2014-01-11

Customer Service

Work never ceases to amuse me. Some people care about their customers. Others merely engage in commercial interactions where making someone feel at home is a matter of pressing all their social buttons - feeding them, smiling with them, smysing with them, counselling them, removing their guards, breaking down their paranoia, giving them something to believe in... all for a few bucks, and change. Change - what a loaded morpheme. In any event, I've often said that the only pets I keep are human. I guess I genuinely like people in the same way that some of my colleagues like kittens. :)
Services delivered. No further value obtainable today. Need to be stricter about clocking out on time. Unit returning to base, hello traffic. Ready to launder.

2014-01-10

Proofs

ProofWiki: good stuff here - finally having time to appreciate it, now that I basically know how to deploy an equivalent cloud service. All this needs is an automated theorem prover feature... maybe it's there, but I don't see it.

Anyway, all in line with further studies in demonstrating why people are boring.
Demonstrate a machine that can beat humans in an argument.

Step one... learn how to encode human arguments in a machine-readable language.
Well, at least as of today, I've managed to link up my understandings of LaTeX, and Haskell, via mathematical proofs and the functionality of Agda.
What did you think... of course I'm not interested in math for math's sake... pooh pooh...
Also of interest, some other stuff in automated theorem solving, automated theorem checking, and proof assistance:

Mizar system
Metamath

Annoying as it seems (either my own ignorance of the counterexample, or the fact that) there is no standard, like an ISO, for computer readable mathematical proofs.

HOL (subject: higher-order logic) also looks like it needs serious looking into - for my education, on how this stuff is currently treated by the world at large. Ah, time is always short. (A family member of HOL being Isabelle, and Haskabelle.Haskabelle by the HOL-ML-Haskell project group.)

I already kinda know Haskell - but OCaML seems to have been more closely associated with this sort of software in the past. Also checking out HaskHOL. Much to learn.

Proof Assistants - history, ideas, and future

You Want... What?

The choice of a management toolkit is completely dependent on how you prioritise end-goals, such as:

a) making a monetary profit
b) having fun
c) learning something
d) building a brand
e) doing it within the law
f) helping people other than yourself
g) creating a quality product
h) making people happy

Any are possible, but none are necessary.
Right. Done helping all the people I can think of for now. Better stop distracting myself with social work, and focus on more important stuff.

7.3 hours to first alarm. Let's study some math...

Questions Answerable by (or with) a Coffee Roaster

Going to over-reach a bit and describe an espresso roasted single-origin - and guess about why it is the way it is. Feel free to comment if anything rings a bell.
  1. Moisture content: unmeasured; bean appears to be on the drier side, as clumping is minimal and grinds fly out of the chute like loose sand.
  2. Hardness: unmeasured; bean appears to be of moderate hardness, as it hardly channels with medium-pressure tamping and a moderate (4-5s) preinfusion.

    (Supporting 1) and 2), the bean crunches easily if you chew it.)
  3. Baking: no apparent baking in the roast, as there is an absence of grainy / malty flavours, and very little astringency of any sort however...
  4. Scorching: no apparent scorching as there's very little bitter stuff to complain about.

Now, I am wondering the following phenomena:

  • while the overall flavour profile is floral and pleasant at low-temperature extractions (90.5C) , extraction yield appears to be low: lower than in other recent roasts of the same origin (evident from taste, and crema yield);
  • if we pump up extraction temperature to 92C, we get a rapid increase in sour, umami, and browned flavours, but not a whole lot more crema - it is near impossible to serve this temperature's brew as a black coffee, to anyone who dislikes the slightest acidity in their coffee (see (4 above) - we'd normally pull a 23s ristretto, but even at 40s lungo, acidity remains unmasked by whatever else can be extracted from the bean;
Anyway, I have zero experience with roasting coffee, and was just wondering if this could be due to some sort of roast under-development between first and second crack.

In light/medium roasting, is it possible that you can achieve a favourable flavour profile, at the cost of reducing the quantity of extractable flavonoids and oils in the bean?

Thoughts?

2014-01-08

Unions & Intersections

Some of the more interesting exercises (a purely subjective qualification). Easy concepts, difficult language. Boo.

Theorem:
The following are equivalent:
- the union of C and (the union of the B-alphas, over the alphas in lambda,)
- the union of (the union of C and the B-alpha), over the alphas in lambda.
Proof:
\begin{align*}


p(x, C, \{B_\alpha\}_{\alpha\in\Lambda})& = \underbrace{x \in
\left\{C \cup
\left[\bigcup_{\alpha\in\Lambda}B_\alpha\right]
\right\}}\\[1em]

& = \underbrace{(x \in C)} \vee
\underbrace{\left( x \in
\left[\bigcup_{\alpha\in\Lambda}B_\alpha\right]
\right)}\\[1em]

& = \underbrace{
(x \in C)} \vee \underbrace{(\exists(B_\alpha), x\in B_\alpha)
}\\[1em]

& =
\underbrace{
\exists(C\cup B_\alpha), x \in (C\cup B_\alpha)
}\\[1em]

&=\underbrace{x\in \left\{\bigcup_{\alpha\in\Lambda}(C\cup B_\alpha)\right\}}\\[1em]

\end{align*}
Less than seven hours till first alarm. I really should get out of the office.

Theorem:
The following are equivalent:
- the intersection of C and (the intersection of the B-alphas, over the alphas in lambda,)
- the intersection of (the intersection of C and the B-alpha), over the alphas in lambda.
Proof:
\begin{align*}

p(x, C, \{B_\alpha\}_{\alpha\in\Lambda})& = \underbrace{x \in
\left\{C \cap
\left[\bigcap_{\alpha\in\Lambda}B_\alpha\right]
\right\}}\\[1em]

& = \underbrace{(x \in C)} \wedge
\underbrace{\left( x \in
\left[\bigcap_{\alpha\in\Lambda}B_\alpha\right]
\right)}\\[1em]

& = \underbrace{
(x \in C)} \wedge \underbrace{(\forall(B_\alpha), x\in B_\alpha)
}\\[1em]

& =
\underbrace{
\forall(C\cap B_\alpha), x \in (C\cap B_\alpha)
}\\[1em]

&=\underbrace{x\in \left\{\bigcap_{\alpha\in\Lambda}(C\cap B_\alpha)\right\}}\\[1em]

\end{align*}


De Morgan's law:

Theorem:
Where all \((A_\alpha)\)s are subsets of U, the following are equivalent:
- the complement in U of (the union over the alphas in lambda, of the A-alphas)
- the intersection over the alphas in lambda of (the complements in U, of the A-alphas)
Proof:
\begin{align*}

p(x, U, \{A_\alpha\}_{\alpha\in\Lambda})& = x\in\left[\bigcup_{\alpha\in\Lambda} A_\alpha\right]^{\mathsf{\complement}}_{U}\\[1em]

&=(x\in U)\wedge
(\underbrace{\forall A_\alpha, A_\alpha \in U)}\wedge \underbrace{\neg(\exists A_\alpha, x\in A_\alpha)}
\\[1em]

&=(x\in U)\wedge
(\underbrace{\forall A_\alpha, A_\alpha \in U)}\wedge \underbrace{(\forall A_\alpha, x\in (A_\alpha)^{\complement}_U)}
\\[1em]

&= x\in
\left[\bigcap_{\alpha\in\Lambda} (A_\alpha)^{\mathsf{\complement}}_{U}\right]\\[1em]


\end{align*}
Theorem:
Where all \((A_\alpha)\)s are subsets of U, the following are equivalent:
- the complement in U of (the intersection over the alphas-in-lambda, of the A-alphas)
- the union over the alphas-in-lambda of (the complements in U, of the A-alphas)

Proof:
\begin{align*}

p(x, U, \{A_\alpha\}_{\alpha\in\Lambda})& = x\in\left[\bigcap_{\alpha\in\Lambda} A_\alpha\right]^{\mathsf{\complement}}_{U}\\[1em]

&=(x\in U)\wedge
(\underbrace{\forall A_\alpha, A_\alpha \in U)}\wedge \underbrace{\neg(\forall A_\alpha, x\in A_\alpha)}
\\[1em]

&=(x\in U)\wedge
(\underbrace{\forall A_\alpha, A_\alpha \in U)}\wedge \underbrace{(\exists A_\alpha, x\in (A_\alpha)^{\complement}_U)}
\\[1em]

&= x\in
\left[\bigcup_{\alpha\in\Lambda} (A_\alpha)^{\mathsf{\complement}}_{U}\right]\\[1em]


\end{align*}

Theorem:
Proof:

Brand Bitchin

At some point in every brand's development - you get to decide if you want to be TI or Apple. Don't complain if you pick the former and sell truckloads of goods but no one knows your name.

If you want to be Supreme, you can't be UNIQLO, or ARMANI, and you certainly can't be a rockstar on Project Runway - it's a different positioning, and you have to deploy completely different strategies to get there

Gossip

DVDA jokes? I guess no one's in a hurry to get rich around here. Hahaha. Moral of the story: don't piss off suppliers.

2014-01-06

Record Inefficiencies

11% of today spent on the road. So far, it's worth it. We've hit the big-time, baby...

Monday: 0700 out the gate - 0840 arrived - 100 minutes
Tuesday: 0640 out the gate - 0755 arrived - 75 minutes

Wednesday: cheat day
Thursday: MAJOR FAIL

Friday: 0605 out the gate - 0655 arrived - 50 minutes
(We're approaching the limit of 25 minutes!)
An hour late to bed. Got records to break tomorrow, if I can just get up on time... it's my off-day, but who'd want to break stride during training

Slacked off on cheat day.
My availability sucks: it's only gotten more constricted since it cost me the last relationship, haha. But one lives with one's choices. This is barely the beginning of busy.

Thu Jan-23. 0620 out. Jam starts 2.2km, odometer dies. 0735 arrived. 75 minutes. 
Fri Jan-24. 0605 out. Jam starts 3.2km, no odometer. 0645 arrived. 40 minutes.
100 minutes.
75 minutes.
50 minutes.
40 minutes. (today)

My $15 in daily sales was a lot cheaper to deliver today. We are officially cost-hacking this operation. Now rinse and repeat, and then maybe soon enough we can worry about growing sales.

First things first. Feed. Heading to my locker for some soylent right now. Meanwhile, keeping a close watch on the output from my autonomic nervous system. Running on three-ish hours of sleep today. :P

News reading. Barely enough time to pay the bills and stay in touch with the world at large. Hardly any time for science. :(

But today is just one of the slower days. There have been many faster ones, and there will be yet faster ones to come.

8am. Pack up readings. Time to feed (second time today). Caffeine was useful for a quick reading sprint through every syllable of a few paragraphs from the Wiki on the Archimedean property. Now, to parlour tricks and janitorial duties for the next ten hours - so far, still the perfect day job for staying healthy.