I am fed up.
Linux as a Desktop OS pros:
Linux as a Desktop OS cons:
- Does not work
- Have you ever used something free that does not work?
My bona fides: I installed my first Slackware Linux v3.0 distribution back in 1995. I have used extensively various Linux distributions since then for studying, research, and work. I have compiled kernels, implemented modules, written countless applications, and administered hundreds of boxes. Linux has been my primary desktop OS for work for the past 15 years. I have been through the nonsense of using the command line to do everything (including wasting five years “programming” with emacs; geeky remark: “I don’t want to take my fingers of the keyboard”, reasonable person’s reply: “ok then, push 10 keys like a monkey just to copy/paste a few words, rather than doing a swipe and click…”), and having to remember countless insignificant details (command line tools, flags, configuration files, etc.) for configuring various aspects of the OS. At the same time I have been an avid Windows user, mostly for fun stuff, like playing games and using devices initially not supported on Linux (e.g., sound cards, webcams, tuners, fancy printers). Hence, I have been through the pain of having to re-install Windows 98 every other week, and experiencing blue screens every other day. I switched to using Windows XP as my desktop OS, at work, in 2006 and I am still using it very productively to this day. I also switched to using Ubuntu as my desktop OS, at home, in 2008, and I am still using it as I type this post.
Here is the bottom line of my experiences: Irrespective of which Linux distribution I have tried the past four years (including Mint, Ubuntu and CentOS), the state of Linux as a desktop OS is worst than what Windows 98 used to be, more than thirteen years ago. This is appalling. I can honestly say that every time I use my Ubuntu machine to try to do something new (and sometimes when trying to do ordinary tasks that I have performed before), there is some kind of annoyance I have to deal with, which can usually take more than an hour to resolve, searching the Internet and trying different ideas — in other words a complete waste of time. More importantly, one would expect things to get better with time, issues to get resolved. This has not been the case, in my experience. Each and every time I upgraded to a newer version of some distribution, a long list of things stopped working or stopped working as well as they used to. Even worse, in certain cases, bugs that had been fixed after applying incremental updates, resurfaced in a newer version of the same distribution. Finally, the stability of Linux as a desktop OS is a myth. I cannot even count the number of times that X applications crash or become unresponsive, each day; especially when it comes to dealing with multimedia applications and plug-and-play devices.
At the same time, my experience with Windows XP has been stellar. I can honestly say that I do not remember ever seeing a blue screen on my Windows XP boxes. I am sure I had various applications crash or become unresponsive at times, but it has not been very often, except the X server of Cygwin, which crashes all the time (guess why?). There is not a single thing that I haven’t been able to accomplish using Windows, and that is without having to buy an application (pretty much all major open source applications for Linux can be used with Windows as well), and without having to waste more than a couple of minutes to figure out how. On the contrary, I find it often necessary to use Virtualbox or dual-boot on my Linux boxes, to run Windows to perform certain tasks, every now and then. I have to admit, that Windows without Cygwin would be useless to me. Cygwin works fairly well (except the X server, that is). Of course, Windows is not free, and it is not very cheap either. But it seems to be worth it, given the level of frustration I have to deal with when I am trying to accomplish certain tasks on Linux.
And just to try to silence Apple fans, I own a Powerbook G4 running OS X Tiger v10.4 (which is a BSD based OS). My experience was terrible. I can deterministically cause blue screens (yes, there is such a thing in OS X) by plugging in and out a network cable while disabling wireless networking. I can crash iTunes and other apps on demand. I am not thrilled with the user experience either (I find the menu bar on the top of the screen causing unnecessary long mouse travelling and the one mouse key silly, since the need for a right-click using control-click is as frequent as a normal click). Furthermore, Apple does not support my machine anymore since it has a PowerPC CPU and Apple releases newer version of OS X for Intel CPUs only. Finally, within a month after buying the laptop the F11 key broke off. Within two years, the hard drive failed. Within three years, the little magnetic latch that is used to hold the screen tightly closed, broke off. Why does Apple have customer loyalty is beyond me. Then of course, I have no experience with newer versions of OS X…
Much has been made lately about how the Internet is ruining our memory. Well, you shouldn’t worry. Humans have been outsourcing their memories for centuries (think of the journey of the Odyssey from spoken word to written word). Jonah Lehrer explains in more detail here: Is Google Ruining Your Memory?
There is a great article by Tim Wu on the May 2011 issue of the Communications of the ACM about the advantages and caveats of monopoly in innovation.
The author argues that corporate research lab environments usually yield more innovation than academia, but tend to also stifle disrupting technologies that can harm existing business.
As an example, the author mentions Bell Labs inventing magnetic tape technology in the 1930s, and practically wiping the results because of fears that tape recordings would cut into the telephone business (somehow!), In fact, Bell Labs at the same time had also invented the answering machine (hence they needed tape!).
Magnetic tape technology resurfaced in the 1950s and led of course to magnetic storage (your hard drive) and much more.
You can read more intriguing examples in the full article here.
I was not able to find online a simple, concise explanation of Einstein’s Special Theory of Relativity. I am not a physicist, am I not an expert, I might even be naive to be trying this, but here is my take nonetheless:
Consider two coordinate systems:
- S: A train station.
- T: A train moving with constant velocity v on a straight railway line passing in front of the train station.
There are two postulates:
- The laws of nature should have the same expression irrespective of the coordinate system used to derive them.
- The speed of light is constant, and equal to c, in vacuum.
The first postulate makes intuitive sense. All coordinate systems should be equivalent for expressing the laws of nature. We should be able to derive exactly the same laws for any natural phenomenon irrespective of if we are observing that phenomenon positioned on the train station or on the train.Why would any particular coordinate system have any particular advantage over expressing the laws of nature in a simpler form?
The second postulate has been shown to be valid through various observations and experiments (here is an argument by De Sitter).
Based on these, let a ray of light travel parallel to the train at the direction of the train, and assume that an observer on the train station determines that the speed of light is c. Then, based on classical mechanics, an observer on the train should determine the speed of light to be c-v, which contradicts postulate 1. Hence, either postulate 1 is not true or classical mechanics is not correct.
Einstein claimed that the latter is true. Here is why:
- The relativity of simultaneity: Two strokes of lightning hit points A and B on the railway line simultaneously with reference to an observer standing on the train station. Assume that at the time the strokes hit with reference to the train station, a passenger on the train sits exactly on the midpoint between A and B. Are the strokes of lighting simultaneous for the passenger as well? The answer is no. Since light travels with constant speed c and the passenger travels towards point B with speed v, as far as the passenger is concerned a stoke of lightning hits B before another stroke of lighting hits A. Hence, events that are simultaneous with reference to the train station are not simultaneous with reference to the train. Every coordinate system has its own particular time.
- The relativity of distance: Consider two points A and B on the train, travelling with velocity v. We can measure the exact distance of A and B while travelling on the train using a tape. We can also measure the distance between A and B from the train station, using the same tape, by simply determining two points A’ and B’ on the rails that are being just passed by A and B at a particular time t with respect to the train station. Suppose that the distance of A and B on the train is w. Then, is it necessary that the distance of A’ and B’ on the train station is also w? Lets assume that it is NOT necessary.
Let the coordinates of an event with respect to S be x, y, z, t. Then, based on classical mechanics, the coordinates of the same event with respect to T are x’ = x – vt, y’=y, z’=z, t’=t.
But, based one the two observations above, if we allow time and distance to be relative, what are the values x’, y’, z’, t’ of the same event with respect to T such that the law of the transmission of light in vacuum is satisfied with respect to both S and T? The values can be expressed using a system of equations known as the “Lorentz Transformation“:
This transformation between the two coordinate systems does not violate postulate 1. A light ray travelling parallel to the rails advances in accordance with x=ct, and by substituting into the Lorentz transformation for x’ and t’, we get x’=ct’. The transformation was specifically derived so that it will satisfy the constancy of the propagation of light.
What are the consequences of this transformation?
- Distance contraction: Let the distance of points A and B be 1 meter on the train. Based on the Lorentz transformation we will find that the distance of A’ and B’ is . The distance has contracted as far as the observers on the train station are concerned. Nevertheless, as far as the passengers are concerned, everything is fine. Distance is relative!
- Time dilation: Similarly, consider a clock permanently fixed at position x’=0 of T. Let t’=0 and t’=1 be two consecutive ticks of the clock. As judged from the passengers on the train the time that elapses between the two ticks is 1 second. As judged by the observers on the train station, by using the Lorentz transformation, the time that elapses between the two consecutive ticks is seconds. That is, a somewhat larger time. Time is relative!
To recap, if we assume that postulates 1 and 2 make sense, then classical mechanics cannot be correct. Nevertheless, there is a simple framework (called the Special Theory of Relativity) based on the Lorentz transformation that can satisfy both postulates. All we have to do is to alter our beliefs that distance and time are absolute across all coordinate systems.
Notice that the special theory of relativity talks only about coordinate systems that move with uniform velocities with respect to each other (no acceleration, no circular motions, etc.). Einstein believed that postulate 1 should be true irrespective of the relative movement of coordinate systems. Hence, he had to come up with the general theory of relativity, but that is a different beast.
Numerous people have come up with examples that seem to lead to contradictions based on the special theory of relativity. Einstein and others were able to show that the special theory of relativity was consistent and did not lead to any contradictions in all cases. The most famous example is the twin paradox, and Wikipedia has a nice simple example. I would recommend reading it, but just to give an intuition, either twin can apply the Lorentz transformation and assume that the other twin is travelling in the opposite direction. Hence as far as either twin is concerned, the other twin has aged slower than themselves. The paradox is false, simply because the twin on earth always remains in the same coordinate system, while the twin on the ship has to change to at least three coordinate systems (one travelling towards the destination at maximum speed, one doing a u-turn, and one travelling at maximum speed back to earth).
I also recommend a book on the theory of relativity written by Einstein himself. It was the most clear explanation of the theory I could find, and all of the ideas here are based on this book.
Pointless but neat :)
If you missed Jeopardy! the past couple of days, do not miss it tonight! Watson, the IBM supercomputer (90 IBM Power 750 servers, 16 Terabytes of memory, 2,880 processor cores) is playing against two of the best Jeopardy! players, for an exciting match-up between humans and machines.
There are undertones of 2001: A Space Odyssey here, but if one looks below the surface, they realize that we are still not very close to HAL 9000. Watson is capable of understanding natural language, and particularly adept at deciphering the difficult language constructs used in Jeopardy (double meanings, expressions, puns,etc.). It (or should I say “he”?) cannot see or listen; the questions are provided to Watson in text format, so the seemingly natural interaction between Watson and the host is unfortunately not real. Nevertheless, so far Watson has been able to outplay both human opponents.
My intuition tells me that this is not because Watson is intelligent, but only because it is fast. First, humans face the electromechanical limitation of a buzzer (the thumb has to physically move to press the button). Second the majority of questions in Jeopardy! are fairly simple to answer using traditional search mechanisms (a la Google). A combination of important (high IDF) keywords directly lead to an answer. On the other hand, Watson seems incapable of answering questions correctly when the answer is not simply a keyword, but a phrase construct. Watson’s creators claim that they have used very sophisticated natural language processing algorithms. Personally, I doubt that anything very sophisticated is necessary for finding documents related to Jeopardy! questions. Of course, in order to understand the question and be able to provide the right answer, a basic understanding of parts of language is needed, but again, no new technology is necessary here. Nevertheless, the achievement is still significant. Being able to identify the actual answer fast is not trivial.
In the first match-up, Watson did fairly well. It tied for 1st place with Brad Rutter. In the second match-up Watson annihilated the competition by answering almost all of the questions on the board. Watson did make mistakes in both games, all of them silly and easy to fix. For example, in the first game Watson repeated the incorrect answer of another player. If Watson could hear or be given in text the answers of other opponents this could be avoided. In the second game Watson messed up the final Jeopardy! round question: It gave “Toronto” as an answer to a question about U.S. cities. This seems like a bug. On the other hand, Watson knew that the answer was probably wrong, so it bet a very small amount of money.
You can find the questions of the past two episodes here:
You will notice that the majority of questions are not very difficult to answer, provided that one has the information stored somewhere. A simple Google or Bing search would provide a highly relevant document for most questions. Watson is not connected to the Internet of course, but it is storing a huge database of information about everything. Here are some questions that Watson got wrong on the first day:
- Olympic Oddities: It was the anatomical oddity of U.S. gymnast George Eyser, who won a gold medal on the parallel bars in 1904. Watson’s answer: What is leg. Correct answer: he is missing a leg.
- Final Frontiers: From the Latin for “end”, this is where trains can also originate. Watson’s answer: What is finis. Correct answer: terminus
- Final Frontiers: It’s a 4-letter term for a summit; the first 3 letters mean a type of simian. Watson did not answer. Correct answer: apex
- Alternate Meanings: Stylish elegance, or students who all graduated in the same year. Watson’s answer: What is chic. Correct answer: class
- To push one of these paper products is to stretch established limits. Watson did not answer. Correct answer: envelope
The last question is instructive. It requires the creative type of reasoning that only humans are currently capable of. Watson could have identified the answer if and only if it “knew” that a common expression for stretching established limits is “to push the envelope”. A simple Google search for “to push the envelope” reveals a plethora of different definitions, none of which contain the words “established” or “limits”. I would expect Watson to be able to find an answer here by using some reasoning on synonyms, but I guess that it doesn’t (or it was not fast enough).
In the category “Name the decade” Watson failed to answer any of the questions. This is odd, given that it should be easy to identify the decade that a particular event mentioned in a question took place in. Humans were faster here.
Overall, Watson had 15 correct and 4 incorrect answers the first day. On the second day things were different. The questions, in my view, are much easier to answer. There are a lot of important keywords in each question. The language constructs are also fairly standard and easy to parse. Watson answered 23 questions correctly. It made only one mistake:
- The Art of the Steal: In May 2010 5 paintings worth $125 million by Braque, Matisse & 3 others left Paris’ Museum of this art period. Watson’s answer: Picasso. Correct answer: modern art
Watson here failed to correctly analyze the question and understand what it was really asking for. It does not seem like a difficult question to me. Watson should have been able to give a more meaningful (if not correct) answer.
Overall, I believe that the effort so far has been admirable, but I wouldn’t go so far as to say that Watson possess any sort of intelligence. Standard technologies, cleverly arranged together, should be able to perform as well as Watson. Notice also that Watson does not learn from its mistakes, neither does its understanding improve the more it plays the game. It would be interesting to see a match-up between Watson and an open source system!
Here is a link to a paper by Watson’s principal investigators, describing their approach.
I recently watched Inception for the second time, and I managed to convince myself that Cobb is not dreaming.
Here is a statistically sound way (albeit not very cost-effective) of finding out whether I am in a dream or not. We know for a fact that the map of a dream in finite, made to look infinite. All I have to do is find the end of the map. So, I drive to a random airport and take a plane to a random location. If I ever get there, I take another plane and fly to a second random location, and then a third, and so on. The more random locations I visit the more improbable it becomes that I am in a dream. Why? Because it is statistically improbable that the architect of the dream would have simulated so many random locations on the globe.
In the movie, Cobb travels from Japan, to France, to Kenya, to the United States. All along, he uses trains, helicopters and cars to travel long distances inside very complex cities. Additionally, he is the one that decides to go to France and Kenya specifically. Nobody suggested that to him. Hence, these locations are somewhat random. What are the chances that an architect had simulated such a complex environment in great detail in advance? Thus, with very high probability, Cobb is NOT dreaming.
David Owen has a fascinating article (subscription required) in The New Yorker about energy efficiency and “rebound” effects. Quoting from the article, the basic idea is that:
“When we talk about increasing energy efficiency, what we’re really talking about is increasing the productivity of energy. And, if you increase the productivity of anything, you have the effect of reducing its implicit price, because you get more return for the same money — which means the demand goes up.”
To give a concrete example, more efficient cars means more miles driven by an ever increasing number of people. The cheaper cars get, and the cheaper it gets to drive them, more people can afford to buy them, and more people can afford to use them for more and more miles per day.
The article gives another example, related to energy efficiency in refrigeration that had repercussions on a grand scale. Refrigerators are a role model of energy efficiency. Modern refrigerators are vastly more energy efficient than earlier models. Pretty much every household in the U.S. owns a modern refrigerator. But many households, instead of unplugging and recycling older models, they “demote” them to the basement and use them as freezers for meat or beverages. More importantly, refrigeration is so cheap that nowadays the refrigerated produce section of even the most remote gas station unnecessarily keeps sodas and even water refrigerated continuously. Clearly, the tremendous increase in refrigeration efficiency led directly to an explosion of refrigerator usage and inevitable waste.
The obvious question is: does the rebound effect plague all new technologies? Can we use clean energy technologies, biomedical advancements, better materials, etc. for the betterment of our society, or are all our efforts futile?