Weeknotes: maintenance, open government, land and climate
A great listen with many Festival of Maintenance friends - BBC Analysis looking at Maintenance. We're being mis-sold a culture of innovation, and forgetting how essential looking after things is. Maybe we can have more positive framings - less make do and mend, more make the best of what we have - and make it better. Also check out this long but lovely piece from last year by Shannon Mattern (who is speaking at this year's Festival of Maintenance) - I love the themes of dust, and rust.
Zeynep Tufekci on how we tell individual stories rather than social ones:
The first trade has occured in the Open Credit Network, a new UK mutual credit network.
Tim Davies has two thoughtful and provocative articles this week. I read this one on the recent Open Government Partnership summit first - Tim calls out the conflation of technology with open government, and has several examples where it seems we are looking for change in the wrong places.
Tim's second piece looks at the poor performance of the UK on open government now, partly but not entirely due to Brexit, and the diminishing space for effective civil society engagement.
Via Francis Irving, an article about land ownership driving inequality based on a new report, Land for the Many. The Guardian piece says
Scary figures about the carbon impact of machine learning models. (If this sort of thing interests you, check out climateaction.tech - a lively community of technologists and others striving to reduce carbon emissions from tech and to create tools and advocacy to drive this more widely.)
Maker Media, who produce Make magazine and Maker Faire, has ceased operations. "“It started as a venture-backed company but we realized it wasn’t a venture-backed opportunity” Dougherty admits." It was always hard to see how this sort of business could generate the returns needed by VCs; but then also hard to see where capital to get this sort of large scale educational and community programme started might come from.
Ton Zijlstra writes briefly about corporations as single purpose algorithms, and how we forget that they, like AI, should be tools serving our needs, not overlords.
Don Norman has a nice rant about how unuserfriendly modern design has become. I took Nielsen Norman Group training many years ago, on user research and interaction design. It's stuck with me - useful material.
Another rant, again well-justified, about how routing algorithms, which are potentially so useful, are still generally very fixated on car journeys, and very bad at multimodel ones. (I have no way to tell Google Maps that I am happy to cycle some distances from home on my own bike, and might be interested in hiring bikes if it makes a short journey the other side of a train trip easier.)
A fascinating essay about whether or not it's OK to publish the machine learning system that can automatically generate plausible-seeming text - from someone who claims to have replicated the unpublished results of OpenAI. It's not deeply technical - the author's points are mostly about truth and trust and the nature of human writing. (If you like this stuff, register for the 2019 Trust & Technology Initiative Symposium on 19th September!) Also good on the nature of 'curious hackers' and how Silicon Valley is full of them.
An interesting study on the ethics of innovation and experimentation in humanitarian relief, the limits of the "do no harm" principle, noting the demands of donors and others for innovation. Really interesting notes on colonialism, the constructions of emergency which enable and permit different actions, and public-private partnerships being used to dilute oversight. HT Sean McDonald, one of the co-authors, and always insightful about data issues like this.
Some of these themes probably occur in this article by Ryan Burns, but it's behind a paywall.
A nuanced look at 'tech for good' / 'data for good', whether governance is adequate, whether benefits are really realised, and when it may be 'good-washing' - the case of phone records and disease.
A study of the downsides of social impact bonds - partly an exploration of how they are usually incentivised by measuring the wrong things, but also raising questions about the things they target.
 |
| https://www.bbc.co.uk/sounds/play/m0005mrj |
Zeynep Tufekci on how we tell individual stories rather than social ones:
In fact, the souring of Game of Thrones exposes a fundamental shortcoming of our storytelling culture in general: we don’t really know how to tell sociological stories.A new tool for co-operatives to draft and manage their articles of association.
At its best, GOT was a beast as rare as a friendly dragon in King’s Landing: it was sociological and institutional storytelling in a medium dominated by the psychological and the individual.
The first trade has occured in the Open Credit Network, a new UK mutual credit network.
Tim Davies has two thoughtful and provocative articles this week. I read this one on the recent Open Government Partnership summit first - Tim calls out the conflation of technology with open government, and has several examples where it seems we are looking for change in the wrong places.
Tim's second piece looks at the poor performance of the UK on open government now, partly but not entirely due to Brexit, and the diminishing space for effective civil society engagement.
Via Francis Irving, an article about land ownership driving inequality based on a new report, Land for the Many. The Guardian piece says
Since 1995, land values in this country have risen by 412%. Land now accounts for an astonishing 51% of the UK’s net worth. Why? In large part because successive governments have used tax exemptions and other advantages to turn the ground beneath our feet into a speculative money machine.I've wondered what the University plans are for this sort of eventuality.
 | ||
| https://twitter.com/smithsam/status/1132964479799631872 |
Scary figures about the carbon impact of machine learning models. (If this sort of thing interests you, check out climateaction.tech - a lively community of technologists and others striving to reduce carbon emissions from tech and to create tools and advocacy to drive this more widely.)
Maker Media, who produce Make magazine and Maker Faire, has ceased operations. "“It started as a venture-backed company but we realized it wasn’t a venture-backed opportunity” Dougherty admits." It was always hard to see how this sort of business could generate the returns needed by VCs; but then also hard to see where capital to get this sort of large scale educational and community programme started might come from.
Ton Zijlstra writes briefly about corporations as single purpose algorithms, and how we forget that they, like AI, should be tools serving our needs, not overlords.
Don Norman has a nice rant about how unuserfriendly modern design has become. I took Nielsen Norman Group training many years ago, on user research and interaction design. It's stuck with me - useful material.
Another rant, again well-justified, about how routing algorithms, which are potentially so useful, are still generally very fixated on car journeys, and very bad at multimodel ones. (I have no way to tell Google Maps that I am happy to cycle some distances from home on my own bike, and might be interested in hiring bikes if it makes a short journey the other side of a train trip easier.)
A fascinating essay about whether or not it's OK to publish the machine learning system that can automatically generate plausible-seeming text - from someone who claims to have replicated the unpublished results of OpenAI. It's not deeply technical - the author's points are mostly about truth and trust and the nature of human writing. (If you like this stuff, register for the 2019 Trust & Technology Initiative Symposium on 19th September!) Also good on the nature of 'curious hackers' and how Silicon Valley is full of them.
An interesting study on the ethics of innovation and experimentation in humanitarian relief, the limits of the "do no harm" principle, noting the demands of donors and others for innovation. Really interesting notes on colonialism, the constructions of emergency which enable and permit different actions, and public-private partnerships being used to dilute oversight. HT Sean McDonald, one of the co-authors, and always insightful about data issues like this.
Some of these themes probably occur in this article by Ryan Burns, but it's behind a paywall.
A nuanced look at 'tech for good' / 'data for good', whether governance is adequate, whether benefits are really realised, and when it may be 'good-washing' - the case of phone records and disease.
A study of the downsides of social impact bonds - partly an exploration of how they are usually incentivised by measuring the wrong things, but also raising questions about the things they target.