Funny I would not want to publish my code either, as apparently it was buggy, they would have been lash wipped by Linus!
Linus will only rant at bad code being submitted to the kernel mailinglist for integration into the mainline kernel. If you publish code on your own website, he's not even going to look at it.
Do we have any hard facts other than the unfounded assumptions of the masses that this information is being somehow data-mined?
I have no information about this doll other than the article, so no hard facts. I do consider it very likely it will be data-mined in some way, since that would actually be useful for improving the product. See my other response above.
So, is a manufacturer that respects privacy while delivering a requested product evil? Or is it just fear of the unknown operating here?
There is no proof they will respect privacy and there is no proof they won't. But when it comes to large companies, I am no longer willing to give them the benefit of the doubt. That's not fear, it's cynicism. I don't like having that attitude, but I can't justify any other attitude to myself either.
The creepy part is not a doll that listens, it is the manufacturer listening as well.
Not if you understand the technology, because you know that in order to do the first, it ALSO has to do the second.
It doesn't require a super computer to do voice recognition, as demonstrated by the $75 retail price (which would cover a few years of server-side processing, unless the article failed to mention a mandatory subscription) and a game like There Came An Echo. My guess is that a modern ARM processor would be able to handle it just fine. If the processing would drain the battery of the doll too quickly, it could be done in an in-home base station (plug server, for example) that communicates over local WiFi. There are advantages to doing the processing on a central server, but I see no reason to claim it is not feasible to do it locally.
The other way it can be phrased, which is more accurate and less creepy, is to say a SERVER processes the audio data in order to form a response. That's not creepy; it's necessary - it says nothing about humans or "the company" (whatever that means) listening in.
How I imagine it will work is that the large amount of data coming in from all the dolls in the field will be used to improve both the voice recognition and the dolls' responses. For that purpose, they would build a big data set that proposed algorithm tweaks can be tested against. Collecting this data from customers rather than dedicated test sessions is cheaper, allows for quicker iteration cycles and produces much larger data sets, which generally improves the results.
The question is whether the stored data will be anonymized early and properly or whether it will be possible to connect the stored data to a specific doll. Even if they would have no intention to violate anyone's privacy, I wouldn't trust them to get this right for a talking doll, when so many institutions dealing with medical data manage to screw it up.
Technology in itself is neither good nor bad, it all depends on how you use it.
Right, so lets not assume anything network connected that requires a server to function is bad. That's what I;m saying.
A completely different concern with an essential server-side component is the same as with always-online games: if the manufacturer pulls the plug, this doll would revert back to being just a plastic doll.
Keep a neutral attitude, and evaluate what it does rather than the scariest thing it MIGHT do.
How would you evaluate it? Even if you have the knowledge to be able to evaluate such a complex process, they wouldn't let you audit their systems.
Even apart from the data collection angle, evaluation is an issue: with a book, a parent can have a quick look at it to check whether its content is something they want their children exposed to. With an interactive online service, how the doll behaves a year from now might be different from how it behaves today. What to do if Barbie suddenly starts begging her owner for a horse or a boyfriend?
I'm not saying the potential for misuse means no-one should buy this doll, that's for every individual parent to decide. What I'm objecting to is being called a luddite over what I think are very valid concerns.
Embracing every new development out of love for technology
It's less wrong because it advances, instead of retards, humanity. The only real enemy is chaos and stagnation.
I generally consider myself progressive, but I don't consider every new thing desirable just because it hasn't been done before.
Having a doll that can talk back to you, that can intelligently respond to what you are asking and learn what you want to talk about is not creepy. That is actually really interesting. It could be really cool.
The creepy part is not a doll that listens, it is the manufacturer listening as well. An interactive doll that operates without an internet connection would be a great piece of technology.
Can the rabid un-thining pitchfork-wielding crazed mobs that roam Slashdot now please take a step back and think about the future at least once?
Technology in itself is neither good nor bad, it all depends on how you use it. Embracing every new development out of love for technology is just as irrational as rejecting it out of fear.
One of the problems is that a LED lamp can be tweaked to give a high CRI value while the color quality can still be bad. Also LED lamps with a good color quality with high saturated colors can result in a low CRI value.
The National Institute of Standards and Technology (NIST) has developed the Color Quality Scale (CQS) that it believes will offer a superior indication of color quality compared to CRI specifications.
Recently one of the largest independent lamp measurement labs in Europe started using the CSQ in all of their new lamp measurements. They also published an very informative video Color Quality Scale (CQS): Measuring the color quality of light sources which explains the problems with the CRI and why the CQS is better suited for LED lamps.
In-depth knowledge is easier to test for than general skills, so it could be that they're asking you the questions that are easiest to formulate, not the ones that are the most important. Another reason could be that it is easier to judge the competence of someone below your own level than to judge the competence of someone who is better at something than yourself; by asking you about subjects that they have mastered they feel sure they can judge your response accurately.
It is possible to create extremely complex systems by combining very simple building blocks. A game like chess has rules you can explain to a child but takes a lifetime to master. Fractals are complex images computed by repeatedly applying a simple mathematical function. People can be complex even if they are built from nothing but physics.
You're assuming that if people consist of nothing but physics, that physics would be able to explain people. In other words, that the properties of the building blocks would explain all properties of a complex system. I think that is not the case, certainly not in practice and perhaps not even in theory.
First of all, deriving the properties of the whole from the properties of the components could be too difficult for people to perform in practice. Half a century of AI research might sound like a lot, but it is really far from an exhaustive search. Even a much simpler problem like computer chess hasn't been fully solved; it took half a century just to get chess algorithms at the same level as human players, but they're still far from perfect.
Second practical problem is that tiny variations in the configuration of building blocks can have big impacts on the system as a whole. For example, weather systems are chaotic. In a computer program, changing a single instruction among millions can have big consequences. So studying physics would at best give you the ability to predict how an unconfigured brain works (like an embryo's?), not an arbitrary brain that has formed lots of connections based on the experiences of the person in question.
A reason to doubt deriving all behavior is possible in theory is that there are unprovable statements in math. So even if you know all the rules, for non-trivial rulesets (such as arithmetic) you cannot prove all the consequences of those rules.
Peak oil is the moment when the maximum production rate is reached. If you say there is no such thing as peak oil, that means you're predicting the production rate will forever increase. Even if that would be possible on the supply side (which I sincerely doubt), the demand for oil will decrease when the price increases, leading to lower production.
We're not running out of oil, we're running out of cheap oil. As oil gets more expensive, people will find ways to use less oil: other fuels, renewable energy, better home insulation, fuel efficient cars, less air travel etc. We already saw the demand for oil drop when the economic crisis hit, showing that there is flexibility in the demand.
I think Microsoft is adapting to being one of several vendors instead of being a dominant force. They have to play better with others because they don't have the market power they had 10 years ago.
Another thing that bothered me is that the documentation consists mostly of examples. However, if I read documentation I don't want a code fragment to copy-paste, I want to read the specification for a particular method. In particular, how it handles edge cases. That information was usually missing. Of course you can test the behavior, but there is no guarantee the next release will have the same behavior if the behavior was never documented. All in all, it didn't feel like a good platform for writing reliable applications.
I don't think they mean an actual device, but a product that will make the company look hip and relevant again to the masses, like the iPod was for Apple.
I'm all for electric and the end of burning fuel to drive around but you have to ask the question of WHERE that electricity is coming from to charge up your car?
Is the problem just being shifted?
In France, a lot of electricity comes from nuclear power plants, so in terms of CO2 reduction switching to electric driving would help. But that transition is going to take a while; it's unrealistic to expect everyone who now drives a diesel to buy an electric as their next car.
As far as I know, and this seems to be supported by your links, modern diesels don't pollute more than modern petrol cars. So if this would be about reducing pollution, they should crack down on old and poorly maintained diesels.
I don't know about France, but in the Netherlands students get free (but limited) public transport use, so very few students drive a car. That keeps a lot of old and poorly maintained cars off the road.
So they want a complex problem solved in 2 months (first test on Feb 4 and there are holidays inbetween), for which they will pay a relatively low amount and only to the winners. Even if the result wouldn't be used for spying, I don't think there would be many takers.
In Holland, everyone pays into the state health care system during their working years, with the money then disbursed to pay for later-in-life expenses
The Dutch health care system never worked like that. They might be confused with the pension system, where people save for their generation's retirement. While heavily regulated, the pension system is not run by the state.
The Dutch health care system is actually a lot like Obamacare, with private insurance companies which are not allowed to turn down people who apply. The system was redesigned in 2006: before that, people with low to moderate salaries were insured via their employer, while now all people pick and pay their own insurer. The increased competition in the new system hasn't stopped medical costs from rising though.
Here in the Netherlands, if something is named "juice", it can by law only contain actual juice. The water + sugar + juice mixes are named "nectar". The real stuff is more expensive (by about 50%; depends on the type of fruit), but it's available in any super market.