To suggest there is a 'fair right' to data behind encryption for goverment use, to my mind, somewhat misses the point of encryption and privacy in general. This is the point I believe we should be arguing.
Your representaiton of the article as a 'declaration against encryption' somewhat undermines this argument, and polarises it into an 'us vs them' debate.
For accuracy, this document is titled:
"Draft Council Declaration on Encryption
- Security through encryption and security despite encryption"
Its pretty handwavy, overly general, and seems to call for some sort of 'back door' from the tech companies. Missing the point really, anybody wishing to use strong encryption for criminal purpoes can do so so with very few resources, and quite independently.
If 14 year olds know more about the security vulnerabilities of our most highly adopted computer systems than the professionals which use and safeguard/build them, then it would say more about those professionals than the teens (a fact that I am highly sceptical of).
Or could it be a self reported survey says more about the perception of 'cool' subversive behaviours amongst that group than it's prevalence? I'd buy the latter over the former. You only have to wander around a university campus nowadays to understand the the days of overindulgence in drink/drugs are over -- and that being a self-identified entrepreneur or hacker is de rigeur.
What WB discusses is a common misconception. Just because it doesn't appear to work for yourself, doesn't mean that targeting is not worthwhile. It's not really designed for an improved user experience, but to provide an advantage to those running the ad campaigns.
The crucial thing that people forget, is that that targeting only has to be a little bit better than random to be worth doing (from the ad operations perspective). Given a large number of ads shown every day, with nominal cost, we've only got to push a fraction of those toward users with a slightly better propensity to interact to increase the interaction rate, and get paid more from the advertiser (presuming a model where the advertiser pays us based upon interactions that we generate).
One of the reasons that it is profitable is that the amount we might get paid for an interaction could 50p, £1 or even £10 depending upon the client and media. The cost of showing that ad is a few pence per thousand! So, we can get it wrong many many times, and still show a statistically significant improvement in interaction rates and make a huge dent in profitability for ourselves.
It really worries me that the leader of our nation does not appear to have all the information required in order to make a decision on a topic as important as this. I would have thought that a domain expert within government would have be consulted before Cameron goes off half cocked in debates and discussions.
With technology a key and growing industry within the UK, shouldn't we expect our leaders to at least attempt to understand the issues around governance?
I concur -- travelling in Boston can be frustrating for those who are inexperienced. They use 'inbound' and 'outbound', but without knowledge of the centre point for that line this information is useless!
It was written in 2012 and is a fairly well known and accessible text! Pedro is an excellent speaker, if you ever get chance to hear him I highly recommend it.
Isn't part of the problem that we place this artificial wall between academia and industry? I often found myself confused in this respect. I attained my PhD a number of years ago, and found this an invaluable experience, which has no doubt helped me to excel in industry, however I still meet people (developers and senior level executives especially) who think academia is about locking yourself in a room and writing stuff without building anything useful. Ironically, these are the same people who build products and code without surveying the field, understanding what is already out there and building on top of existing solutions. Their naivity and ignorance is such that they believe they are building something truly unique, and will get it right first time, every time.
It seems to me that we should stop standing on each side of the fence and looking down on each other, and start working together. We might actually achieve something great it if we do that.
Regards the OP, if it was the right time for him/her to leave, that's great. Nobody should continue on a path that they do not find fulfilling. It says nothing about academia, or industry though. It only reveals the feelings of the author.
The true travesty here, is that he/she will find it extremely difficult to pick up that work again and find a place in a university if they change their mind in the future. Our system in the UK is broken, since you only get a single shot at academia -- unless of course you are independently wealthy.
Perhaps this is the cause of the divide. You are either in, or you are out -- and once you make your choice there is no going back.
I think it's more to do with the application of parallel MC to approximate inference over Bayesian networks. In the normal case of approximate inference, each node would operate over the incoming data to approximate maximum likelihoods (this is the MC bit) then pass them out on each link. The key contribution of this work seems to be to tie the MC algorithms together from difference nodes. Difficult to say any more without reading the paper though.
My key question would be whether there are independence considerations that Parallel MC breaks? (Message passing works because all nodes need only information from their neighbours to summarise the rest of the network, due to the modelled independence).
I think the most dangerous thing here is the branding of research as something new, that Musk has invented. I worked in academic and industrial research for a number of years, and any work that was worth it's salt typically went back to first principles to make its point.
The problem with this is that it's hard and time it's time consuming to do this. Most companies and people don't have the time to go back to first principles for every product they build, and it would not be financially viable to do so if they did. So we compartmentalise and we make assumptions about the outcome based on prior knowledge to assess the risk of the project.
Software development is illustrative of this point. Let's say you are developing a beautiful front end for your product. You use a toolkit, based upon an existing language, which eventually will get interpreted/compiled down through multiple existing systems and run on the processor of your computer.
If I wanted to go back to first principles and make the whole thing faster, do I optimise my code the language, the compiler, the browser, the os stack, the hardware stack, CPU assembly? No, as I'd no longer be a front end developer.
Research is hard, expensive, and by it's nature, high risk. If Musk can sit on his pile of money and do it, great for him, but for most, going back to first principles for everything is not a viable option.