I doubt that. First of all excesive battery use by means of things like busy waits are grounds for rejection, as are failing to follow the HIG. Assuming the cross compiling system does not use busy waits and polling as much as flash on other platforms do, then there is no battery life issue unless the programmer has added his only busy wait code, which would be indisputable grounds for rejection.
Regular apps would be a pain to implement with flash, since it would require simulating the standard widgets.
But there is one category of applications that is not really expected to follow the HIG, namely games. A large number of games on the iPhone are ports of games implemented in flash on other platforms.
A few I can think of off the top of my head (there are many, many, more though):
5 Minutes to Kill Yourself, Amateur Surgeon, Bloons, Boomshine, Cannabalt, Crush the Castle, FarmVille, Gravity Hook, My Li'l Bastard, Shift, Sushi cat, and Vector Runner.
Furthermore, since PopCap games has Flash versions of many of its games, which may provide an an easier way of porting its games to the iPhone, just adding in the remaining features of the full version as desired for the iPhone and iPad, and possibly tweaking the form factor and controls a bit.
True. But the method of porting flash apps to the iphone may introduce differences from how flash is run as an applet. Do you have any evidence that the recompiled flash apps for iPhone have the same issues as flash applets? (Barring of course programmers using explicit busy waits.)
Interesting. On the other hand this new source code contains several tables that map the channel numbers to frequency. If i went in and changed the value to something reasonably close, but outside the current frequency range, I suspsect that might indeed cause it to broadcast on the new frequency.
The source code also has tables for things like the transmission gain, which might make it entirely possible to broadcast above legal power, etc. That would probably depend on the hardware itself.
That AFAICT covers a completely different chip. The drivers they just released do not download firmware blobs to the chip.
These new drivers do contain some binary blobs in the form of tables without sufficent documentation as to what they are, and any hint of why for the values, but many look like they are just empirically derived parameters for the radio.
They do have a few tables of data to load as initial values for registers, and those really need to be documented. Similarly the codebase is huge, and while parts are reasonabby well documented, other parts are not. The code does not have any structure that is clear to me, although perhaps it makes good sense to those who develop wireless drivers. I really don't know.
Generally speaking, no they cannot discriminate against potential customers. They could refuse to serve black food, but must be willing to serve black people.
The hosting site may refuse to host content about black people, but they must be willing to host content for black people.
There are some exceptions though. In many places private groups, especially those with religious affiliations, but even sometimes secular groups, may discriminate against certain protected attributes. If a restaurant is affiliated with such a group, they may legally have a policy of only serving members. If the membership policy of the group in question discriminates based on said protected attribute, the end result is a restaurant that is legally discriminating on that protected attribute.
The exact laws differ from place to place, since the anti-discrimination laws are generally a combination of the federal laws and state laws.
But Rackspace is already limited in how they can control the use of their resources. Would they be allowed to say "We don't allow websites about black people on OUR resources." No.
Actually, they could. It would be a terrible idea for business, but it would be perfectly legal. What they cannot do is refuse to hire black people, nor can they deliberately create a hostile working environment, or discriminate in pay or benefits.
True. The real issue though is that they seem to be doing this ith the intention of starting a fight. Unlike most other actions, there is little legitimate reason for a church to be burning copies of the Quran.
That is unlike the infamous mohommad cartoons, which were intended to provide humor. Unlike building a mosque near ground zero, which would have a legitimate function as a place of worship.
If the only plausible purpose is to incite hate or violence, then we must assume that is the intent. Attempting to incite hate or violence is a crime, and for good reason too.
Very interesting idea! The ipad does sound like it could wok well for that sort of thing. Have an app that displays charts, and optionally plots your location (acording to GPS) on one. Then you could tap a radio-navigation beacon on the map to open up a page giving the details in nice big print, Tap on an airport to get a similar information screen, but this one letting you pick a runway to get a screen with runway specific information.
Wow. That does sound nice. Combined with a directory style look-up mechanism for the airports and navigation beacons that shows the detail page and lets you jump to the appropriate chart, and that would be one handy tool.
Not to mention the ability to have other (potentially) helpful applications, like a flight computer, or one that can display the various flight checklists.
My doctor happens to own an old (pre-iPad) HP tablet, that has a detachable keyboard that can alternatively swivel and lock into place under the main device for when using it as a tablet. That sort of thing was common in pure tablets. Then you also had the pseudo-tablets, which were laptops (all the main electronics under the keyboard) but with screens that could pivot around and fold against the keyboard.
Those older tablets always used digitizer technology (like a wacom pad) for the screen,requiring the use of a stylus. the iPad's biggest innovation here was to use a touchscreen and special written applications. Further they decided not to feature things like handwriting recognition that was considered a major selling point in existing tablets, despite never working very well.
But have it reviewed by 120 academics in CS, and you would likely have more subject matter experts reviewing the article. I can't imagine that very many of the 120 philosophers are experts in quantum computing. Indeed many of them probably have almost no knowledge or interest in computational theory, while all computer scientists have (or should have) at least a passing interest in computational theory.
That is obviously not the only thing it can do. In P time it can solve P problem (much like a classical computer, but potentially using $\sqrt{classical}$ time, if it meets the above requirements. You can use quantum computing to find (with any probability of your choice which is less than one) the solution to a BPP problem in P time, which is again just like classical computers. Something new here is the ability to solve BQP problems (with any chosen probality less than one) in P time.
That last one is the killer. That is because two of the "hard" problems we use in asymmetric cryptography are BQP, namely integer factroization and discrete logarithms are in BQP.[1]
What we really want is asymmetric encryption based on an NP-complete problem where many instances can be shown to take no less time (asymptotically) than the hardest instances to solve (i.e. many instances are tied for the hardest), and an easy way to generate instances of this hardest problem, and corresponding solution. That is really tricky, as many FNP problems that are not optimization problems (not NPO) have many instances that can be solved in only P time.
Footnote: [1] Actually that is not strictly true. The problems have more than a yes or no answer, making them FBQP problems. But FBQP-complete problems take no longer to solve than BQP-complete problems. So quantum computers can solve FBQP with any given probability of success in only P time.
All three phases? I'm no pilot, but I'd argue that there are many more than that! Pre-flight Taxi. Takeoff. Ascent. Level flight. Descent. Holding. Approach. Landing. Post-flight taxi.
A jetliner's main autopilot system, combined with the autothrottle, are more than able to do ascent, level flight, and descent, and probably also holding, with the pilots doing little more than entering the ATC instructions, and monitoring everything.
My understanding is not using autopilot at all during level flight in jetliner is unusual. After all, the pilot would prefer to avoid continuous adjustments, and having to monitor all the major instruments constantly, while with the autopilot they don't need to worry much about whatever attribute the autopilot is maintaining, and thus can check the corresponding instrument a bit less often, with discrepancies generally indicative of some problem with the autopilot.
The ILS-based autoland systems can do approach and landing.
I've never seen or heard much about autopilots for take-off. Honestly, I would just as soon have the pilot in control then. A smooth take-off is an easy enough thing for the pilots to do by hand, and there is no system for guiding takeoffs that I know of, besides potentially using ILS in reverse.
True, which is why most TVs do not have one. What thy do have is a color saturation control, which is basically an amp offering gains in the range of 0-2 or so on the color signal, before it is applied to the luminosity signal. This actually makes sense, as the luminosity and color signals are not broadcast on the same carrier frequency, so different antenna designs may have the color signal come in with greater or less amplitude relative to the luminance than is expected.
The ability to have the amp go all the way down to zero would have been easy enough to design, and would allow color televisions to be used to determine how an image would appear on a B&W television, so being useful and very easy to add since the amp was already there, it was added.
How would onion routing not prevent that? The messages to the onion router are encrypted, so the ISP can see that the user is using encrypted comms, but has no idea who is receiving the encrypted message.
Onion routing is based on an incomplete earlier anonymous remailing system created by a subset of the TOR team (the type II system: Mixminion). The routing was similar technically, except that a message would normally have ten relays rather than 3. Besides the messages being encrypted, the links between nodes were also encrypted with TLS. This does not add any security that I know of, but was such an easy thing to add, and might help mitigate any weaknesses in the core system.
Under mixminion, a relay only knew the encrypted contents of the message, the previous relay, the next relay, and if it was the last relay in the chain. It had no idea if it was the first or next-to-last relay. Only the very first node (other than he sender) has any idea who the sender is, but that node does not even know that the sender is really the sender.
A complicated system of single use reply blocks would permit having a pseudonym server, which collects reply emails sent to a pseudonym e-mail address, and stores them for later retrieval by sending single use reply blocks, which allows the pseudonym server to encrypt the reply and put it in the system, all without knowing anything more than the first relay to use.
Yahoo's web search is powered by Bing, which explains the extreme similarity between the two.
I do find it absurd that Altavista (which was the best engine before Google thanks to it's advanced boolean search) ranks higher than Google in Bing results. Nobody uses altavista anymore.
Page rank is actually only one of several systems used in deciding the ordering of the results. Pagerank could not account for the grouping of links from one site (including hiding any after the first two or three), nor for the search history based re-ordering, etc.
Granted those other ranking systems are also certainly not very complex, almost certainly less so than pagerank. Similarly, how the various ranking systems are combined is probably quite trivial.
That said, the exact details are not published, quite intentionally, so as to avoid people gaming the system.
However, I don't see any unfair action on Google's part. If I search for image search, Google's comes up first, but I'm not surprised, Google's is the most popular. The other two major search engines are in the top six, so they are hardly buried. A search for news brings up Google as the third result, but the rest of the results in the main page are the major news sites. The sites do seem sorted in approximate order of popularity. Yahoo and Bing's news aggregator are not on the first page, because they are not very popular. (Most people using Yahoo News are just using the news stories on a Yahoo home page, and I've never heard of anybody using Bing News).
He might have been citing wikipedia's paraphrasing of the contents of the DSM, such as in many articles where it lists the DSM diagnostic criteria. I would agree that is completely inapporiate. Wiki is not an expert in DSM diagnostic criteria, nor can we even temporally presume its paraphrasing is accurate. If that was the case, the judge made the right call.
The other (less likely) possibility is that the prosecutor was citing the DSM, but in a roundabout way. Rather than take the time to procure a copy of the DSM and find the relevant passage, he found the passage on Wikipedia, and cited Wikipedia quoting the DSM. I suspect the court would agree that the DSM is an acceptable source to counter an expert. He then attempted to use Wikipedia as prima facie evidence as to the contents of the DSM. That is questionable, since there happens to exist an ultimate authority on the contents of the contents of the DSM, namely the DSM itself. I would think that it could still be acceptable, as long as the expert has a chance to dispute the validity of the wiki quote, providing the actual DSM as a better authority than the wiki quote of it. If the expert were given such an opportunity, and did not take it, that would indicate his belief that the wiki quote of the DSM is accurate.
He was describing a system that gives a result probabilistically, with the probability of a correct response being proportional to the ease of verifying it. There are two cases, one in which the result can be easilly verified. That case would be NP, and realistically BPP. The other case has no easy way to verify, making it emphatically not NP, but the exact category is not determined. Needless to say though that it has problems much harder than NP-complete problems.
I believe the part that said "interfere with, discriminate or interfere in any way" forbids both blocking and discriminatory QOS.
Also see the phase "and offer a service that does not distinguish content, applications or services, based on the source of it or their property", which also implies discriminatory QOS is forbidden.
For now that is the case. But Lightpeak by design is a bus that can carry multiple protocols and expose them to the end system. So It can carry USB data for example, but at higher bandwidth than even USB3. That will likely be one of the more common uses of the technology, since the USB HID and mass storage specs already have driver support in just about everything.
So I would expect that chips with USB Logic and LightPeak PHY will be common. You connect it to the port with wires, just like USB, and the port will house the LED and photo-transistor (or whatever optical sensor they are using). If they have the version where the optical fiber is bundled with wires for providing power ready at the time of release, that would be very useful.
Lightpeak's main advantage should be its multiple protocols. AIUI It is basically intended to be a near universal Layer 1 replacement, usable with existing layer 2 protocols, but at higher bandwidth. So it could carry USB packets, ethernet frames, SATA frames, HDMI packets, etc.
Since it supports multiple protocols an the same fiber simultaneously, it will have two layers. One is layer 1. The other is a layer 1.5 protocol. It would probably have a few different types of frames.
One that has merely added a header to the frames/packets of the layer 2 protocol. The header would indicate which device is to receive the packets. There would also be a frame that gets sent when a device is connected that indicates the address of the device, and the protocol it speaks. At that point the host knows how to talk to the device, and can use that protocol to finish the discovery process, and begin regular communication.
Such a system if designed in that fashion would be extremely generic. All that would be needed to support another protocol is the assignment of a protocol number for the frame that is sent when the device is connected. All other data would be defined by the packet/frame format of the underlying protocol.
Is it just me, or has Gmail strayed pretty far from its original purpose.
Yes. It was created to be a web interface to a standard SMTP-based email account and should never be altered or updated to take advantage of shifting trends in communication.
Did I say that? I don't think so. Gmail was unique and innovative for email from the start. It started out with one of the largest account sizes known, and the tagging system, and a spam filtering system that tends to be more accurate than even the best stand alone systems (although admittedly is less configurable).
But I always though the chat was weird. First it assumes it is quite common to keep Gmail open in a separate tab or window. While some certainly do, not everybody does, making the chat feature less useful to them. Second, since the text chat of Google Talk has always been pure XMPP, users could use any XMPP (Jabber) client they want, including any of the far nicer (IMHO) stand alone clients. Even web-based XMPP clients from other sites could be used.
I will however not dispute that Google Talk's and Gmail Chat's ability to log the conversation, and store the logs in with the e-mail is valuable feature. But I don't really see the need to have the ability to make the conversation in Gmail itself.
Google Buzz just seems worthless. As far as I can tell, you can only see the Buzz status of a person if you are a Gmail user. Furthermore, the Buzz interface only shows the Buzz statuses of users. If it acted as a micro-blogging aggregation, that would be more useful. Similarly, if you could set it up so posting a Buzz status would update external services, that would also be useful, but right now you can only do the reverse.
If Google wanted Google Voice integration with Gmail, having it archive and search revived text messages would make good sense so you don't need to go to the Google voice site to see them. Similarly, allow the transcripts of recorded/archived calls be searchable from Gmail. That would be a new and useful feature.
I don't see the value of being able to send and revive phone calls from Gmail. Gizmo5 can already be used with Google Voice, so this new feature does not give additional capabilities to the service. It merely offers the general public a way to use a VOIP with Google Voice, ssince Gizmo5 is not accepting new registrations.
Is it just me, or has GMail strayed pretty far from its original purpose.
It added chat, using Google Talk's XMPP servers.
It recently added voice/video chat using Google Talk's chat protocols.
It added that Buzz feature which is a micro-blogging service like twitter. (You can have your twitter posts become buzz posts, but you cannot subscribe to non-gmail user's twitter feeds, so it is not very useful).
Now it added this talk feature, which is basically a web based VOIP system. You use the GTalk voice chat for in-network talk, this feature for PC to POTS, and optionally use a pre-established Google Voice account for POTS to PC.
Interesting. I recently graduated from a Well known univerisity, that required basically the same thing. One department offered both a CS and a Computer Engineering degree. The degress are very similar in required courses and electives, but with some slight differences. Both required Logic Design, and Computer Architecture I, but Computer Architecture II was only required for Computer enginneers.
Computer Architecture I required that everybody work with a RISC processor design that was developed over the course of the class. The final project required making certain changes to the existing Verilog for the design. t was not nessisary to run the code on an FPGA.
Computer Architecture II required the construction of a complete RISC microporcessor, including integration with IO peripherals, and test it on the FPGA. While students were permitted to use the processor design from Computer Architecture I, it would need substantial modifications, and some form of parallelism (be it SIMD, multi-core, hardware threading support, or something else) was required. Some students created entirely new processor designs which was also acceptable, but slightly discouraged due to time constraints.
The Lectures for this class covered in more detail the lower level functioning of processors, but the CA I class had enough of the basics that the CS students were left with a good understanding of the internals of a reasonably simple processor. The only thing that the CS students should have been exposed to in more detail but were not was floating point, which was mostly covered in CA II, at least at the hardware level. (Unfortunately no lectures on the common pitfalls of floating point were covered. I hope that one of the CS-only courses touched on that more.)
Beats me. For computers I would recommend using TIA, with the Unix epoch being defined as 1970-01-01T00:00:00 (TIA), with the stored value being the number of seconds elapsed since that time. Display time for human consumption could be in UTC, using a timezone file to determine the offset, much like is currently done for local time, and the inane Daylight Saving Time rules. Displaying values in the future for human consumption would be a bit problematic, if the exact seconds are important, since leap seconds (or changes to timezones or DST rules) cannot be accurately predicted ahead of time. However, for most human purposes, giving a time of day accurate to the second for some future event is not necessary or useful.
Of course, time in general is a very tricky thing. Special and General relativity do indicate that the concept of keeping a universally synchronized timebase is never going to work the way we would hope. We will continue to have more and more time standards created in the future, as we need them. It is just the way of things.
Even if they don't combine multiple types of plastics in the same product, requiring consumers to separate the different plastics is never going to work very well, with the result being that either an automated separation process is needed, or the collection centers will need people to manually separate the plastics.
Slashdot Mirror
I doubt that. First of all excesive battery use by means of things like busy waits are grounds for rejection, as are failing to follow the HIG.
Assuming the cross compiling system does not use busy waits and polling as much as flash on other platforms do, then there is no battery life issue unless the programmer has added his only busy wait code, which would be indisputable grounds for rejection.
Regular apps would be a pain to implement with flash, since it would require simulating the standard widgets.
But there is one category of applications that is not really expected to follow the HIG, namely games. A large number of games on the iPhone are ports of games implemented in flash on other platforms.
A few I can think of off the top of my head (there are many, many, more though):
5 Minutes to Kill Yourself, Amateur Surgeon, Bloons, Boomshine, Cannabalt, Crush the Castle, FarmVille, Gravity Hook, My Li'l Bastard, Shift, Sushi cat, and Vector Runner.
Furthermore, since PopCap games has Flash versions of many of its games, which may provide an an easier way of porting its games to the iPhone, just adding in the remaining features of the full version as desired for the iPhone and iPad, and possibly tweaking the form factor and controls a bit.
True. But the method of porting flash apps to the iphone may introduce differences from how flash is run as an applet. Do you have any evidence that the recompiled flash apps for iPhone have the same issues as flash applets? (Barring of course programmers using explicit busy waits.)
Interesting. On the other hand this new source code contains several tables that map the channel numbers to frequency. If i went in and changed the value to something reasonably close, but outside the current frequency range, I suspsect that might indeed cause it to broadcast on the new frequency.
The source code also has tables for things like the transmission gain, which might make it entirely possible to broadcast above legal power, etc. That would probably depend on the hardware itself.
That AFAICT covers a completely different chip. The drivers they just released do not download firmware blobs to the chip.
These new drivers do contain some binary blobs in the form of tables without sufficent documentation as to what they are, and any hint of why for the values, but many look like they are just empirically derived parameters for the radio.
They do have a few tables of data to load as initial values for registers, and those really need to be documented. Similarly the codebase is huge, and while parts are reasonabby well documented, other parts are not. The code does not have any structure that is clear to me, although perhaps it makes good sense to those who develop wireless drivers. I really don't know.
Generally speaking, no they cannot discriminate against potential customers. They could refuse to serve black food, but must be willing to serve black people.
The hosting site may refuse to host content about black people, but they must be willing to host content for black people.
There are some exceptions though. In many places private groups, especially those with religious affiliations, but even sometimes secular groups, may discriminate against certain protected attributes. If a restaurant is affiliated with such a group, they may legally have a policy of only serving members. If the membership policy of the group in question discriminates based on said protected attribute, the end result is a restaurant that is legally discriminating on that protected attribute.
The exact laws differ from place to place, since the anti-discrimination laws are generally a combination of the federal laws and state laws.
But Rackspace is already limited in how they can control the use of their resources. Would they be allowed to say "We don't allow websites about black people on OUR resources." No.
Actually, they could. It would be a terrible idea for business, but it would be perfectly legal. What they cannot do is refuse to hire black people, nor can they deliberately create a hostile working environment, or discriminate in pay or benefits.
True. The real issue though is that they seem to be doing this ith the intention of starting a fight. Unlike most other actions, there is little legitimate reason for a church to be burning copies of the Quran.
That is unlike the infamous mohommad cartoons, which were intended to provide humor.
Unlike building a mosque near ground zero, which would have a legitimate function as a place of worship.
If the only plausible purpose is to incite hate or violence, then we must assume that is the intent. Attempting to incite hate or violence is a crime, and for good reason too.
Very interesting idea! The ipad does sound like it could wok well for that sort of thing. Have an app that displays charts, and optionally plots your location (acording to GPS) on one. Then you could tap a radio-navigation beacon on the map to open up a page giving the details in nice big print, Tap on an airport to get a similar information screen, but this one letting you pick a runway to get a screen with runway specific information.
Wow. That does sound nice. Combined with a directory style look-up mechanism for the airports and navigation beacons that shows the detail page and lets you jump to the appropriate chart, and that would be one handy tool.
Not to mention the ability to have other (potentially) helpful applications, like a flight computer, or one that can display the various flight checklists.
Nice use of an iPad.
My doctor happens to own an old (pre-iPad) HP tablet, that has a detachable keyboard that can alternatively swivel and lock into place under the main device for when using it as a tablet. That sort of thing was common in pure tablets. Then you also had the pseudo-tablets, which were laptops (all the main electronics under the keyboard) but with screens that could pivot around and fold against the keyboard.
Those older tablets always used digitizer technology (like a wacom pad) for the screen,requiring the use of a stylus. the iPad's biggest innovation here was to use a touchscreen and special written applications. Further they decided not to feature things like handwriting recognition that was considered a major selling point in existing tablets, despite never working very well.
But have it reviewed by 120 academics in CS, and you would likely have more subject matter experts reviewing the article. I can't imagine that very many of the 120 philosophers are experts in quantum computing. Indeed many of them probably have almost no knowledge or interest in computational theory, while all computer scientists have (or should have) at least a passing interest in computational theory.
That is obviously not the only thing it can do. In P time it can solve P problem (much like a classical computer, but potentially using $\sqrt{classical}$ time, if it meets the above requirements. You can use quantum computing to find (with any probability of your choice which is less than one) the solution to a BPP problem in P time, which is again just like classical computers. Something new here is the ability to solve BQP problems (with any chosen probality less than one) in P time.
That last one is the killer. That is because two of the "hard" problems we use in asymmetric cryptography are BQP, namely integer factroization and discrete logarithms
are in BQP.[1]
What we really want is asymmetric encryption based on an NP-complete problem where many instances can be shown to take no less time (asymptotically) than the hardest instances to solve (i.e. many instances are tied for the hardest), and an easy way to generate instances of this hardest problem, and corresponding solution. That is really tricky, as many FNP problems that are not optimization problems (not NPO) have many instances that can be solved in only P time.
Footnote:
[1] Actually that is not strictly true. The problems have more than a yes or no answer, making them FBQP problems. But FBQP-complete problems take no longer to solve than BQP-complete problems. So quantum computers can solve FBQP with any given probability of success in only P time.
All three phases?
I'm no pilot, but I'd argue that there are many more than that!
Pre-flight Taxi. Takeoff. Ascent. Level flight. Descent. Holding. Approach. Landing. Post-flight taxi.
A jetliner's main autopilot system, combined with the autothrottle, are more than able to do ascent, level flight, and descent, and probably also holding, with the pilots doing little more than entering the ATC instructions, and monitoring everything.
My understanding is not using autopilot at all during level flight in jetliner is unusual. After all, the pilot would prefer to avoid continuous adjustments, and having to monitor all the major instruments constantly, while with the autopilot they don't need to worry much about whatever attribute the autopilot is maintaining, and thus can check the corresponding instrument a bit less often, with discrepancies generally indicative of some problem with the autopilot.
The ILS-based autoland systems can do approach and landing.
I've never seen or heard much about autopilots for take-off. Honestly, I would just as soon have the pilot in control then. A smooth take-off is an easy enough thing for the pilots to do by hand, and there is no system for guiding takeoffs that I know of, besides potentially using ILS in reverse.
True, which is why most TVs do not have one. What thy do have is a color saturation control, which is basically an amp offering gains in the range of 0-2 or so on the color signal, before it is applied to the luminosity signal. This actually makes sense, as the luminosity and color signals are not broadcast on the same carrier frequency, so different antenna designs may have the color signal come in with greater or less amplitude relative to the luminance than is expected.
The ability to have the amp go all the way down to zero would have been easy enough to design, and would allow color televisions to be used to determine how an image would appear on a B&W television, so being useful and very easy to add since the amp was already there, it was added.
How would onion routing not prevent that? The messages to the onion router are encrypted, so the ISP can see that the user is using encrypted comms, but has no idea who is receiving the encrypted message.
Onion routing is based on an incomplete earlier anonymous remailing system created by a subset of the TOR team (the type II system: Mixminion). The routing was similar technically, except that a message would normally have ten relays rather than 3. Besides the messages being encrypted, the links between nodes were also encrypted with TLS. This does not add any security that I know of, but was such an easy thing to add, and might help mitigate any weaknesses in the core system.
Under mixminion, a relay only knew the encrypted contents of the message, the previous relay, the next relay, and if it was the last relay in the chain. It had no idea if it was the first or next-to-last relay. Only the very first node (other than he sender) has any idea who the sender is, but that node does not even know that the sender is really the sender.
A complicated system of single use reply blocks would permit having a pseudonym server, which collects reply emails sent to a pseudonym e-mail address, and stores them for later retrieval by sending single use reply blocks, which allows the pseudonym server to encrypt the reply and put it in the system, all without knowing anything more than the first relay to use.
Yahoo's web search is powered by Bing, which explains the extreme similarity between the two.
I do find it absurd that Altavista (which was the best engine before Google thanks to it's advanced boolean search) ranks higher than Google in Bing results. Nobody uses altavista anymore.
Page rank is actually only one of several systems used in deciding the ordering of the results. Pagerank could not account for the grouping of links from one site (including hiding any after the first two or three), nor for the search history based re-ordering, etc.
Granted those other ranking systems are also certainly not very complex, almost certainly less so than pagerank. Similarly, how the various ranking systems are combined is probably quite trivial.
That said, the exact details are not published, quite intentionally, so as to avoid people gaming the system.
However, I don't see any unfair action on Google's part. If I search for image search, Google's comes up first, but I'm not surprised, Google's is the most popular. The other two major search engines are in the top six, so they are hardly buried. A search for news brings up Google as the third result, but the rest of the results in the main page are the major news sites. The sites do seem sorted in approximate order of popularity. Yahoo and Bing's news aggregator are not on the first page, because they are not very popular. (Most people using Yahoo News are just using the news stories on a Yahoo home page, and I've never heard of anybody using Bing News).
The prosecutor could have done one of two things.
He might have been citing wikipedia's paraphrasing of the contents of the DSM, such as in many articles where it lists the DSM diagnostic criteria. I would agree that is completely inapporiate. Wiki is not an expert in DSM diagnostic criteria, nor can we even temporally presume its paraphrasing is accurate. If that was the case, the judge made the right call.
The other (less likely) possibility is that the prosecutor was citing the DSM, but in a roundabout way. Rather than take the time to procure a copy of the DSM and find the relevant passage, he found the passage on Wikipedia, and cited Wikipedia quoting the DSM. I suspect the court would agree that the DSM is an acceptable source to counter an expert. He then attempted to use Wikipedia as prima facie evidence as to the contents of the DSM. That is questionable, since there happens to exist an ultimate authority on the contents of the contents of the DSM, namely the DSM itself. I would think that it could still be acceptable, as long as the expert has a chance to dispute the validity of the wiki quote, providing the actual DSM as a better authority than the wiki quote of it. If the expert were given such an opportunity, and did not take it, that would indicate his belief that the wiki quote of the DSM is accurate.
Not quite.
He was describing a system that gives a result probabilistically, with the probability of a correct response being proportional to the ease of verifying it.
There are two cases, one in which the result can be easilly verified. That case would be NP, and realistically BPP. The other case has no easy way to verify, making it emphatically not NP, but the exact category is not determined. Needless to say though that it has problems much harder than NP-complete problems.
I believe the part that said "interfere with, discriminate or interfere in any way" forbids both blocking and discriminatory QOS.
Also see the phase "and offer a service that does not distinguish content, applications or services, based on the source of it or their property", which also implies discriminatory QOS is forbidden.
For now that is the case. But Lightpeak by design is a bus that can carry multiple protocols and expose them to the end system. So It can carry USB data for example, but at higher bandwidth than even USB3. That will likely be one of the more common uses of the technology, since the USB HID and mass storage specs already have driver support in just about everything.
So I would expect that chips with USB Logic and LightPeak PHY will be common. You connect it to the port with wires, just like USB, and the port will house the LED and photo-transistor (or whatever optical sensor they are using). If they have the version where the optical fiber is bundled with wires for providing power ready at the time of release, that would be very useful.
Lightpeak's main advantage should be its multiple protocols. AIUI It is basically intended to be a near universal Layer 1 replacement, usable with existing layer 2 protocols, but at higher bandwidth. So it could carry USB packets, ethernet frames, SATA frames, HDMI packets, etc.
Since it supports multiple protocols an the same fiber simultaneously, it will have two layers. One is layer 1. The other is a layer 1.5 protocol. It would probably have a few different types of frames.
One that has merely added a header to the frames/packets of the layer 2 protocol. The header would indicate which device is to receive the packets. There would also be a frame that gets sent when a device is connected that indicates the address of the device, and the protocol it speaks. At that point the host knows how to talk to the device, and can use that protocol to finish the discovery process, and begin regular communication.
Such a system if designed in that fashion would be extremely generic. All that would be needed to support another protocol is the assignment of a protocol number for the frame that is sent when the device is connected. All other data would be defined by the packet/frame format of the underlying protocol.
Is it just me, or has Gmail strayed pretty far from its original purpose.
Yes. It was created to be a web interface to a standard SMTP-based email account and should never be altered or updated to take advantage of shifting trends in communication.
Did I say that? I don't think so. Gmail was unique and innovative for email from the start. It started out with one of the largest account sizes known, and the tagging system, and a spam filtering system that tends to be more accurate than even the best stand alone systems (although admittedly is less configurable).
But I always though the chat was weird. First it assumes it is quite common to keep Gmail open in a separate tab or window. While some certainly do, not everybody does, making the chat feature less useful to them. Second, since the text chat of Google Talk has always been pure XMPP, users could use any XMPP (Jabber) client they want, including any of the far nicer (IMHO) stand alone clients. Even web-based XMPP clients from other sites could be used.
I will however not dispute that Google Talk's and Gmail Chat's ability to log the conversation, and store the logs in with the e-mail is valuable feature. But I don't really see the need to have the ability to make the conversation in Gmail itself.
Google Buzz just seems worthless. As far as I can tell, you can only see the Buzz status of a person if you are a Gmail user. Furthermore, the Buzz interface only shows the Buzz statuses of users. If it acted as a micro-blogging aggregation, that would be more useful. Similarly, if you could set it up so posting a Buzz status would update external services, that would also be useful, but right now you can only do the reverse.
If Google wanted Google Voice integration with Gmail, having it archive and search revived text messages would make good sense so you don't need to go to the Google voice site to see them. Similarly, allow the transcripts of recorded/archived calls be searchable from Gmail. That would be a new and useful feature.
I don't see the value of being able to send and revive phone calls from Gmail. Gizmo5 can already be used with Google Voice, so this new feature does not give additional capabilities to the service. It merely offers the general public a way to use a VOIP with Google Voice, ssince Gizmo5 is not accepting new registrations.
Is it just me, or has GMail strayed pretty far from its original purpose.
It added chat, using Google Talk's XMPP servers.
It recently added voice/video chat using Google Talk's chat protocols.
It added that Buzz feature which is a micro-blogging service like twitter. (You can have your twitter posts become buzz posts, but you cannot subscribe to non-gmail user's twitter feeds, so it is not very useful).
Now it added this talk feature, which is basically a web based VOIP system. You use the GTalk voice chat for in-network talk, this feature for PC to POTS, and optionally use a pre-established Google Voice account for POTS to PC.
Interesting. I recently graduated from a Well known univerisity, that required basically the same thing. One department offered both a CS and a Computer Engineering degree. The degress are very similar in required courses and electives, but with some slight differences. Both required Logic Design, and Computer Architecture I, but Computer Architecture II was only required for Computer enginneers.
Computer Architecture I required that everybody work with a RISC processor design that was developed over the course of the class. The final project required making certain changes to the existing Verilog for the design. t was not nessisary to run the code on an FPGA.
Computer Architecture II required the construction of a complete RISC microporcessor, including integration with IO peripherals, and test it on the FPGA. While students were permitted to use the processor design from Computer Architecture I, it would need substantial modifications, and some form of parallelism (be it SIMD, multi-core, hardware threading support, or something else) was required. Some students created entirely new processor designs which was also acceptable, but slightly discouraged due to time constraints.
The Lectures for this class covered in more detail the lower level functioning of processors, but the CA I class had enough of the basics that the CS students were left with a good understanding of the internals of a reasonably simple processor. The only thing that the CS students should have been exposed to in more detail but were not was floating point, which was mostly covered in CA II, at least at the hardware level. (Unfortunately no lectures on the common pitfalls of floating point were covered. I hope that one of the CS-only courses touched on that more.)
Beats me. For computers I would recommend using TIA, with the Unix epoch being defined as 1970-01-01T00:00:00 (TIA), with the stored value being the number of seconds elapsed since that time. Display time for human consumption could be in UTC, using a timezone file to determine the offset, much like is currently done for local time, and the inane Daylight Saving Time rules. Displaying values in the future for human consumption would be a bit problematic, if the exact seconds are important, since leap seconds (or changes to timezones or DST rules) cannot be accurately predicted ahead of time. However, for most human purposes, giving a time of day accurate to the second for some future event is not necessary or useful.
Of course, time in general is a very tricky thing. Special and General relativity do indicate that the concept of keeping a universally synchronized timebase is never going to work the way we would hope. We will continue to have more and more time standards created in the future, as we need them. It is just the way of things.
Even if they don't combine multiple types of plastics in the same product, requiring consumers to separate the different plastics is never going to work very well, with the result being that either an automated separation process is needed, or the collection centers will need people to manually separate the plastics.