Weekly Roundup 001 - Luke Davis

all in pod
tech should just work
the perfect game
matt pocock on ai agents
link dump

This is the first edition of a weekly (?) roundup of interesting links, tweets, youtube videos, articles and really any other media I find. I tend to send lots of links to friends, family and coworkers, so I figured why not make a big roundup of all of those in one place. In no particular order:

“no” ⟶ “yes”

I wrote a blog inspired by this tweet from Adam: Spreadsheet to Web App with Claude Code

Loading tweet...

all in pod

on party of government (49:15)

David Sachs:

… the party of government doesn’t really care where the money goes. It’s all just patronage to them. Whether it’s to Somali daycares or whether it’s to this high-speed rail line that doesn’t have one piece of track.

Look, these politicians don’t care how the money is spent at all. All they care about is rewarding their supporters. That’s the whole game. And then those supporters round trip a portion of the ill-gotten gains they’ve gotten from the government back to those politicians. That’s the whole game.

on CA’s wealth tax (1:10:05):

normalizing the seizure of private property
tax on work vs tax on individual private property that hasn’t been realized

David Friedberg:

We are talking about for the first time a tax on individual people’s private property not on your income. […] the way taxes work today is when you earn income either via labor so you collect a paycheck or via the sale of an asset. […] So for the first time ever we are talking about placing a tax on people’s private property that hasn’t turned into cash.

Takeaway #1: This is a new tax system. Not simply “taxing the rich” as the slogans imply.

In France in 1988, 1989, they passed this millionaire tax. Anyone with 1.3 million euro or greater net worth had to pay a tax of like 1% a year, 1 and a half% a year of their total assets. And it raised 4 to 5 billion euro a year in new tax revenue. But what happened? €200 billion of net worth left France and the total income tax went down by 8 billion.

Takeaway #2: this kind of tax causes net worth to leave the state to avoid future taxes, hurting the tax base and the local economies.

In the state of California, these billionaires, there’s 200 of them. The total net worth of billionaires in the state of California is $2 trillion.

There’s about 900 billionaires in the United States. The total net worth of billionaires in the United States, $8 trillion. So if you took, let’s say 1% a year, every year you take 1% of the billionaire’s worth, that’s $80 billion of income for the federal government every year. That will pay 20 days of interest on our debt. 20 days.

Takeaway #3: this kind of tax is a tiny fraction of what’s actually needed.

The real money is in the middle class. The net worth of the United States, the middle class, and everyone else is 170 trillion compared to 8 trillion of the billionaire. They need a way to open the door so that they can go after the real honeypot. The real honey. So the real goal of this, just so everyone understands, the real goal of this is not to tax billionaires because there are other ways to tax billionaires. The real goal of this is to create for the first time in American history a private property asset seizure tax and to use that as a way to take a percentage because they’re going after the 170 trillion.

Takeaway #4: If the goal were purely revenue, you’d tax the middle class because that’s where the money is ($170T vs $8T). It seems more likely that this is a play at normalizing this kind of tax so it can be applied en masse, not just to billionaires.

Takeaway #5: The real concern is precedent. The US legal and political system runs on precedent; once a wealth tax exists for billionaires, expanding it becomes a matter of adjusting thresholds, not breaking new ground. This is the kind of long-term structural thinking that the under-25 voting bloc doesn’t seem to be aware of at the moment. “tax the rich” is a simpler, and let’s be honest much easier, rallying cry than working through these implications.

tech should just work

My wife has learned that when I start verbalizing my frustrations with tech not working as expected, it’s a “nod and pretend to care” moment. Too many times I’ve ranted about tech just working. I spend so much time configuring and weighing options that when something stops working, alarm bells go off in my head. Just work GitHub. Damn.

Loading tweet...

the perfect game

Factorio is a perfect game. This is a well-known fact. Karpathy makes some really good points; being effective in AI prompting is more about being able to operate in realtime while keeping your head and eyes at 30,000 feet. In Factorio, you start the game with basically nothing but your ability to gather resources.

Gathering resources is boring and tedious, but necessary. Once you have enough resources, you can start building tools and factories. This is the key. Once you have factories and the means to automate, the game becomes less of a resource game and more of a logistics and planning game.

Loading tweet...

With AI, the same is true. You start with basically nothing but your ability to prompt. Prompting is boring and tedious, but necessary to build tools like custom agents, skills, slash commands, memory and more. With these building blocks in place, you can start building factories that will allow your focus to shift from basic mechanics to higher-level concerns and concepts.