Stone age tribesman? Take a look at the questions they're asking: who is on top on football, popular NBA players, info on the Obamas, quality of life in different regions of Uganda, the causes of sexual health problems, transmission of diseases, etc. They live in an underdeveloped country, but that doesn't mean they're underdeveloped people.

If a service like this could be sustained long-term and made accessible to more people, I think this could be a great tool. In particular, the questions about conflicting religions and sexual health are striking - there's a lot of ignorance about health, religion, and science in Africa... but that ignorance is a reflection of the state of region, not the willfull behavior of the people. Access to the Internet can provide an "out" for those that want to learn but have limited options in their village.

While it is true that whether or not a particular packet will be discarded is the result of a probabilistic process, it is unfair to call it "random". Based on a model of the queue within the router and estimation of the input parameters the probability of a packet being discarded can be calculated. In fact, that's how they design routers. You pick a bunch of different situations and decide how often you can afford to drop packets, then design a queueing system to meet those requirements. Queueing theory is a well-established field (the de-facto standard textbook was written in 1970!) and networking is one of the biggest applications.

You wouldn't expect uniform delays. A queueing system with a uniform distribution on expected number of customers in the queue is a very strange system indeed. Those sorts of systems are usually related to renewal processes and don't often show up in networking applications. That's actually a good thing, because systems with uniform distributions on just about anything are much more difficult to solve or approximate than most other systems.

"Substantial" is the key word here. Effectively the concept of managing "flows" just means that the router is caching destinations based on fields like source port, source IP address, etc. By using the cache rather than recomputing the destination the latencies can be reduced, thus reducing the number of times you need to use the queue. In queueing theory terms you are decreasing mean service time to increase total service rate. Note however that this can backfire: if you increase the variance in the service time distribution too much (some delays will be much higher when you eventually do need to use the queue) you will actually decrease performance. Of course assumedly they've done all of this work. In essence "flow management" seems to be the replacement of a FIFO queue with a priority queue in a queueing system, with priority based on caching.

Personally, I'm not sure how much of a benefit this can provide. Does it work with NAT? How often do you drop packets based on incorrect routing as compared to those you would have dropped if you had put them in the queue? If this was a truly novel queueing theory application I would have expected to see it in a IEEE journal, not Spectrum.

And of course, any time someone opens with "The Internet is broken" you have to be a little skeptical. Routing is a well-studied and complex subject; saying that you've replaced "packets" with "flows" ain't gunna cut it in my book.

Equal consideration is all that is needed. Unequal consideration means that an open source solution is considered a better choice than a closed source solution, before the individual merits of each solution are examined.

When you put open source and closed source solutions side-by-side on equal footing, you can make comparisons such as TCO, vendor lock-in, support options, timeliness of updates/upgrades, and so on. If you simply assume that open source should be preferential you're making as big a mistake as assuming that closed source should be preferential and you'll end up picking some forked project with a fading community and no documentation simply because "nobody ever got fired for choosing open source".

An equal footing is necessary - no preference either way until all the cards are laid out.

It depends what you're worried about. If you're concerned about it being proven that you were part of the distribution of the document, then any of the suggestions posed so far will work. Buy a new hard drive, encrypt it, receive the document to it from a USB device (destroy the USB device afterwards). Upload to Wikileaks from the encrypted drive. Destroy the drive.

However if you are more concerned with being thought to have been part of the distribution (as that is, after all, what's going to get you raided) then you have a bigger problem. You don't care so much whether in 5 years anybody can show you were involved, you only care about right now, can you stay under the radar.

I would build a suitable alibi (get out of town) and then receive the document from the source. Afterwards, have a change of heart, convince the source that you are NOT going to host it anywhere, convince them you have destroyed the media, and lay low for a while.

Then upload it to WikiLeaks at your leisure. If your source is convinced that you didn't upload it the Man hopefully won't think so either.

For the record, replacing memcpy with memcpy_s isn't going to magically fix any code.

For instance, let's say you wanted to do memcpy(target, src, 100), but copying 100 bytes from src to target would cause a buffer overflow. So instead you use memcpy_s(target, 100, src, 100) and cause your buffer overflow anyway.

memcpy_s makes it easier to do the right thing, but doesn't make it harder to do the wrong thing. Therefore this is a good change to C/C++.

To be honest, a single semester's worth of students at the author's own school isn't going to convince a nation-wide publisher of anything. Professors use their own texts because they think they're the best, naturally.

You're saying that in every course you've taken, each professor made you buy a $250 textbook written by himself/herself, updated every year with insignificant changes? I find that hard to believe.

In many fields there are established texts; if you find yourself in a school in which every professor writes his/her own texts and each text costs $250, you are being scammed.

That said, there is definitely a policy of "changing the figure numbers" and frequent new editions, and textbooks are often priced extremely high. But at any serious institution professors are (when appropriate) willing to let students use older editions, provide free, supplementary material, and stick to established texts.

The new edition may cost $250 the year it came out but the old edition, used and bought online, will cost you $30. I've never had a professor who wouldn't allow you to use an older edition if money was a problem, and I wouldn't take a course with one who wouldn't.

While certainly interesting, that would probably not be discussed in a textbook on data compression.

I would say that the biggest change in the past 10 years is that technology made available algorithms that previously were untenable. For instance (this is from communication and not data compression, but that's lemons to limes) LDPC codes were discovered/invented back in the 70s, and while they were theoretically the Best Thing Evar they were practically useless. Come the 90s they were rediscovered and hey presto, they're the Best Thing Evar.

Many fields move slowly, if at all - especially for the content that would be presented in a first or second text. I mentioned in another post that the standard textbook for queueing theory is a first (and only) edition of a book published in 1975! The only difference between then and now is that the companion textbook, on implementing queueing systems on "state of the art" technology, is no longer used.

It's very reasonable to assume, in the lack of contradictory evidence (which yours is not) that the textbook is still relevant.

Anecdotal to be sure, but I started reading MacKay's book online, found that it was excellent, and ordered a copy (it's a very well made hard-cover at a reasonable price).

I don't think the "10 year old" part of Wayner's book is the big deal - theoretical information theory hasn't changed too much in 10 years. Many areas of study are well-established and (for a first or second course) don't need to be terribly modern. Kleinrock's de facto standard textbook for queueing theory was published in 1975!

To me this is an issue of "would they have paid for it otherwise?" There are a lot of choices out there for textbooks in that field. With MacKay's book, I could use it first before deciding if I wanted/needed a hard copy. I'd be inclined to purchase a hard copy of a book I already knew rather than risk it with a book I have no knowledge of, even if the latter might be better than the former.

Anyway, the submitter asks " Do I (1) get another job, (2) sue people, or (3) invent some magic spell?"

Easy answer: (2). (1) is assumedly facetious and (3) will probably turn you into a toad instead.

He also asks "Is society going to be able to support people who synthesize knowledge or will we need to rely on the Wikipedia for everything?"

This is a stupid question, posed in frustration. Obviously society will support those who synthesize knowledge or there would be no knowledge to post on Wikipedia. Assuming that people want knowledge (maybe a dubious assumption!) there will be support for research.

A better question is "can I expect to make a living off of sales and royalties from a single textbook?" The answer there is "sorry, no, technology makes it too easy to pirate your work to make this a viable option anymore." That's just the way things go. Most textbook writers are paid university employees. Maybe you make a lot of extra cash from a textbook... or maybe it gets pirated and you don't make much at all... but you get paid by your employer.

Typically contraceptive drug trials require test subjects in stable, long-term relationships that have at least one child and are interested in another. The idea being that if the drug doesn't work, you're ok with a "mistake".

It's not terribly surprising that after 2.5 years a third of these couples might decide that they actively want a child, and (if the drug works) decide to stop.

I'm sure he realizes it, he just feels that it's at odds with how the world should work.

I think software-as-a-service is just one of the many difficulties that are going to be facing RMS and his black-and-white world (software) view in the near future (software-as-a-service has been around forever of course).

How is RMS going to handle on-board FPGAs for runtime code offloading into hardware? Your software, let's say it's Free, cranks through some possibly non-Free algorithm to produce (assumedly non-Free) hardware description language (softare? hardware?) which then turns into actual hardware on a non-Free device. Is it ok to use such a feature?

Let's say he answers that question in the affirmative. Now let's put that FPGA on each board in a cluster half-way around the world and let's pay for monthly access. How about now? Are we paying for software-as-a-service or hardware-as-a-service? Is it ok to use?

Again, let's say he answers in the affirmative. Now let's say the company has a phone system I can use to dial in and reboot the cluster. The software that runs the phone system is non-Free. Can I use the cluster? If my software requires a reboot as a part of its nature (and can only run on this particular system) is my software now non-Free?

Software, hardware, software-as-a-service, and other related concepts are only going to get more muddled as time goes on. The questions are going to get harder.

Free software has many practical benefits; I'm an everyday user thereof... but I'm not sure there's any moral rightness about it. As the questions get harder, if we stick to thinking of software as a moral question, as RMS tends to, we're going to elevate computer programming to a religion. When there's a desire for Free software (the open source community, for example) the need will be met (home-brew projects, Sun, IBM, Red Hat sponsored projects, etc). But is there some innate need for ALL software to be Free? I don't think so.

(Note that here I'm talking only about RMS's desire to classify each and every computing model as Free or non-Free - I'm not addressing the concept of IP ownership in general.)

I have found caffeine pills (taken sparingly, and on a full stomach) to be a handy delivery mechanism. If I have the time, coffee is definitely a tastier choice, but the pills are certainly effective.

Taken on a full stomach I find them to be only slightly stronger than a large mug of coffee.

One interesting thing I've noticed is that when I drink a cup of coffee it takes time (time, or second degree burns). During this time my mind can rest or wake-up, meaning that not only do I get the caffeine but I get a more natural rejuvenation. The caffeine pills provide the former but not the latter; I find myself very alert but my brain thinks as slow as if I just woke up. A very interesting sensation - as if I can recall being quick witted but just can't seem to muster it in the moment.

The caffeine burst helps with the immediate functions (speech, writing, parsing conversations, rote math) but not with deeper thought (argument, analysis, complex math, etc).

What, is the US government my mom? I don't have to explain myself. It's none of the US government's business if I want to have candy after dinner and stay up late on a school night.

I think part of The Problem is that people are beginning to think how you are thinking: "But how will you justify your legal actions to the government?" The proper sentiment is "Why on earth does the government care why I'm doing perfectly legal things?"

I pay my taxes. I am not a felon. I've paid my "rent" so to speak and in return I get a handful of rights, a secure nation to live in, and (theoretically) a solid economy to base my finances on. And that is where my relationship with the government ends.

(Though as a side-note: no affordable health care, no decent public transportation, poor economy, limited social rights, and a questionable education system... I'm not sure I'm getting my money's worth anymore!)

I've never really understood this. The MPAA and RIAA (and every company ever) employ people whose job it is to figure out how to maximize profit. The MPAA and RIAA, however, just keep pushing the same tune. They're searching for the peak of the parabola but they're already too far to the right - rather than backtrack they just keep going further: "maybe if we just push the same idea a little harder... the slope has to change sometime!"

The parent blames it on "egos, control, and middle manager charts that are out of touch with the real world" but to have two cartels that are from the top down completely lost? At any legitimate company that's when the board fires the CEO or the CEO shuffles the board and you hire a consultant to tell you "you're idiots, the money is all over there!" (Granted these organizations are structured differently but the concept still applies).

With every further lawsuit and lobbying attempt, I get the idea that the MPAA and RIAA just hate movies and music, and they don't want anybody else to enjoy it either. After all, that's about the only goal they're achieving.

The bots described do sound impressive, but the questions asked are mostly trivial ("How is the weather?") or predictable ("Are you the human or the computer?").

A good bot-killer should be linguistically simple, culturally independent, but require analysis of human concepts based on personal feelings.

Some bot-killers:

• How do you feel about your father's relationship with his mother?
• As a child, did you ever wonder what sort of parent you would make?
• Do you think you could ever change your opinion of economic policy?