recentpopularlog in

timcowlishaw : data   265

« earlier  
Digital Hygiene — Real Life
Metabolic metaphors ignore the structural factors that place internet users in peril, putting the burden on individuals to know where their data exists, how they’re being tracked, who has access to the data, and how it is being used to make decisions about them. And while these habits might make people feel like they have a modicum more control, it distracts from the real issue, which is the corporations actually doing the extracting, and the systems that allow this in the first place.
labour  solidarity  digitalhealth  race  wellbeing  dirt  metaphor  diet  digitalwellvekng  digital  privacy  data  culture  criticism  politics 
july 2019 by timcowlishaw
‘Raining clicks’: why we need better thinking on technology, data and journalism
Here’s a thing: looking at page views doesn’t actually mean you only care about pieces with numbers in the millions. It also might lead you to notice that, while populist topics have a wider potential audience (just as they always have in any medium), your long-form piece on Turkmenistan was read in full by 30,000 people. It might lead you to spot that you haven’t even properly promoted it yet and that even more people might engage with something you’re incredibly proud of. Imagine that. Imagine a world in which looking at page views doesn’t only lead you to write about kittens and completely renege on your own stated editorial ambitions and beliefs. Imagine a world in which you use data to put your excellent journalism in front of a wider audience.
data  analytics  technology  web  society  journalism  psi  news 
january 2018 by timcowlishaw
Cargo cult analytics
Here’s one weird trick I learned from Eric Ries. No, it’s actually more like a four-step program.

figure out what is important to your organization, what your goals are
think of a couple of ways in which you could move the needle on one of those goals, pick a project
assemble a team that will actually execute said project
then, and only then, think about a good metric the team can use to see whether they’re making progress.
analytics  measurement  datascience  data  journalism  media  business  psi  metrics 
october 2017 by timcowlishaw
Detecting Tanks
There's a story that's passed around to illustrate the ways machine learning can pick up on features in your dataset that you didn't expect, and probably gained the most exposure through Yudkowsky using it in "Artificial Intelligence as a Positive and Negative Factor in Global Risk" (pdf, 2008):

Once upon a time, the US Army wanted to use neural networks to automatically detect camouflaged enemy tanks. The researchers trained a neural net on 50 photos of camouflaged tanks in trees, and 50 photos of trees without tanks. Using standard techniques for supervised learning, the researchers trained the neural network to a weighting that correctly loaded the training set—output "yes" for the 50 photos of camouflaged tanks, and output "no" for the 50 photos of forest. This did not ensure, or even imply, that new examples would be classified correctly. The neural network might have "learned" 100 special cases that would not generalize to any new problem. Wisely, the researchers had originally taken 200 photos, 100 photos of tanks and 100 photos of trees. They had used only 50 of each for the training set. The researchers ran the neural network on the remaining 100 photos, and without further training the neural network classified all remaining photos correctly. Success confirmed! The researchers handed the finished work to the Pentagon, which soon handed it back, complaining that in their own tests the neural network did no better than chance at discriminating photos.
It turned out that in the researchers' dataset, photos of camouflaged tanks had been taken on cloudy days, while photos of plain forest had been taken on sunny days. The neural network had learned to distinguish cloudy days from sunny days, instead of distinguishing camouflaged tanks from empty forest.
bias  machinelearning  ai  tanks  science  datascience  data 
august 2017 by timcowlishaw
Datafication and ideological blindness — Cennydd Bowles
Replacing strategy with metric optimisation is stupid enough, but it’s even more dangerous for companies that choose the same metric as competitors.

Social networks typically make engagement their primary target, and consider it a proxy for user success. It’s now clear that among the strongest drivers of social network engagement are rich media (images and video), contemporaneity, and easy feedback mechanisms. Little wonder then that all social networks are headed toward the same territory of videos, live streaming, and push-button social grooming. It’s the preordained endgame of a battle for engagement, and so every social network starts to look the same.
metrics  data  ideology  positivism  design  products  bullshit  management  managerialism  culture  psi  parp  truthiness  product  kpis 
june 2017 by timcowlishaw
Selfwork — Real Life
Because artists’ work is not always seen as work, they are accustomed to exposure to potentially exploitative labor conditions and practices. They often know by experience the implications of having their work abstracted and systematized into market data. So they may be well-placed to recognize similar practices in other areas and offer strategies for coping, resisting, or rejecting them.
taylorism  data  measurement  efficiency  timemanagement  quantifiedself  bigdata  surveillance  politics  management  managerialism  culture  criticism 
february 2017 by timcowlishaw
We are data: the future of machine intelligence - FT.com
At the moment, Artificial Intuition is just you and the Cloud doing a little dance with a few simple algorithms. But everyone’s dance with the Cloud will shortly be happening together in a cosmic cyber ballroom, and everyone’s data stream will be communicating with everyone else’s and they’ll be talking about you: what did you buy today? What did you drink, ingest, excrete, inhale, view, unfriend, read, lean towards, reject, talk to, smile at, get nostalgic about, get angry about, link to, like or get off on? Tie these quotidian data hits within the longer time framework matrices of Wonkr, Believr, Grindr, Tinder et al, and suddenly you as a person, and you as a group of people, become something that’s humblingly easy to predict, please, anticipate, forecast and replicate. Tie this new machine intelligence realm in with some smart 3D graphics that have captured your body metrics and likeness, and a few years down the road you become sort of beside the point. There will, at some point, be a dematerialised, duplicate you. While this seems sort of horrifying in a Stepford Wife-y kind of way, the difference is that instead of killing you, your replicant meta-entity, your synthetic doppelgänger will merely try to convince you to buy a piqué-knit polo shirt in tones flattering to your skin at Abercrombie & Fitch
ai  bigdata  machinelearning  technology  politics  privacy  data 
august 2015 by timcowlishaw
New Left Review - Evgeny Morozov: Socialize the Data Centres!
for. On the one hand, we can foresee these companies extending their reach ever further into everyday life, to a point where it would become difficult to even articulate why you would want a different model, since our use of these technologies and the politics embedded in them also permits or restricts our ways of thinking about how to live. On the other hand, we can speculate about a utopian future in which technology plays the role that back in the sixties Murray Bookchin accorded it in his essays in Post-Scarcity Anarchism: helping us to live with abundance.
morozov  politics  data  privacy  righttothenetwork  marxism  internet  ownership 
february 2015 by timcowlishaw
Probable Points and Credible Intervals, Part 2: Decision Theory - Publishable Stuff
“Behind every great point estimate stands a minimized loss function.” – Me, just now This is a continuation of Probable Points and Credible …
bayesian  statistics  data  analysis  loss  utility  expectedloss  expectedutility  decisiontheory  machinelearning  optimisation 
january 2015 by timcowlishaw
Big Data Needs a Big Theory to Go with It - Scientific American
An overarching predictive, mathematical framework for complex systems would, in principle, incorporate the dynamics and organization of any complex system in a quantitative, computable framework.
complexity  complexadaptivesystems  science  data  bigdata  guff  information  statistics  haveyouheardofthomasbayes 
november 2014 by timcowlishaw
Getting Started with Big Data Architecture — blog.cloudera.com
What does a “Big Data engineer” do, and what does “Big Data architecture” look like? In this post, you’ll get answers to both questions. Apache Hadoop has come a…
bigdata  data  architecture  systems  engineering  operations  hadoop  apache 
september 2014 by timcowlishaw
Why Zipf's law explains so many big data and physics phenomenons - Data Science Central
The Zipf's law states that in many settings (that we are going to explore), the volume or size of entities is inversely proportional to a power s (s > 0…
statistics  data  zipf  paretoprinciple  pareto 
august 2014 by timcowlishaw
« earlier      
per page:    204080120160

Copy this bookmark:





to read