
                                  Washington on 13 September 2019.
                                  Photograph: Nicholas Kamm/AFP/Getty
                                  Images</figcaption> </figure>
                            </div>
                            <p>These are just a few possible sources of
                              inspiration for a broader movement for
                              decomputerization, one that pursues social
                              and ecological goals simultaneously. The
                              premise of the Green New Deal is that we
                              can make society greener and more
                              equitable at the same time – that we can
                              democratize as we decarbonize. We should
                              apply the same logic to our digital
                              sphere. Preventing a local police
                              department from constructing an ML-powered
                              panopticon is a matter of algorithmic,
                              social <em>and</em> climate justice. As
                              they used to say in the 1960s: one
                              struggle, many fronts.</p>
                            <p>For such a struggle to be successful,
                              however, resistance is not enough. We also
                              need a vision of the future we want.
                              Again, the history of the Luddites can be
                              helpful. In 1812, a group of Yorkshire
                              Luddites sent a factory owner a letter
                              promising continued action until “the
                              House of Commons passes an Act to put down
                              all Machinery hurtful to Commonality”.
                              Following their example, we might derive a
                              simple Luddite principle for democratizing
                              technology: we should destroy machinery
                              hurtful to the common good and build
                              machinery helpful to it.</p>
                            <p>What does this mean in practice? It’s
                              hard to think of anything more hurtful to
                              our common life than heating large
                              portions of the planet beyond habitable
                              levels. Privacy advocates have long called
                              for companies to restrict their collection
                              of data to the minimum necessary to
                              perform a service – a principle now
                              enshrined in the GDPR, the EU’s omnibus
                              data regulation. A 21st-century Luddism
                              should embrace this principle but go
                              further. What matters is not only how much
                              data a service collects, but what imprint
                              that service leaves upon the world – and
                              thus whether it should be performed at
                              all.</p>
                            <p>Decomputerization doesn’t mean <em>no</em>
                              computers. It means that not <em>all</em>
                              spheres of life should be rendered into
                              data and computed upon. Ubiquitous
                              “smartness” largely serves to enrich and
                              empower the few at the expense of the
                              many, while inflicting ecological harm
                              that will threaten the survival and
                              flourishing of billions of people.</p>
                            <p>Precisely which computational activities
                              should be preserved in a less computerized
                              world is a matter for those billions of
                              people themselves to decide. The question
                              of whether a particular machine hurts or
                              helps the common good can only be answered
                              by the commons itself. It can only be
                              answered collectively, through the
                              experiment and argument of democracy.</p>
                            <p>The zero-carbon commonwealth of the
                              future must empower people to decide not
                              just <em>how</em> technologies are built
                              and implemented, but <em>whether</em>
                              they’re built and implemented. Progress is
                              an abstraction that has done a lot of
                              damage over the centuries. Luddism urges
                              us to consider: progress towards what and
                              progress for whom? Sometimes a technology
                              shouldn’t exist. Sometimes the best thing
                              to do with a machine is to break it.</p>
                          </div>
                        </div>
                      </div>
                    </div>
                  </div>
                </div>
              </div>
            </div>
          </div>
          <br>
        </div>
      </div>
    </div>
  </body>
</html>