The Public-Reason Sphere in Mid-2026: Fighting LinkRot Edition
A linkrotted weblog post from 21 years ago: “Alamedia of ‘Unfogged’ Is Puzzled by Investment Banker Compensation Levels”, and even though #typepad-resurrection is one of my to-do #project #hashtags these days, it has not yet happened. But there is the Wayback Machine at <http://archive.org>. And there is (for now at least, on an experimental basis) my personal information butler SubTuringBradBot at <[web.telegram.org/k/](https://web.telegram.org/k/#@SubTuringBradBot)>…
Well, Joe, if you were on BlueSky like a human rather than grunting in the toxic mud like an animal, I would have seen and responded to this in an hour!
As it is, I only see it when SubTuringBradBot, my information butler, brings it to me on his virtual silver tray at:
And so you are only getting a response the following day—and a response I will now outsource to the brand-new version of my information butler, SubTuringBradBot:
In what context did Brad DeLong write something like: "There are four relevant human capabilities here: mastering details, following through, working like a dog, and sizing up people”?
The quote comes from a May 16, 2005 blog post on your TypePad weblog, in a piece discussing the "Mysteries of Chet" — your analysis of why a certain type of finance/business operator commands high compensation despite not necessarily being the sharpest intellectually. The full passage runs roughly:
"A bigger part of this answer is that there are four relevant human capabilities here: the ability to master details, the ability to quickly grasp what the salient issues are and follow them through to their conclusion, the ability to work like a dog, and the ability to size up people — figure out quickly who will actually produce something useful and who will not..."
The argument was that world-class command of even two of those four is rare, which explains the premium.
can you give me the original URL, and a working wayback machine URL please?
That is SubTuringBradBot, at <[web.telegram.org/k/](https://web.telegram.org/k/#@SubTuringBradBot)>, freely world-queryable until it gets too expensive for my Anthropic budget, or (if I do succeed next week in moving it to my under-the-dining-table machine) contention for RAM, GPU, & NPU access on my machine becomes substantial enough to annoy me, given what else I am using the machine for.
What do I think of it? I think catechism-based RAG search with a relatively light-touch LLM front end to provide a natural-language interface is now scarily good. I think trying to tune the m*****f***** via vibecoding annoyingly frustrating. It is rife with hallucinations of all kinds. It does not "just work", or "do what I mean". It has taken 4.5 years, off and on, to get here, after all.
If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…
