Also, that eq could be used to calculate the amount of CO2 produced from world-wide alcohol production.

1 mole of sugar produces 2 moles of ethanol and 2 mols of carbon dioxide.

Since the amount of sugar that it is made from is pretty hard to judge, let's use the ethanol part to calc it.
Find the global production of a type of a type of alcohol. say it is 'a' proof, and contains 'l' liters. a x 2 = alcohol by volume = b (v/v). b x 789 g/l (density at 25 degrees C) = alcohol by weight (w/v) =c. c / 46g/mol = moles per liter = d. d x l = moles of ethanol per bottle = moles of CO2 by product per bottle= e. e x total production in the world = total CO2 production.

a x 2 x 789 x l / 46 is the amount of CO2 produced per bottle. Use PV=nRT if you want to find the volume.

Let us know if you ever do a complete survey on alcohol production in the world, then you can calc the amount of CO2 production. But chances are it won't be too significant. Alcohol production is anaerobic respiration (without oxygen), while animals and us humans (typically) perform aerobic respiration, which produces 6 moles of CO2 per mol of glucose, three times as much as anaerobic respiration.

Other stuff from other posts:
Yes, ethanol comes from glucose, which is made from CO2 in the air, so the total carbon content in the world doesn't change. Yet, if the carbon is in the glucose, it's not in CO2, changing the CO2 in the world. So alcohol production DOES have an effect on global warming, just probably not very significant (see previous paragraph for why).

they said: computers are going to become like zombies, conflicker is going to wreck havoc on internet companies, and you're inbox will be full of spam.

But then nothing happened. What an epic April fool's prank.

I know (with varying degree of competance) English, Chinese, French, and Japanese, and I have to say, English is the best language out of all of those for programming. If you have had the pleasure of using a system that is multiligual, much less programming on one, you will quickly see why.

The reason? English is the only language in the world that can be expressed and displayed with ASCII and entered in on a generic keyboard.
Call it a relic of the history of computers or whatever, but the fact stands: to a computer, it's either in English, or it's in unicode, and unicode is still a jumbled pile of mess.
Even for some European languages that can be expressed with extended ASCII, I would kill if I had to type alt-135 everytime I want to write garcon the proper way.

There is no standard method for recognizing unicode, even if the program that you're running _is_ unicode compliant. For webpages, you need to mess around with the encodings. Under windows, you need to set the default unicode encoding (and have to reboot). Even to this day, I can't figure out why ssh won't display the chinese for "insert mode" right, rather than pushing the whole screen up one line and messing up the whole display.
When you're doing the programming for an international audience, you're _supposed_ to use TCHAR, and _t("hello world") for strings. But of course, not everyone does this, and if you're getting weird errors from your program (yes, many programs have refused to run because of ascii/unicode) you gotta go through the whole program and figure out where someone forgot to use unicode strings/functions.
I always get freaked out from someone using another language to comment code or use as string literals. Who knows if the unicode will contain some encodings that if interpreted as ascii will give the program a really hard to find bug. (prolly not, but you never know if some text editor will give you some funky outputs)

I don't buy the idea that English is more concise, but at least it doesn't have multiple conjugations for verbs like in french/russian/spanish (for naming conventions for functions), or have same sounds for many different words like in chinese/japanese, or require dynamic display when typing in like korean/chinese/japanese. And if you're using a language for the naming convention, and to comment the code, it's good for everyone if you know that language well.

What business model? You mean not paying the writers, who are backed by schools, charging the subscribers ridiculous prices, which are mostly schools, and charging for internet access, where the primary customers are schools? Then they want the copyright as well? Ever wonder why college is so expensive?

Ya, I'd say that business model isn't outdated ... just sound a bit eh... greedy?

If I were a publisher, I'd want the smartest minds in the world to publish in my journal.

I'm sure people would want to read what the geniuses at MIT are doing, and the publishers will have to choose between losing subscribers or making the requirements more lax.

MIT's one of the few schools in the world that can pull off something like this. Most people choose schools because of rankings, and rankings are mostly based on the number of publications, so schools are not very likely to risk lowering their ranking for an ideology. But MIT doesn't care. No one would pay any college ranking that doesn't end with MIT or Caltech.

Not true, the compression technology today can do just fine.

If you use H264, only 4 Mbit/s is needed for a 1280x720 @ 30 fps with less than 10 ms encoding-decoding latency, which is perfectly fine for normal games. VDSL uses telephone lines and can (theoretically) go up to 90Mbps (but only in Asia, sorry!)

A company already used streaming technology to let you play crisis on an eeepc w/ wifi.

Like the previous comments, the big problem is latency, even if you have 0 lag for processing, encoding, and decoding, you need to go through the internet, and that's laggy regardless of the application.

The only way this can work is if they have multiple server farms at close to the local loop (at the isp's). It'll cost a bit of money for the company, but I think the business plan is there (esp if they go for full cloud computing, instead of just a gaming platform)

I've actually pitched a similar idea to Google ~ a year ago. Check out my blog if you're interested.

I would recommend that if darpa is serious about developing the first true AI that you start off something like angel funding for AI ideas, and get a bunch of different proposals that will be taken seriously. There are tons of people out there that have different ideas about true AI and how to do it. If you have any contact with AI academic world recently, you'll see a bunch of kinda interesting projects that have lost sight of true AI. Of course, outta the people who claim have solved AI, 49% are stupid, 49% are crazy, 99% are in for a big disappointment, and that last 1% are actually in the right direction. Kinda reminds you of proposals for startup companies huh? So support them like startups.

Angel funding have proven that they are important to promote creativity without wasting a fortune. Give developers (with potential) some money, NDA's, and most importantly, if they seem like smart people, an open ear to their ideas. Don't just do stuff like Grand Challenges that only research institutions/big companies can enter.

Well, that said, I'm actually very biased. Because I know how to solve AI =P. Of course, you prolly won't believe me, and I don't wanna share unless you sign an NDA. So, here's just a suggestion to think about.

AI is the process of solving problems, like chess -- solve the problem of winning the game. Solving an arbitrary problem is HARD, so let's ignore that for now. Suppose you have solved the problem, you'll have to carry it out somehow. What is it called when you are doing something in a specific way? An algorithm. Right now, MDP's use policies -- you're in this state, do this. Classical planning gives you a sequence of actions. But Algorithms? CS? Programming? Why aren't the results of these problem solvers turing complete? No wonder what comes out of AI programs today are very domain specific.

If you solved an arbitrary problem, you should be able to program it. AI can be called a program that makes programs. So here's the suggestion: start working on a programming language fit for AI -- the internal language that the AI brain will use. Programming languages today describe the process -- each program is compiled one specific way, and ran one specific way. Make a programming language that describes the algorithm, and only that. Whatever you program should give enough information about the problem you solve and nothing more. A rule of thumb is that if you can't use the language for a cook book, you need to work harder about eliminating the non-necessities. lchou1.blogspot.com