Saturday, October 11th, 2008

Time	Event
2:20a	I'm not looking for a bytecode VM language with great performance - my performance will be in how I use the language, not in what language I pick. Something slow is okay! I need simple-elementary. I'm not porting Windows NT to it. For that matter, I'm not porting anything to it. I want a VM bytecode that is a little like RISC, a little like LISP. Something academic, something Wolfram might come up with. Something Turing-complete, but unable to natively parse XML Unicode. Is Xen what I'm looking for? I'm trying to understand what Xen is but I doubt it. about 9,980,000 for "virtual machine". about 4,260,000 for "virtual machine" -java. about 2,880,000 for "virtual machine" -java -vmware -"virtual pc". about 28,400 for "virtual machine" -java -vmware -"virtual pc" -x86 -pvm bytecode. Ah! That's more like it. I'm told there are two VMs that may be what I'm looking for. They're called Neko and Parrot. Python and Xen might help, too, but I've checked out Python's bytecode and it's not really what I want. To answer the question, all you must do is ask it correctly. I love LISP but... I need a bytecode for this one. (Comment on this)
3:17p	Understanding Sarah Palin: Or, God Is In The Wattles It can't have escaped notice that there are a lot of people engaging in sloppy thinking. Now it's well-known that the brain is a survival engine that lies *all-the-time* if it thinks that said lie will improve your chances of having fertile grandchildren, but here's the question of the day and it's one that Hitchens, Dawkins, and Harris have Completely Dropped the Ball in not asking: Why did superstition evolve? Well, there's some unspoken assumptions in that question that will dissatisfy some people. Let's re-word. Why is it that superstitious people haven't all killed themselves off yet? It's not for lack of dumb. These people have a significant and measurably higher rate of blowing themselves to kingdom come than rational types. So there's got to be a trade-off, some compensation, some mitigating factor that rewards them. http://rifters.com/real/2008/10/understanding-sarah-palin-or-god-is-in.html -- Understanding Sarah Palin: Or, God Is In The Wattles (Peter Watts) He puts forth the notion (and supports it well) that when you see someone behaving in an amazingly retarded way, that you can conclude pretty quickly that they are genuinely not faking it. (A "wattle" is metaphorically a way for one creature to learn some true things about another.) So this extra factor that rewards people for their superstitions... could it be that people trust them better? Being stupid is a people skill? We will find out, and we will know for sure very soon. (Comment on this)
4:04p	...and then I found out that this is an old idea The handicap principle is a hypothesis originally proposed in 1975 by biologist Amotz Zahavi[1][2][3] to explain how evolution may lead to "honest" or reliable signaling between animals who have an obvious motivation to bluff or deceive each other. The handicap principle suggests that reliable signals must be costly to the signaller, costing the signaller in the trait being signalled in a manner that an individual with less of that trait could not afford. For example, in the case of sexual selection, the theory suggests that animals of greater biological fitness signal this status through handicapping behaviour or morphology that effectively lowers this quality. The central idea is that sexually selected traits function like conspicuous consumption, signalling the ability to afford to squander a resource simply by squandering it. Receivers know that the signal indicates quality because inferior quality signallers cannot afford to produce such wastefully extravagant signals. http://en.wikipedia.org/wiki/Handicap_principle (Comment on this)
6:11p	I'm going to write a program and it will contain no bugs. f(false,false):=return(false) f(true,false):=return(false) f(false,true):=return(false) f(true,true):=return(true) I think I will call it... "and". Look, maybe it is impossible to debug a 100 megabyte program. Fair enough. I've never wanted to write one. Point being, if you make the problem small enough, simple enough, well defined enough, Perfection is Possible. (Comment on this)
7:11p	Ah! There it is. Programming | +- Imperative | | | +- Procedural (Subroutine) | | | | | +- Structured (Goto considered harmful) | | | | | +- Object Oriented (Encapsulation) | | | +- BrainF*ck | +- Declarative | +- Logic (expert systems?) | +- Functional (no side effects) 8 commands, each one byte long. Perfect. Well, to be exact, it's horrific. BrainF*ck was designed as a joke. Its resemblance to a Universal Turing Machine is more of a flaw than a feature. Nobody would write a research project in this language. For that matter, I can't type BrainF*ck source code directly in LJ because < and > make up 1/4 of what I'd be typing. I'll need a pretty-printer, which I can write in Sed, and a code tester, which I can write in Python, and I'll want to compress my source code with gzip... This could work. What is this emotion? Could it be... fear? (Comment on this)
9:33p	The cool thing about language is that the same code can be run in all kinds of different contexts. I can add apples to apples, and oranges to oranges... and yet despite the fundamental differences between adding apples and adding oranges, we discovered enough parallels between the two to *generalize*, to abstract. To say... maybe oranges and apples and bottlecaps have some property in common - a thing that behaves a certain way under addition. A thing that eventually we decided to call "number". 2 apples + 3 apples = 5 apples 2 oranges + 3 oranges = 5 oranges 2 bottlecaps + 3 bottlecaps = 5 bottlecaps 2 + 3 = 5 We realized that there was a shortcut. So we used it. AI must be able to do that. (Comment on this)
11:36p	"This is called an oubliette. That's French for a place of forgetting." "If the physical representation of the bit is part of a thermalized distribution (so that thermodynamics does apply), then this bit erasure necessarily is associated with the dissipation of energy in a heat current of kTlog2 per bit. This is due solely to the act of erasure of a distinguishable bit, and is independent of any other internal energy and entropy associated with the physical representation of the bit. Bennett went further and showed that it is possible to compute reversibly with arbitrarily little dissipation per operation, and that the apparent paradox of Maxwell’s Demon can be explained by the erasure in the Demon’s brain of previous measurements of the state of the gas" -- MIT - Signal entropy and the thermodynamics of computation by N. Gershenfeld Boltzmann constant = 1.3806503 × 10^-23 m^2 kg s^-2 K^-1 body temperature = 37 degrees Celsius NIST's STP: 293.15 K = 68 °F = 20 °C 101.325 kPa = 14.696 psi Idle adult human output about 100 watts at 310 Kelvin, with a kTlog2 of 3*10^-21 Joules (the cost of erasing one bit). So from the perspective of information theory, the human body has a bandwidth of about 3*10^22 bits per second. 30 /trillion/ gigabits per second. (Upper limit, of course) I should mention that most of that bandwidth is spent hashing out the finer points of which beneficial bacteria get to rule every little fold of your small intestine and is unimportant to the human condition. No doubt, at least 99% is worthless. Maybe the next 99%, too. I'm sure a low-res jpegged version of me would have just as much fun. (Comment on this)

<< Previous Day

2008/10/11 [Calendar]

Next Day >>