nu23

There seem to be a lot of defensive comments, which I guess, seem to be coming from the focus on individual guilt. Focussing too much on individual choice is examining a particular solution approach and not the problem itself which needs to be solved. Lack of awareness is not a valid excuse, because even when some people speak out on the harmful consequences of technology(people have long talked about the surveillance state), they are ignored since we dont have systems in place to do a cost-benefit analysis and coordinate around a solution. Sure, there are many problems with the individual approach(what if somebody else is hired?, what if opponents develop a weapon technology?). But this only highlights the difficulty of the problem. This is why the 'we' becomes appropriate, because individuals are weak(unless they are at some critical stage).

There are potentially many developments in areas like nanotech, bioengineering, ai with which humanity might face big troubles. For this, there needs to be some strategy identifying such potential problems, and putting precautionary measures on r&d. Something like this happened with nuclear scientists hiding their research before WW2. This is harder to replicate now, with the science and engineering community distributed more widely between competing nations. So, a first step could be to acheive a shared understanding of common interests and ways to coordinate with each other.

Alternatively, energy can be poured into defensive measures for defeating harmful technologies. We already have many engineers developing counter-technologies. Also, Snowden himself was an example of exposing something which he considered harmful.

One important thing to note is that cyclone hit Orissa at low tide. Floods, more than wind speed, seem to be the the important factor. There was flooding some time after Phailin in Andhra Pradesh due to heavy rains and the casualty count was much higher than the cyclone itself. That said, the Orissa government did a good job. There was a terrible cyclone ten years ago with huge devastation, and this time they organized mass evacuations. Hope the disaster relief is strong and the Filipinos get all the help that they will be needing.

Interesting. This is a very detailed effort, with encyclopaedic ambitions across different languages. The question is if there is a lightweight version of this - someway of chosing a reasonably small set of phonemes which can represent, most English words accurately. The second part is to represent the phonemes by either using a small number of new letters and diacritics (which is what the IPA already does), or to map the phonemes directly to single letters or a string of two or three letters. With the second option, there would still be ambiguity in reading a spelling (where are the phoneme breaks?). But, the accent situation would improve and one can get used to the phoneme boundaries by looking it up for each new word.

Maybe, we should have an international language reform to introduce a phonetic spelling system (WYSIWYHear) which would also lead to a somewhat standardized accent. The local ways of pronouncing words, can still continue in each region. There are probably many aesthetic considerations for this. But when two people across different regions speak, they can use this system.

This would require linguists to document the phonemes, and to find a minimal set of alterations of either the spelling or the pronounciation of a word so that you have a phonetic system. There are already variations in spelling (American/British English). This would be another variation which is phonetic.

Of course, even if this this work is done, it would still be yet another standard needing popular adoption. The phonemes would also be more in number than the current letters which means either using diacritic symbols (requiring change in keyboards, slower typing) or mapping a phoneme to multiple letters (longer spellings and possibly some ambiguity in detecting phoneme boundaries in spellings).

The title seems apt actually. I think 'universe' here refers to the concept of a set so large that it is closed under all set theoretic operations and since mathematical operations reduce to set theoretic operations (in the usual foundation scheme) one can start with some elements in this universe set and perform any arbitrary manipulation and the output will still be inside this set. http://en.wikipedia.org/wiki/Grothendieck_universe Inter-universal presumably refers to geometrical statements which hold across different such universes. At least, that's my understanding. The paper is well beyond my knowledge.

To me, 'hack' and 'unintuitive' carry different meanings. 'Hack' indicates some kind of ad-hoc patch to fix the model. Even if it fits the data, there could be a theoretical irregularity. I suspect that programmers experience this quite often when making a change that keeps the program working. Sure, it is possible that ultimately the hack is indeed how nature works, but its still worth exploring and making precise the nature of the irregularity.

This shouldn't be surprising given that the traditional maps of the Vipassana explicitly talk about negative stages. The problem is that somehow this information is not commonly known. It would benefit practitioners a great deal. In this vipassana community, for instance, it is considered important enough to be mentioned as a sticky post right at the top, http://dharmaoverground.org/web/guest/discussion/-/message_b...

Reasoning from basic principles is a valuable tool to evaluate conclusions. Sure, it has strong weaknesses (hidden assumptions which are wrong, insufficient imagination about what could happen, rationalizing one's biases). But it is useful, when you dont have complete trust in the quality or the scope of the experiment. For instance, claims about quantum computing solving NP complete problems are legitimately held in doubt because of theoretical reasons. Also, whenever there are short term positives hiding a long term negative, like say unsustainable financial or ecological behavior, the negatives might be only seen by a chain of reasoning and not by direct experiments.

I agree in the sense that we see so many cases in the other direction - reasoning full of holes being trusted over empirics. The interesting thing is in any given situation, how much trust to give to the different tools that we have to evaluate a claim.

We need a higher resolution in our vocabulary when talking about technological risks. Someone can be in favour of most forms of technology, and yet even when dealing with technologies with extinction risk, the word that comes up to describe the opposition is 'Luddite'. Which is not to say that I am opposed to AI research.

Well, I'd love to be wrong about this and hope that you are right. What worries me is that key parts of the technology, like the processor, sensors and software might be available cheaply due to their civilian uses. Manufacturing ability could well be scarce, but these weapons could pass through illicit distribution networks which today distribute machine guns and rocket launchers. I didn't fully understand your postscript, but regarding visibilty, I admit I dont know much here - for instance, how small of drone would be realistically possible for a non-state producer. But I dont see a clear knockdown argument, given that drones are used effectively today. Again, I hope you are right and that there are more technical obstacles out there.

In twenty years or so, this could easily be widely available, just like machine guns today. Even if there was an explicit ban on the endproduct itself, the basic technology could be understood well enough to allow underground manufacturing. If true, this would allow anonymous murders and a complete breakdown of law and order. The more advanced miniature variants with face recognition software, might elude detection and jamming technologies. Popular officials might find it too dangerous just to go outside. This would provoke even more intrusive surveillance systems.

There seems to be a broader trend where our offensive technologies are moving much faster than our protective systems. At a state level, this has been handled by deterrence. Fortunately, for nuclear weapons, the necessary materials were rare enough for regulation to be possible. But deterrence wont work when these technologies become popularly available. Hopefully, there is some non-brutal solution that we can find for this problem.

In your metamorphosis example, is the finished project essentially be a robot? Would you have the same problem when a software agent is transferred from one computer to another?

Assume a negative answer. Then, the interesting implicit belief, is that there is something special about body composition. Whereas in physical law, there is nothing unique in principle about the body. So, the laws would have to be wrong or incomplete(fail to include consciousness) in an important way.

> he talks about how its impossible to make an algorithm tile a grid in a certain way.

My guess, is that here he is talking about undecidability which appears quite often in mathematical problems, maybe something like this http://en.wikipedia.org/wiki/Wang_tile or http://www.nd.edu/~svandend/Tilings.pdf (pdf).

I guess the problem is you need to pay a significant amount of money to file a claim. Interestingly, some have advocated the same method for preventing spam in email by requiring you to pay a very small amount every time you send one.

Otoh, one of the early tests for general relativity was a minor deviation in Mercury's orbit from what was predicted by classical mechanics. My rough intuition is that inconsistencies can show up initially as faint flickers, but once you investigate more closely, one can find interesting and prominent counter examples.

Sometimes though, not finding good names might just be an indication of the newness or non-standardized nature of the situation you are using the name in. Even in natural languages, there are concepts/situations which have names in one language and must be described using sentences in others. Trying to fit something into the currently popular wordlist might then be an artificial requirement.

Someone already mentioned PLOT. This would be a good thing to lookup when designing your hygienic macro system, http://users.rcn.com/david-moon/PLOT/index.html. The author was also involved in Dylan, which is another infix language with hygienic macros.

The counterpoint here would be that memorization happens more naturally when practicing, as frequently used knowledge is automatically memorized.

I remember school courses in history and chemistry where requirements of memorization was the largest component of the course credit. What I suspect is the real reason is that measuring student understanding and engagement is a hard problem, and memorization is easily measured proxy for the above. The problem is that it is easily gamed, learning decays into this gaming process and then, students promptly forget material after the exams. In my experience, though, universities mostly do allow for notes in exams.

My solution to this problem was to print the todo file whenever a terminal is opened using .bashrc. It works much better as I see the terminal more often than the desktop.

Yes, a place to take exams could be one of the important services the center provides. There could be an independent exam standards organization, which would check that the exams at center are fairly conducted.

Interestingly, since a distributed model is now possible, the content of the exams and grading procedures could be decided by the course designer. The center and exam standards body would just administer it. And the course designer could be anybody just as anybody can write a book. The only constraints would be students should be willing to take the course, which in turn could depend on the ability for the course to count as credit in a virtual major(apart from other factors like student intrinsic interest and usefulness of the course). There would probably be different organizations with different criterion for including a course in a major. A single course could be in many such major programs. So course design, physical infrastructure, exam/grading standards and formal accumulation of course into a virtual degree could all be handled independently, though I guess, in practice, some of them will be handled by one organization. Some of this has already happened.

Once we have a well-indexed high quality video and course material archive (not fully there yet), there is a lot of potential for learning centers where students can go for immersion into a study environment. These centers can offer services like meeting places for study groups to meet up, tutoring from people who have already mastered the course, and advice on what to study. Software can be used to coordinate the learning needs with expertise. The software will probably be independent of the learning center, so that students can move to different places and still continue their existing learning tracks.

These centers could provide the environment that this article talks about - essentially, many of the services that a university used to, but in cheaper and more flexible ways. Many constraints in traditional systems would become artificial in the new environment. Students could be of different ages, learn at their own pace and fit things into their own schedule. New courses could be designed based on local expertise and interest.

> It's one of the biggest metropolis areas in the world and it still has a small-town feel.

Offtopic, but this is something that was very striking when I watched, what is one of my favorite TV shows, http://wiki.d-addicts.com/Haikei,_Chichiue-sama. The show is set in Tokyo, and centers around the lives of people who run a ryotei restauarant. It has beautiful shots of the streets, and it feels very much like a small town. The whole atmosphere feels so classical that at some point in the show when some of the characters are inside a car with a modern GPS system, it feels bit surprising and one realizes that this is actually happening sometime around 2005.

I thought blocking IP addresses itself was a part of the proposed law. If that happens, this wont help. Using proxies is a solution, but this will make things much slower, especially for videos.

Yes, as long as one still needs to look at the screen there is still scope for ads. On-screen ads, if they are small enough, aren't annoying and sometimes they are useful too. There will be some interesting dynamics regarding how ad money is distributed among all the api backends.

But if voice response is used a lot, then ads(in the audio) will be sufficiently annoying for many people to switch to paying the equivalent small amount every month for ad-free services (which will have to be be distributed to the backend providers using micropayments). If this kind of service works out, then the mechanism exists for getting rid of the on-screen ads too, whether or not people actually use it. I wonder how much a the ad-value for a typical user is every month. A very rough calculation 30 billion in google's revenue/(say 200 million users * 12 months) gives 12.5 dollars/month. I think this should be an upper bound on the average, and that there will be a large variance - users who consume drugs will have larger value for advertisers, for instance.

I guess one reason people see siri as a threat to google is that if voice based interfaces take off, no matter who does it, there is less scope for advertisements - which are more annoying when spoken out loud. It's still too soon to say, of course, maybe voice apps will always be a smaller share or google could possibly find another business model.

My goal was not to defend pure empiricism - that justifications for knowledge can come only from sense data. I just wanted to emphasize that, in your example, when we chose a good explanation instead of a bad one, the goodness of the explanation has a large empirical component.I am guessing that you agree with this but would focus on the non-empirical component. One non-empirical component of its goodness would be the complexity of its specification, the Kolmogorov complexity for instance.

About the example of grass that you talk about, I think that amounts to a criticism not of empiricism but direct hypothesis checking. We have built up a model of human body and grass which tells us that it would be harmful. Now this background model has been subject to empirical checks at lots of times and has evolved in presence of new empirical discoveries. (Though of course, the model has not been derived axiomatically from empirical data).

Almost nothing in science is justified by direct checks. Think of any clinical trial or say, the LHC experiments. There are so many implicitly assumed background theories about how the instruments respond to their stimuli. Some of these instruments are based on principles which were discovered only in this century. So scientists from previous times wouldn't accept the way the results are derived unless they are shown the results derived from previous experiments justifying the way the instruments work. For example, the atomic hypothesis was controversial till 1900 or so. How would one even interpret LHC electron beams if one one doesn't believe in the atomic hypothesis? A useful analogy here, is minesweeper - a new opened square, whose information, via a very long chain can tell us if there is a mine on a far away square.

Reading this discussion on bcrypt and timing attacks, I have a question. Aren't systems like these designed so that even someone who has access to the hashed password, the randomly created salt and the hashing algorithm cannot find the password? If timing attacks help in some way, doesn't this mean some small compromise for this goal? Bruteforce attacks can of course be done faster by someone who has access but this seems to be independent from timing issues.

I just realized that I had a wrong idea about how HFT works. Though I am still not sure about how the pros and cons add up, your comment convinced me that it is more complicated than I thought.

You rightly argue that a rapid change in perceived business value is possible and furthermore that it is in the interest of any given individual investor in performing very quick trades.

But your examples support relative gain, not absolute gain. Even after HFT, there will be some investors who get rid of the hot potato before others (or the opposite acquire an attractive stock). It's just that the one's who are the first to trade will have to pay more resources to access powerful technology. Situations like these arise in many non-cooperative games, like an arms race, where an improved technology leads to increased loss to both sides but is unavoidable unless there is a scope for cooperation via a treaty.

Furthermore, this leads to a systematic bias in favour of large investors and hedge funds who are the ones with resources sufficient to win the arms race.

Note, that after HFT, the market value synchronizes faster by a short interval of time. Maybe there is an argument that this leads to a large absolute gain in some form by the way it affects investors decisions, but I dont see how it does.