Benki β†’ All Posts

β‡  previous page next page β‡’

Haskell library for defining values recursively while avoiding nontermination. Magic.

Imports a variety of build systems into Nix.

Currently supports Rust, Haskell, PHP, Node.js, and Python.

Matthias #

Rent controls

The problem

Rent controls sound nice at first. High housing costs hit poor people especially hard. Funding the building of cheap public housing, which would alleviate exploding rents, doesn’t work long-term because in order to stay funded it needs to make a profit and so can’t stay cheap for long. So instead you just go and make a law that caps housing rents. Problem solved.

Except it isn’t. High rents are the result of scarcity. By capping rents:

  • You disincentivize the building of new housing by restricting profits, which further restricts supply in the long run.
  • You subsidize high earners who live in apartments that are much too large for them, restricting supply even in the short term.

This latter part is a particularly painful realization when (as I do) you identify as relatively pro redistribution. It would be much preferable to subsidize low earners in a more targeted way.

Alternative: direct subsidies

Instead of restricting supply and subsidizing everyone indiscriminately, the solution, then, is to deregulate supply and pay direct subsidies to low earners to enable them to pay the higher rent.

Benefits:

  • This increases profits for people who build new apartments, which is what you want.
  • It nudges people living in apartments that are too large for them to move to smaller apartments.
  • By freeing larger apartments and enabling a more targeted subsidy, it benefits poor families, who can now move to apartments that are big enough for them.

Raising the money needed through Georgism

The obvious question the direct subsidy approach raises is where to get the money from.

The answer is a land value tax: tax the land (not the buildings built on it!) a fair amount to match its profit-making potential.

Here is why it is a good idea:

  • It incentivizes people to build more apartments onto a given parcel of land.
  • It strongly disincentivizes single-family home parcels, which are extremely inefficient and simply have no place in an expensive city these days.
  • It keeps the good incentives (incentive to build more, incentive to move to smaller apartments) in place while moving money from land owners to renters in need of support.

Challenges

The approach outlined here faces a few challenges:

  • It is not obvious how to calculate the value of a parcel of land.
  • Land owners, who have a powerful lobby, will not be pleased by the changes.

I consider neither of these a deal breaker, but they will have to be dealt with. The land value estimation problem is real, but it does not need to be solved in full generality, since some inaccuracy can be tolerated as long as the profit making potential is still positive (i.e. it forces the state to err on the side of a too low tax rather than a too high tax). The lobbying point must be dealt with through politics as usual. A first step could be an education campaign, which I am hoping to contribute to with this article.

On the overapplication of intuitionism.

It rings true to me that classical logic is philosophically underappreciated. In most of mathematics, constructivism does not, I feel, carry its weight. On the other hand computation is intrinsically tied to intuitionistic logic, so it is important to know when one needs to apply it.

An innovative proprietary (possibly patented) logic that can prove your propositions extraordinarily efficiently.

A Linux kernel security module that enables applications to limit their own privileges.

An optimizing back end and intermediate representation for compilers of functional languages that are based on graph reduction (such as Haskell, Agda, or Idris).

A compiler suite for C, Modula-2, and other languages targeting 8- and 16-bit CPUs.

In addition to many low-end processors, also targets the VideoCore IV, the VPU on the original Raspberry Pi.

A DOS object linker that makes use of DPMI to be faster than LINK.EXE and handle larger binaries.

β‡  previous page next page β‡’