The trouble with this type of driverless car tech is that it's going to be as brittle as the AI it's based on. It may work fine for normal, complex or not, situations, but the day a child runs out in front of it in a way it's not been programmed to handle there's going to be a tragedy.
It's one thing if this is going to be used as a glorified cruise control, but another entirely if it's meant to be used without a qualified driver ready to take over whenever appropriate.
For this type of tech to be safe in an unrestricted environment (e.g. on public roads) it needs to be backed by human level AI (i.e don't hold your breath), not just an expert system using lasers and cameras to stay on the road and read the speed limit.
Actually the real news wasn't mentioned in this awful slashdot summary. The discount didn't help him win - it would only have reduced his losses if he lost.
The real reason he was able to win, was because the casinos were willing to drastically negotiate the rules of the game (in addition to the discount) to the point where the house had only the tiniest advantage.
The guy was of course under heavy scrutity at the casinos (gambling $100K per hand), and they didn't detect him card counting, but I suspect he probably was counting but only acted occasionally when the payoff was huge (such as the single hand where he split twice and won $800K, mentioned in the article).
One of the many rule changes he negotiated was a small hand shuffled shoe, so he may well have been tracking cards through the shuffle too, as the top players are able to, thereby giving him a further edge beyond that nominally calculatable per the agreed rules.
There's a distinction between companies that are growing earnings fast (as tech companies often are) and those that are in solid businesses but with little growth (e.g. utlity companies, or mature tech companies like Microsoft and IBM whose fast growth days are probably behind them).
For the slow or non-growing companies that are plenty profitable but not actually growing, then paying out earnigns in the form of dividends to shareholders makes sense - you get little growth in the share price, but get your dividends as the reward for owning the stock.
However, for fast growing companies, a dividend doesn't really make sense - if the comany can reinvest profits and grow the company and share price at, say, 20% a year, then would you really prefer they use their cash to give you a 1-2% dividend instead? If the company can't usefully invest cash to grow the business faster, then maybe a share buyback makes sense if the stock is cheap or reasonably valued.
Of course there are also plenty of slower growing companies that fall in the middle of this spectrum, where a small dividend may make sense.
For Apple, whose stock people are buying because of it's explosive growth, to start pay a dividend, doesn't seem to make much sense. It's essentially saying that they have run out of ideas on how to grow the company (or at least have more cash than ideas), and they anticipate growth slowing such that a dividend is their best use of cash.
Competition and exclusivity (generating hardware sales) are two sides of the same thing. Apple bought Siri to make it an Apple exclusive. It should go without saying they don't want to help a competitor gain any traction by making it available on iPhone.
Siri is about more than just selling hardware though - it's also going to become a huge revenue generator in it's own right. "Siri, book me a hotel" = $$$ to Apple. The smartphone advertizing market is up for grabs, but Siri is even better - it generates actual sales (this is the stuff Apple has temporarily taken out of Siri).
Anyway it seems that Apple may have reconsidered their position on this
Pure speculation, and most likely untrue.
Siri is a huge strategic asset to Apple. They paid $200M to buy the company, and have only just begun to deploy it (what you see today is massively de-featured from what they bought - most of the potential revenue generating interfaces have been removed from it, one can only assume temporarily until referrel fee agreements are in place).
If Apple thought Evi in any way threatened Siri they'd pull it in a heartbeat. The fact that it's still there, and they are obviously aware of it, should tell you that it's not perceived of as any kind of threat to Siri at all in terms of functionality. It's really rather damning fro Evi that they havn't been pulled from the iPhone, since it's a screaming indication that they are in no way a threat to Siri.
I'm really surprised that on a supposedly tech site like slashdot that people still don't understand what Siri is despite there being plently of write ups on it. It's all about the AI - it's nothing you can replace any time soon with a quick speech recognition tied to actions mashup.
The point to be understood here is that Siri is not merely about voice transcription, nor is it about the transfer of voice input. That is just one part of the process. The next part is using the result of its transcription algorithm as input to a natural language processing engine that likely uses various other statistical methods to pick out certain words, analyze the grammatical structure of the input, and determine the sentence's most likely intent. This is what Wolfram|Alpha attempts to do. The final part is to have the computer search what resources are available to it and provide data or perform an action that (hopefully) is what the user wanted. None of these steps are trivial.
Actually you've entirely missed what's at the core of Siri, and you're also wrongly giving Wolfram Alpha the credit for figuring out the intent of what you're asking Siri to do!
The core technology of Siri is the artificial intelligence component which was originally developed by SRI (S.R.I = "Siri") under a US army DARPA contract. The SRI project was called "Cognitive Agent that Learns and Organizes" (CALO), and was then taken by the startup company Siri who extended it into what it is today. Siri was then aquired by Apple.
The DARPA/SRI/Siri AI component is where the intelligence of Siri comes from - how it figures out what you mean (maintaining the conversation context and asking for clarification if needed) and how to do it. In some cases it might do what you ask by interfacing with applications (calendar, e-mail, etc) on your iPhone, in other cases it may do a web search or go to Wolfram Alpha to find or calculate information you've asked for, and in other cases it goes out to specialized web service to do "real world" stuff like ordering taxis or making restaurant reservations that you've asked for.
Wolfram Alpha has nothing to do with the smarts of Siri - it's merely one service that Siri uses once it's done the hard part of figuring out what you want and determining that Alpha is the appropriate way to do what you want. It's no different to Siri sometime using web search to get info for you if it figures out that's what it needs to do.
They certainly could be doing that, but they seem to be claiming to only track/group your logged-in activity, so until that changes I expect they're only doing cross-service IP based preference correlation in an anonymized fashion.... The regulatory blowback of tracking stuff your explicitly claiming not to track would seem to risky for them to do it.
Sounds like a great reason not to choose an Android based phone.
Google mail + Microsoft/Bing search + Apple iPhone seems the way to go for privacy, and arguably gives you the best of each cataegory, especially since Google search seems now to be emphasizing social media / personal search vs generic web search.
Probably the most personal information you're giving away is via search and e-mail, and they're the simplest ones to get from different providers. e.g. keep using gmail, but switch to Bing for search, or keep using Google search and switch to someone else for web mail.
For video there's no real choice other than Google, so just make sure that you're not logged in if you're viewing something you don't want to be tracked and associated with your other Google services.
I'm a bit surprised at some of the comments along the lines of "no biggie - I assumed they were doing this already.... This really does make a difference. For example if you previously limited your Google searches.to stuff you wouldn't mind your girlfriend/children/co-workers or whoever seeing as suggested search completions, but weren't so careful on youtube, now you have to worry about that... You'll be getting google search completions popping up on your screen based on your youtube viewing habits! Not so harmless, eh?
But even a regular FFT requires you to pre-specify the accuracy in terms of (maximum) number of components, so the difference appears a bit more subtle.
I wonder what the difference is between doing a length-k FFT of an n-dimensional signal vs doing a k-sparse approximation to it using this algorithm?
Does anyone know if this technique could be used to get a sparse approximation to a non-sparse signal (i.e. assume the signal only has a small number of components, and calculate what they would be to get a best match)?
That would be really useful - for example rather than JPEG/MPEG compression doing a regular FFT and throwing away the high frequency components, maybe you could use the fast sparse algorithm to quickly calculate an approximation in the first place.
No - he did not go to jail for fraud. He did go to jail because a real energy company he founded (creating oil from waste) was polluting the environment, and for alleged tax evasion, but apparently was subsequently acquitted of those charges.
You do realize that Rossi has working devices that have been inspected by dozens of scientists, right?
You do realize that NASA also has working devices they've built on similar principles, right? The fact that something works rather trumps the fact that you don't yet know how it works.
Note that in the NASA video the scientist is talking about this as a very practical device for home generation of power and heat. The physics may not be well understood, but evidently trustworthy organizations like NASA have no problem creating devices that consistenty produce heat that cannot be explained using known physics (although there are a half dozen theories about what may be happening).
The trouble is that "cold fusion" got a bad name because the voices of those who were initially unable to reproduce the effect has heard over those (such as NASA!) who were able to. These types of process are now being referred to as LENR (low-energy nuclear reactions) instead, and serious research continues.
If you think the reality of LENR heat producing devices is a secret, it's more an indication that you get your news from Fox news than it is a reflection on the reality. Try listening to NASA instead of Fox news.
The TIOBE Programming Community index is an indicator of the popularity of programming languages.
followed by:
The index can be used to check whether your programming skills are still up to date or to make a strategic decision about what programming language should be adopted when starting to build a new software system
It seems that the "popularity" they are trying to measure is one that infers some merit (i.e. popular by considered choice) to the more popular languages, otherwise using this measure to choose a language for a new project doesn't make much sense. Similarly a language isn't going to be widely useful as a career skill if it's only being used on a niche platform, so it would seem that they are expecting their popularity to be measuring breadth of use (presumably due to merit), not depth of use on a single platform (which speaks for itself).
Of course it changes the facts - it's not Objective-C that's popular - it's the iPhone that's popular. If the language itself was popular then we'd see it being used where it was a choice, not a necessity.
Some basic undrestanding of computers isn't really vocational - nowadays they are so pervasive (in all your gadgets as well as computers themselves) that it's really basic knowledge. I'd put knowledge of how computers work (incl. basic programming) in the same class as something like physical geography (how mountains, glaciers form, etc)... If you want to understand the world around you then these are basics you need to know... it's more a matter of foundational knowledge than vocational training.
Arduino is a computer, not in of itself an I/O peripheral, so if you wanted to do I/O from a Raperberry Pi via an Arduino you'd basically be sending software commands from the Pi to the Arduino then have a program running on the Arduino to do the actual I/O... it's way more efficient just to use a Gertboard and do I/O directly from the Pi.
Compare the two:
Pi + Gertboard - runs at Pi speed (>> Arduino speed), and no need for a seperate I/O program
Pi + Arduino - I/O constrained by Arduino speed, amnd requires an I/O program on the Arduino in addition to the main program on the Pi
I can't think of any situation where Pi + Arduino really makes sense. Other than Gertboard for GPIO, using USB peripherals (e.g. servo driver, etc) with Pi makes more sense.
Undocumented != Closed, and anyways you can certainly write apps to undocumented APIs if you can figure/reverse engineer what they do.. but you'd generally be unwise to as undocumented typically also means unsupported so they may disappear without warning in a subsequent release.
The article anyways isn't talking about documented vs undocumented - they are struggling for the right words to describe the trend towards "platformization" of web based products - exposing their functionality via web-services APIs as a way to:
a) directly make money via pay-per-call
b) encourage 3rd parties to help the product become entrenched and attract additional users by building customized apps on top of it
Slashdot Mirror
The trouble with this type of driverless car tech is that it's going to be as brittle as the AI it's based on. It may work fine for normal, complex or not, situations, but the day a child runs out in front of it in a way it's not been programmed to handle there's going to be a tragedy.
It's one thing if this is going to be used as a glorified cruise control, but another entirely if it's meant to be used without a qualified driver ready to take over whenever appropriate.
For this type of tech to be safe in an unrestricted environment (e.g. on public roads) it needs to be backed by human level AI (i.e don't hold your breath), not just an expert system using lasers and cameras to stay on the road and read the speed limit.
OK - I'm curious - why *ONE* share of GOOG ?
Actually the real news wasn't mentioned in this awful slashdot summary. The discount didn't help him win - it would only have reduced his losses if he lost.
The real reason he was able to win, was because the casinos were willing to drastically negotiate the rules of the game (in addition to the discount) to the point where the house had only the tiniest advantage.
The guy was of course under heavy scrutity at the casinos (gambling $100K per hand), and they didn't detect him card counting, but I suspect he probably was counting but only acted occasionally when the payoff was huge (such as the single hand where he split twice and won $800K, mentioned in the article).
One of the many rule changes he negotiated was a small hand shuffled shoe, so he may well have been tracking cards through the shuffle too, as the top players are able to, thereby giving him a further edge beyond that nominally calculatable per the agreed rules.
There's a distinction between companies that are growing earnings fast (as tech companies often are) and those that are in solid businesses but with little growth (e.g. utlity companies, or mature tech companies like Microsoft and IBM whose fast growth days are probably behind them).
For the slow or non-growing companies that are plenty profitable but not actually growing, then paying out earnigns in the form of dividends to shareholders makes sense - you get little growth in the share price, but get your dividends as the reward for owning the stock.
However, for fast growing companies, a dividend doesn't really make sense - if the comany can reinvest profits and grow the company and share price at, say, 20% a year, then would you really prefer they use their cash to give you a 1-2% dividend instead? If the company can't usefully invest cash to grow the business faster, then maybe a share buyback makes sense if the stock is cheap or reasonably valued.
Of course there are also plenty of slower growing companies that fall in the middle of this spectrum, where a small dividend may make sense.
For Apple, whose stock people are buying because of it's explosive growth, to start pay a dividend, doesn't seem to make much sense. It's essentially saying that they have run out of ideas on how to grow the company (or at least have more cash than ideas), and they anticipate growth slowing such that a dividend is their best use of cash.
France had the widely popular Minitel system since the early 80's, and the UK had Teletext.
http://en.wikipedia.org/wiki/Minitel
Competition and exclusivity (generating hardware sales) are two sides of the same thing. Apple bought Siri to make it an Apple exclusive. It should go without saying they don't want to help a competitor gain any traction by making it available on iPhone.
Siri is about more than just selling hardware though - it's also going to become a huge revenue generator in it's own right. "Siri, book me a hotel" = $$$ to Apple. The smartphone advertizing market is up for grabs, but Siri is even better - it generates actual sales (this is the stuff Apple has temporarily taken out of Siri).
Pure speculation, and most likely untrue.
Siri is a huge strategic asset to Apple. They paid $200M to buy the company, and have only just begun to deploy it (what you see today is massively de-featured from what they bought - most of the potential revenue generating interfaces have been removed from it, one can only assume temporarily until referrel fee agreements are in place).
If Apple thought Evi in any way threatened Siri they'd pull it in a heartbeat. The fact that it's still there, and they are obviously aware of it, should tell you that it's not perceived of as any kind of threat to Siri at all in terms of functionality. It's really rather damning fro Evi that they havn't been pulled from the iPhone, since it's a screaming indication that they are in no way a threat to Siri.
I'm really surprised that on a supposedly tech site like slashdot that people still don't understand what Siri is despite there being plently of write ups on it. It's all about the AI - it's nothing you can replace any time soon with a quick speech recognition tied to actions mashup.
Actually you've entirely missed what's at the core of Siri, and you're also wrongly giving Wolfram Alpha the credit for figuring out the intent of what you're asking Siri to do!
The core technology of Siri is the artificial intelligence component which was originally developed by SRI (S.R.I = "Siri") under a US army DARPA contract. The SRI project was called "Cognitive Agent that Learns and Organizes" (CALO), and was then taken by the startup company Siri who extended it into what it is today. Siri was then aquired by Apple.
The DARPA/SRI/Siri AI component is where the intelligence of Siri comes from - how it figures out what you mean (maintaining the conversation context and asking for clarification if needed) and how to do it. In some cases it might do what you ask by interfacing with applications (calendar, e-mail, etc) on your iPhone, in other cases it may do a web search or go to Wolfram Alpha to find or calculate information you've asked for, and in other cases it goes out to specialized web service to do "real world" stuff like ordering taxis or making restaurant reservations that you've asked for.
Wolfram Alpha has nothing to do with the smarts of Siri - it's merely one service that Siri uses once it's done the hard part of figuring out what you want and determining that Alpha is the appropriate way to do what you want. It's no different to Siri sometime using web search to get info for you if it figures out that's what it needs to do.
They certainly could be doing that, but they seem to be claiming to only track/group your logged-in activity, so until that changes I expect they're only doing cross-service IP based preference correlation in an anonymized fashion.... The regulatory blowback of tracking stuff your explicitly claiming not to track would seem to risky for them to do it.
Sounds like a great reason not to choose an Android based phone.
Google mail + Microsoft/Bing search + Apple iPhone seems the way to go for privacy, and arguably gives you the best of each cataegory, especially since Google search seems now to be emphasizing social media / personal search vs generic web search.
Don't use Google for everything.
Probably the most personal information you're giving away is via search and e-mail, and they're the simplest ones to get from different providers. e.g. keep using gmail, but switch to Bing for search, or keep using Google search and switch to someone else for web mail.
For video there's no real choice other than Google, so just make sure that you're not logged in if you're viewing something you don't want to be tracked and associated with your other Google services.
I'm a bit surprised at some of the comments along the lines of "no biggie - I assumed they were doing this already.... This really does make a difference. For example if you previously limited your Google searches.to stuff you wouldn't mind your girlfriend/children/co-workers or whoever seeing as suggested search completions, but weren't so careful on youtube, now you have to worry about that... You'll be getting google search completions popping up on your screen based on your youtube viewing habits! Not so harmless, eh?
Ha ha - scratch that question - it is an approximation method!
But even a regular FFT requires you to pre-specify the accuracy in terms of (maximum) number of components, so the difference appears a bit more subtle.
I wonder what the difference is between doing a length-k FFT of an n-dimensional signal vs doing a k-sparse approximation to it using this algorithm?
Does anyone know if this technique could be used to get a sparse approximation to a non-sparse signal (i.e. assume the signal only has a small number of components, and calculate what they would be to get a best match)?
That would be really useful - for example rather than JPEG/MPEG compression doing a regular FFT and throwing away the high frequency components, maybe you could use the fast sparse algorithm to quickly calculate an approximation in the first place.
No - he did not go to jail for fraud. He did go to jail because a real energy company he founded (creating oil from waste) was polluting the environment, and for alleged tax evasion, but apparently was subsequently acquitted of those charges.
http://en.wikipedia.org/wiki/Andrea_Rossi_(entrepreneur)
So, he's a for-real, engineer and entrepreneur with experience in the energy business.
You do realize that Rossi has working devices that have been inspected by dozens of scientists, right?
You do realize that NASA also has working devices they've built on similar principles, right? The fact that something works rather trumps the fact that you don't yet know how it works.
There's no secret - did you even read/view the NASA presentation and video linked in the slashdot story?
Or how about this - a Nobel Prize winning physicist who believes the Rossi reactor is for real.
http://pesn.com/2011/06/23/9501856_Nobel_laureate_touts_E-Cat_cold_fusion/
Note that in the NASA video the scientist is talking about this as a very practical device for home generation of power and heat. The physics may not be well understood, but evidently trustworthy organizations like NASA have no problem creating devices that consistenty produce heat that cannot be explained using known physics (although there are a half dozen theories about what may be happening).
The trouble is that "cold fusion" got a bad name because the voices of those who were initially unable to reproduce the effect has heard over those (such as NASA!) who were able to. These types of process are now being referred to as LENR (low-energy nuclear reactions) instead, and serious research continues.
If you think the reality of LENR heat producing devices is a secret, it's more an indication that you get your news from Fox news than it is a reflection on the reality. Try listening to NASA instead of Fox news.
I'm not so sure...
All I can see on the TIOBE site is:
The TIOBE Programming Community index is an indicator of the popularity of programming languages.
followed by:
The index can be used to check whether your programming skills are still up to date or to make a strategic decision about what programming language should be adopted when starting to build a new software system
It seems that the "popularity" they are trying to measure is one that infers some merit (i.e. popular by considered choice) to the more popular languages, otherwise using this measure to choose a language for a new project doesn't make much sense. Similarly a language isn't going to be widely useful as a career skill if it's only being used on a niche platform, so it would seem that they are expecting their popularity to be measuring breadth of use (presumably due to merit), not depth of use on a single platform (which speaks for itself).
Of course it changes the facts - it's not Objective-C that's popular - it's the iPhone that's popular. If the language itself was popular then we'd see it being used where it was a choice, not a necessity.
If there are many more like this one:
http://yro.slashdot.org/story/12/01/10/1450255/ibm-snags-patent-on-half-day-off-of-work-notifications
then color me unimpressed.
Next up - IBM patents starting fire by rubbing sticks together.
Some basic undrestanding of computers isn't really vocational - nowadays they are so pervasive (in all your gadgets as well as computers themselves) that it's really basic knowledge. I'd put knowledge of how computers work (incl. basic programming) in the same class as something like physical geography (how mountains, glaciers form, etc)... If you want to understand the world around you then these are basics you need to know... it's more a matter of foundational knowledge than vocational training.
No - it'll show the porn your friends and family like.
Imagine goatse with a google annotation "your dad liked this".
Or worse.
Yee haw!
It was good while it lasted.
Arduino is a computer, not in of itself an I/O peripheral, so if you wanted to do I/O from a Raperberry Pi via an Arduino you'd basically be sending software commands from the Pi to the Arduino then have a program running on the Arduino to do the actual I/O... it's way more efficient just to use a Gertboard and do I/O directly from the Pi.
Compare the two:
Pi + Gertboard - runs at Pi speed (>> Arduino speed), and no need for a seperate I/O program
Pi + Arduino - I/O constrained by Arduino speed, amnd requires an I/O program on the Arduino in addition to the main program on the Pi
I can't think of any situation where Pi + Arduino really makes sense. Other than Gertboard for GPIO, using USB peripherals (e.g. servo driver, etc) with Pi makes more sense.
Undocumented != Closed, and anyways you can certainly write apps to undocumented APIs if you can figure/reverse engineer what they do.. but you'd generally be unwise to as undocumented typically also means unsupported so they may disappear without warning in a subsequent release.
The article anyways isn't talking about documented vs undocumented - they are struggling for the right words to describe the trend towards "platformization" of web based products - exposing their functionality via web-services APIs as a way to:
a) directly make money via pay-per-call
b) encourage 3rd parties to help the product become entrenched and attract additional users by building customized apps on top of it