Stablecoins of America

•  1. Once upon a time, everyone trusted the USA.
• 2. The USA is spooking smart-money, and while smart-money isn't dumping all their US assets, they're definitely dumping some in order to diversify out of USTs.
• 3. The Trump administration's stablecoins-of-America policy is simply a device to harvest dumb-money liquidity so that smart-money can exit first.
• 4. This doesn't mean that crypto is going to crash right away - in fact, it can't, otherwise the smart-money can't exit completely.
• 5. Play the players, and get your money off the table, while you can.

Learning how to talk to machines is difficult.

I studied this very slightly as a teenager, then in college I paused to study the history of ideas, followed by a study of commerce in the developing world for another 20 years. Over the past two years I've managed to find some time to survey the stack, and it is discouraging, but I think I will be able to manage, if I do not fumble my finances. Hah, that is always the biggest risk.

There are four legs to the stool.

1.

A bit of machines memory requires 1 "D flip-flop" : there are 4 gates in a DFF, and each gate needs like 4 transistors & change. For example, an Apple M4 has 28 billion transistors, and the manufacturing process takes like 4 months - fortunately you can make a few hundred at a time, but there will be like 16% waste. These things are arranged in hierarchies of control, like an army of factory bots passing electrons around 4 billion times a second, with many layers of prep, parallelisation, and quality control, on the size of a postage stamp. 

2.

Sending instructions to these bots, is what software programmers do. Instructions written in near-human languages are translated a few times, until the machine understands it perfectly. Most of your friends who are programmers give zero fucks about how this gets done - they leave translation to bots. A programmer is also a bot, albeit a slightly creative one. A programmer decides, from toolkits known as logic, mathematics, statistics, and computer science, how to write programs to get thing done. You've probably noticed that many programmers are quite focused here, and really don't want to know about the other three legs of the stool - and that's totally ok, because some people just need to focus, otherwise nothing gets done.

3.

Telling the programmers what needs to actually get done are a bunch of business and public policy managers, pushing various political agendas : shareholders need profits, customers want entertainment, and someone needs to worry about public health, scientific progress, weapons manufacturing, and all that other stuff.

4.

All politics is rooted in economics, and all economics is rooted in states of mind. Most people you'll meet are in possession of a mind, some are in possession of more than one. Almost none bother to quantify this - most people don't think about quantifying their perceptions of their own consciousness - many think about it, but don't think it's possible. Boy oh boy, this is a whole other issue of it's own. But this post is about machines, not phenomenology.

The ultimate goal is opaque.

All of this proceeds through time, under the bumbling tyranny of fates unknown - again, most people give zero fucks about that, as they are more concerned with the locality of their puny little bodies, hairspray, perfume, family, friends, i.e. lives.

And being aware of this, a civilised person must conduct themselves with some modicum of moderation, i.e. reason ( rootword : ratio ).

I rarely feel up to the task, as I am generally quite far behind my peers. But, I try to comfort myself with the thought that we all have different paths, and we each take different amounts of time to get things done, based on what other things we have to do which others do not.

Such is life ...

Modelling the World : Phenomenology

It is often said we think in 3D - that is partially true. Most of our daily thoughts are in less than 3D.

• - Hearing happens in 1+ dimensions
• - Smell happens in 1+ dimensions
• - Sight happens in 2+ dimensions
• - Haptics happen in 2+ dimensions
• - Kinesthesia is a loop, over the datasets above, and hypotheses about a 4+ dimensional model of object permanence that we have no direct sensory perception of

The brain is a fun thing to play with. Most people do not properly understand it, because phenomenology is not formally taught in schools.

"XX% of my code is written by AI."

"XX% of my code is written by AI."

Folks, nearly all machine code has been written by bots for the last 50 years already. The stuff your programmer writes is some juvey dialect that gets translated two times by other bots before it reaches the CPU. 

The problem with any work delegated, is that you need a way to audit it. It doesn't matter if you delegate it to a bot or a human. Most humans are dumb shits anyway, and need to be checked regularly.

The quality of the design , which permits governance of code, is not mentioned in the quote above. It's the equivalent of saying 

"XX% of my conpany is run by the new guy."

No one cares if the guy is new or old. People need to know if the guy is making you money, or losing it.

Multiple-choice Questions : on tests : done right

 Multiple-choice Questions : on tests : should have variable costs based on the importance of the question.

• 1. Every question should have a "I don't know / abstinence" option.
• 2. Every question should negative points, if the answer is wrong.
• 3. The ratio of negative to positive points should be obvious to the test taker; for simplicity, every correct answer should have +1 points, but wrong answers may be -0.5, -1, -2, -4, etc.

Counting Machines, Language, and Anthropomorphism

I have some pre-hypotheses about literacy across the period from 1900 to 2100.

Somewhere around 1950, counting machines began to understand programming languages that were anthropomorphic, and closer to natural language. Yet computer languages remained formal, even though they were more human.

As we approach 2050, computers are getting better at understanding natural languages. I'm curious about what this means for the human population that had shifted their verbal resources towards formal languages.

Will these people remain lower in the stack, speaking to machines more as they are? Or will these people revert to distributing more of their language style across natural language texts directed at humans?