I haven't posted here in awhile. It has been brought to my attention that there are only 35 Google search results for "iota combinator". This needs to be corrected.
ι = λ x . x (λ f g y . f y (g y)) (λ y z . y)
That is the definition of iota in terms of lambda calculus. Iota is complete for lambda calculus. I would like to build a system on top of iota. Somebody already has. It's called lazy K.
I learned a lot of what I know about iota from Wikipedia and from Iota and Jot: the simplest languages?

sudo ln `which apropos` /usr/bin/how

A busy semester is drawing to a close. Senior year has been hectic so far.
I haven't been blogging much, but I have been reading RSS feeds and sharing things that follow a somewhat specific rule:
If I would want to feed an item into my currently unimplemented event system, I share it. That means my Google Calendar spams it quite a bit, as does a small fraction of my Facebook friends' status updates.
Anyway, my Google page links to ways for people to observe me on the web, and I update it whenever I get a new way. I'm unlikely to blog about the fact that I discovered Twitter over the summer, for instance, but I'll add a permanent link to it on my Google page.

My wiki's down because there's no power at the Manor.

Poor Grendel's had an accident.
So may you all.

Best line of the movie, and repeated more than once with the same inflection: "I'm Beowulf!"

testing google docs blog support
testing update

i am willing to argue the following point here and nowhere else:
all communication is conducted as transmission of shared symbols

23 is the most probable sum-of-digits for a 5-digit base-ten number, just as 7 is the most probable (exactly 50%) roll of 2d6
the smallest possible value of a 5-digit base-ten number is 10000, which adds to one
the greatest is 99999, which adds to 45
nothing adds to less than one and nothing adds to more than 45
there's probably a really nice proof somewhere that shows that the distribution is linear up and then down like it is with dice, but i don't know it
suffice it to say that a six-dimensional surface representing the sum function (dot product with vector whose dimensions are all one) over Z₁₀5 (the five-dimensional natural field) looks like a five-space just as 2d6 is a diagonal plane including the line from (1,1,2) to (6,6,12) and a perpendicular horizon of height seven from (1,6) to (6,1)
2d6: 1+6 = 7
1d9 + 4d10: 1+45 = 23 * 2
or something
so yeah, the number 23 is a statistical thing, and not at all anomalous

hey how would i go about writing a framework for programming emotional agents?
i guess i would need to start by considering the language in which one would write such agents and make an interpreter for that language
so what kind of model would this language need to describe?
i guess it would start with some dimensions of emotion, like the linearly independent dimensions of color (rgb), flavor (those flavor groups), etc; i guess that could be pheremones or something
but why hard-code them into the engine when they could be a data type?
so we need a pheremone atom and some sort of meaningful graph in which they travel
and a language for rules at the nodes of the graph
so we have a set of pheremones, a list of verteces, and a matrix of edges
what type should the edges be? is this just a neural net of real-valued edges or should the edges be objects?
better make the edges objects
so then what?
i still haven't determined any semantics for the node language, but now i know that at each timeslice, each node will act upon each of its edges with parameters of pheremonal levels
i guess that means the easy part is done

"children of workaholics [have the same problems] as children of alcoholics: anxiety and depression" ~the guy on the radio
my dad is always either too busy or too drunk to have deep, meaningful interaction, so what does that make me?
he doesn't care about how his actions affect others, but he considers himself a moral person
today he demonstrated that he doesn't care what others have to say, so he's not subject to feedback of any kind, so he'll never change unless something extrahuman puts him in his place and shows him what he's doing, since people can't do anything to do that

Three Wrong Ideas From Computer Science - Joel on Software:

There are three very major differences between accessing resources on another machine and accessing resources on the local machine:

1. Availability,
2. Latency, and
3. Reliability.

When you access another machine, there's a good chance that machine will not be available, or the network won't be available. And the speed of the network means that it's likely that the request will take a while: you might be running over a modem at 28.8kbps. Or the other machine might crash, or the network connection might go away while you are talking to the other machine (when the cat trips over the phone cord).

The answer, of course, is to do what X11 did: assume a network even when it's all on localhost.
You'll lose some of the benefit of the special case of local use, but the benefit is combining two concepts into one abstraction.
In Java, no file operations are considered safe: all I/O stuff can throw an IOException. Rather than wrapping networked file verbs with local interfaces to make it easier on the programmer, do it the other way around to make a more reliable system.

.CAC DF F8F.1E 47 5F 83E 9C BE C5 D1 E9.0137A 17D 2469EF 39 5038BE 60D 71246 85C 93 A07 B8 CE D9.26 33B 5369 60 7E 957AB.057CE 19E 39 406C 62.31 6369 7A 9B.F1
.06 24 5E 73C 869 97 AA BBD CE D2 E6 F0D.0E 52 C4 D27 E8 F3.11 23 46 7ADF 88 B6 DD E0.01 1A425 658 73 8B A3.3BC 4A 5C 6D 87F A0D C48 E14.C1 E1.78 98 AB CC EC

2F 9D.48 7B 9A BE C4C F9.05 1D 48B 5E 83 91.002 A1B FC.01 4F 8C 936 A8D B06 D3.11 2CE 5C 7D 915 A9F B1 D2.04 26 446 50 66 7B 82 938 C068 D15C E59CE F2.1F4B 58 76 81 CE
2E 3D B59 C7 D3.18 29 32 5A 8C A8 BF C9 DC FB.0F 67 A2 BD E1.0F 1CD 32 78 91.09 2D 3B 66A 77F 84 C3.18 32 56 76 81 B5 C7 D5 F1.09 10EF 32 7A 92 C36 D257 F3CF.23 359 4237 547 725B A5 B1 C7 D3


12 22 42 60 72 8B B1 F4.00A 2C 43 69CF BA CE DC.20 6B DB F1..17 AD B3 EDE.22 34 535 70 C4 E0 F3.13 2D 33F 45E 55F 61 76 9C C2 F3.10 23 99 B0 FA.42 83 92 E0
6A.3F CB D1.38 E0.6F 7C 99 AC D8B.CC 5E 6B 768 95ABDF BBD CAB DE EC F8.10 3DE 402 5F 7F 81CEF A0E CD.0C 113 2245 30 43 61 7124 8AF 98 A0 B2 C6 E8.32 40B 65 70 87 90 C0 E0 F3

.10 58 66A 96 A4 B9 CA.71 92 C03.15D 64.14..FF.46 83 B0
9F AF.22 44 52 80 CC.04 10 43 5A 69E 78 84 D6.4A 7D.CA..1D 69 77 9E AB BA C9 DAC.40



.2BD

some men claim to be intimate with the supernatural without the aid of any one particular supernatural agent telling them what is going on
of these i ask, how well do you even know the hole in your head?
i'm speaking, of course, of sinuses
how well does your mental model of the inner workings of a physical three-dimensional cavity lined with nerves match it itself?
how, then, can you expect to make any claims from observation about anything you can't observe both in its entirety and from outside?

the activation record doesn't need to go in the stack!
data structures are graphs
scope is a data structure
connecting graphs together makes bigger graphs
running programs are graphs
a stack is a list
a linked list is a graph
the activation record does need to go in a stack
but that stack can be a doubly-linked list instead of an array
i think it would be kind of cool if the scope was something more explicit
also if data flow were a graph
but then we're looking at labview, which is fine too

sorry for the ramblings, i'm not very awake

things i've begun to grok this summer

Quotes about Operating Systems:

In pioneer days they used oxen for heavy pulling, and when one ox couldn't budge a log, they didn't try to grow a larger ox. We shouldn't be trying for bigger computers, but for more systems of computers. (Rear Admiral Grace Hopper)

What would you rather have to plow a field — two strong oxen or 1,024 chickens? (Seymour Cray)

so, i found something interesting in my econ book and i wanted to know if it was always true

notation
let [a,b] = the closed interval from a to b
let (integral r) phi = the integral over the interval r of the function phi
let d = lim (epsilon->0+) delta

given a function f over R+
let g(x) = ( (integral [zero, x]) f(t) * dt ) / x

prove g(x) = f(x) => g'(x) = 0

proof
g(x) = ((integral [0,x]) f(t)dt)/x
xg(x) = ((integral [0,x]) f(t)dt)
xf(x) = (integral [0,x]) f(t)dt
d(xf(x))/dx = d((integral [0,x]) f(t)dt)/dx
(dx/dx)(f(x)) + (x)(d(f(x))/dx) = f(x)
f(x) + xf'(x) = f(x)
xf'(x) = 0
x = 0 or f'(x) = 0
x = 0 or g'(x) = 0

and, of course, since
lim (x -> 0+) g(x) = lim(x->0+) (integral [0,x])f(t)dt/x
= 0/0 therefore l'hopital
= (d/dx (integral [0,x])f(t)dt) / (d/dx x)
= f(0) / 1
= f(0)
so g(0) = f(0), or at least its limit does


that's all well and good, but it doesn't help me understand why marginal production crosses average production at the latter's maximum

the anonymous biography is a privacy policy concept that should be used whenever invasive, Google-like records need to be kept to enhance the user's experience via intimacy with the application
to implement the anonymous biography, keep data about the user in two separate databases: one for data that was given in confidence that can never ever ever ever be shared with any third party in any way that can link it to the data stored in the other data base, and one for those "anonymous usage statistics"
the distinction, of course, is transparent to the user, but when it comes time to decide the hazy issue of who to share information with, the separation of confidential data from the anonymous biography (or anonymous usage statistics) will make it easier for policymakers to create laws that treat confidentiality the same between users and machines as it is treated between people and their therapists, clergymen, etc.
just a thought