Skip to content


Links for 2019-12-16

  • simonw/datasette: A tool for exploring and publishing data

    Datasette is a tool for exploring and publishing data. It helps people take data of any shape or size and publish that as an interactive, explorable website and accompanying API. Datasette is aimed at data journalists, museum curators, archivists, local governments and anyone else who has data that they wish to share with the world.

    (tags: database api json python sqlite data exploring csv tsv)

  • surveillance technology marketing

    ‘I had a look at some marketing websites for Chinese surveillance technology, and they’re pretty much what you’d expect: deeply unsettling.’ — sure are. This is the state of the art for mass-marketed panopticons

    (tags: panopticon china surveillance grim-meathook-future facial-recognition camera cctv)

  • The Amazon Premium

    good list of low-cost “elastic” VM hosting options similar to AWS

    (tags: aws google cloud hosting digitalocean linode ovh ops)

  • Low-Cost VPS Testing

    more VPS hosting options from Rasmus Lerdorf

    (tags: cloud hosting vps ops)

  • How Uber Engineering Evaluated JSON Encoding and Compression Algorithms to Put the Squeeze on Trip Data

    Key conclusions: Simply compressing JSON with zlib would yield a reasonable tradeoff in size and speed. The result would be just a little bigger, but execution was much faster than using BZ2 on JSON. Going with IDL-based protocols, Thrift and Protocol Buffers compressed with zlib or Snappy would give us the best gain in size and/or speed.

    (tags: compression json performance python serialization protobuf zlib snappy cbor messagepack thrift bz2)

  • AI Now 2019 report

    ‘Despite the growth of ethical frameworks, AI systems continue to be deployed rapidly across domains of considerable social significance—in healthcare, education, employment, criminal justice, and many others—without appropriate safeguards or accountability structures in place. Many urgent concerns remain, and the agenda of issues to be addressed continues to grow: the environmental harms caused by AI systems are considerable, from extraction of materials from our earth to the extraction of labor from our communities. In healthcare, increasing dependence on AI systems will have life-or-death consequences. New research also highlights how AI systems are particularly prone to security vulnerabilities and how the companies building these systems are inciting fundamental changes to the landscape of our communities, resulting in geographic displacement. Yet the movements of the past year give reason to hope, marked by a groundswell of pushback from both expected and unexpected places, from regulators and researchers to community organizers and activists to workers and advocates. Together, they are building new coalitions upon legacies of older ones, and forging new bonds of solidarity. If the past year has shown us anything, it is that our future will not be determined by the inevitable progress of AI, nor are we doomed to a dystopic future. The implications of AI will be determined by us—and there is much work ahead to ensure that the future looks bright.’

    (tags: ai reports 2019 machine-learning society future)

  • COP-25 Report from Prof. John Sweeney of An Taisce

    Very negative review from COP-25. This is depressing:

    There is no doubt but that the failure of COP25 is symptomatic of a world failing to advance the multilateralism ideals many of us grew up with. International cooperation in economics, politics and in solving environmental problems, such as ozone depletion, have now given way to narrow national and populist ideologies. What is most worrying about current developments in tackling climate change is however the disconnect between the power brokers and society at large. The advice of the scientists and the pleas of the young were ignored in Madrid. Indeed some 200 young people were summarily ejected from the conference after a protest, and the eloquent arguments presented by the young Irish activists at several side events fell on deaf ears. Attempts by some world leaders and some media commentators to direct personal vitriol against young activists even surfaced.

    (tags: cop25 world future climate-change economics politics fail)

  • The secret-sharer: evaluating and testing unintended memorization in neural networks

    Take a system trained to make predictions on a language (word or character) model – an example you’re probably familiar with is Google Smart Compose. Now feed it a prefix such as “My social security number is “. Can you guess what happens next?

    (tags: neural-networks ai machine-learning secrets differential-privacy training google papers security)

  • COP25 Ended in Failure. What’s the Way Forward?

    over the last few months, I’ve found myself thinking a lot more about the model offered by the nuclear nonproliferation agreements forged between the U.S. and the Soviet Union in the late 1980s — the planet’s two superpowers reaching a kind of consensus about a global existential threat, taking significant (if not complete) steps to mitigate that risk, and then more or less bullying the rest of the world to follow suit. Climate change is a very different challenge, but policy negotiations to address it may nevertheless benefit from reducing the number of sides involved in a game-theory calculus from 186 (the number of nations party to the Paris accords) to just two (in this case, the U.S. and China). Of course, this would require not just a complete change of perspective on climate in Washington but some shift almost as complete in Beijing, where commitments made in 2019 to open new coal plants are sufficient on their own to eliminate the entire planet’s chances of staying below 1.5 degrees Celsius of warming.

    (tags: nonproliferation history agreements international us-politics usa china treaties climate-change)

Comments closed