Good post, but I think the "don't collect it", "don't store it", "don't keep it" warnings may fall on deaf ears.
The reality is that businesses:
(1) aren't sure of the value of their data. Perhaps there is a company willing to buy their data and they might need that money someday. Maybe if they had more historic data, reporting would show a clearer trend in some way that would help drive them in the right direction- if they had the resources to analyze all of it.
(2) may not fully understand the laws and regulations they may be bound to that relate to how long they need to keep data.
I think it's not a bad thing to spend the time to understand your legal obligations, clean up your data, and remove what you'd never need or never realistically be able to sell. But many companies don't have that time.
The used census data from the 1930's innocently collected by govts and collated by IBM subsidiaries. The subsidiaries then passed this on to the Nazis a few years later when the govts were conquered.
Then the Hollerith machines did a query : "select person if parent=Jew or grandparent=Jew"
you didn't thinnk they did it with pencil and paper now did you ?
Thomas J. Watson, chairman of IBM, didn't get his 1937 Order of the German Eagle for nothing
Part of the brilliance of idleword's approach in this talk is in his avoidance of Godwin-level rhetoric. You don't need to posit an authoritarian dictatorship to see big data as hazardous. "Nixon in your datacenter" is the perfect image for this purpose, because it's funny in a way that "Hitler in your datacenter" never could be, and also easier to imagine.
Yeh, I realised that, hence my apology. But IBM's involvement with the Third Reich is an example where the nightmare of data retention leading to mass murder has already happened.
Every time you tick the boxes on in the equal opportunities compliance area you could be setting yourself up for disaster.
If you were a Ugandan student in Britain in 2000 you might have ticked the "[ ] Homosexual" box and had the nightmare of The Uganda Anti-Homosexuality Act, passed by the Parliament of Uganda on 20 December 2013 with the death penalty proposal dropped in favour of life in prison. [1]
It was a really huge conference, so I unfortunately didn't have a chance to interact with people after the talk. I'm also very curious about whether anyone found the argument persuasive.
I will bite. The issue with this presentation is that it create a worst of all worlds narrative, without any space for nuance. Where everything that can go wrong goes wrong.
For every russian LGBT person worried due to acquisition of LiveJournal, there are countless in other countries who have benefitted from Social Networks, who have organized & furthered their cause. Also the state apparatus is so powerful that any additional knowledge gained from LiveJournal data is likely meagre.
These slides espouse the same kind of luddite thinking that brought us Drug Wars, where rather than sane policy making we ended up with war on drugs fueled by morality and lack of introspection. One can easily construct similar set of slides showing why "Free Speech" is evil and ought to be regulated like "Radioactivity".
Also the writer has zero training in Pharmacy or BioMedicine and is talking complete non-sense when it comes to Pharma industry. Over last three decades medicine has had significant improvement thanks to innovation in devices, surgeries and adoption of "Evidence based medicine" and no the evidence comes from "Data" collected on large scale and not randomly injecting patient with chemicals. We have made huge progress in curing some types of cancers, converted HIV-AIDS into a manageable chronic condition from a death sentence. For all the visions of doom, and cherry picked examples of poor leaders, World has been a much better place.
The growing fear is that the data-driven approach is inherently a poor fit for life science. In the world of computers, we learn to avoid certain kinds of complexity, because they make our systems impossible to reason about.
But Nature is full of self-modifying, interlocking systems, with interdependent variables you can't isolate. In these vast data spaces, directed iterative search performs better than any amount of data mining.
All these statements are complete bullshit and presented with zero evidence. This article is so spectacularly bad that I am amazed. The author majored in Russian and Studio Art, and other than some programming (seemingly very proficient) has no knowledge or background in Medicine, Healthcare or Bio-Medical Sciences.
I am not zero-trained in BioMedicine. The writer is actually not too wrong about it. That field has been always evidence based, for example, you can always enjoy the evidence in Morgan or Mendel's 100 years ago paper. And the data collected in that field is usually much bigger scale than so called big-data, do you know how much DNA needed to be purified, crystallized, imaged and be discovered as double helix? And don't forget to check how drug candidates are "screened by library" -- that's one fancy name of "randomly try with chemicals".
BTW I don't think it's correct to say "curing some type of cancers", it's actually some treatments that extending the dying path.
Glivec, Chemo, Marrow Transplants are extremely effective against several types of cancers. Sure we are powerless against a significant proportion, yet we do "Cure" few types of cancers. and in spite of what Pop-Sci article you might have read, all treatments don't "just extend the dying path".
The reality is that businesses:
(1) aren't sure of the value of their data. Perhaps there is a company willing to buy their data and they might need that money someday. Maybe if they had more historic data, reporting would show a clearer trend in some way that would help drive them in the right direction- if they had the resources to analyze all of it.
(2) may not fully understand the laws and regulations they may be bound to that relate to how long they need to keep data.
I think it's not a bad thing to spend the time to understand your legal obligations, clean up your data, and remove what you'd never need or never realistically be able to sell. But many companies don't have that time.