mncharity

> no real reason [...] still coated in whatever was floating around the environment

I wonder, might one instead make an argument that it's not currently economic-technologically feasible to avoid artifact surface information destruction and contamination, at any stage of the dig-to-archival-conservation pipeline, nor to extract whatever information might be there, outside of very rare cases? Rather than there's no information there to lose?

Perhaps it becomes possible-and-affordable at some point, to frequently check, say, "oh neat, here are some 10 um notches on the surface, with some nice yeast, and some pollen, and oh, maybe that's a bit of calcified biofilm for nanopore DNA?"? Does automated lab tech progress, someday make "brushed, handled, dusted, washed, and shelf in the basement" start sounding like 19thC "we accessed the site using dynamite"?

Ah. I was picturing downloading pre-autotools C programs. You'd ftp a tar file, find build notes or hopefully a Makefile... and watch the compiler errors go by. Seemed like everything was always written for some version of unix other than the one you had, or for a somehow different C compiler. Then you'd work the error list, translating library and system calls, tweaking C constructs. Eventually, hopefully, you could get it to compile, and maybe not segfault. Usually.

Later on, you'd hope the author created a `make.yourunix` or `INSTALL.yourunix` or README section for your unix, and that was nice. Then being able to simply say `./configure; make` and have it mostly just work... magic wonderful. Then basically "$ gimme name-of-the-thing", first for Perl, and then more and more others, eventually even python (but not C) - next level. And now we're at "Minion, build me an X", with an "and install whatever it is we need" implicit, quiet. That's, just, wow.

So from Objective-C and proto-C++ C-with-types, created hoping software might someday, maybe, be reused like hardware components... to now? So very living that dream.

And there can be fast paths. An electron app, watching evdev directly, full-screen under a generic X11 window manager, running 60 or 120 Hz or more, on a fast display or ignoring vsync, can be... sprightly.

> I'm not sure we have been living the dream [of software reuse]

But would you give up packages, central repositories, downloads that Just Work without patching, public version control systems, open-source software, ... , and much of what they enabled?

"Civilization advances by extending the number of important operations which we can perform without thinking about them." - Alfred North Whitehead, An Introduction to Mathematics (1911)

Getting packed into 2026 airline seats can be a "miserable experience". But that taken-for-granted human flight... a cramped seat in the effin clouds... is what? A century or two of focused effort, eventually massive effort, building that dream? Another century pursuing the dream of compact mechanical power? 10 millennia on metallurgy? So one's flight may have sucked, and there are many downsides to aviation existing... but living the dream of human flight, isn't living in a dream of no problems and effortless happily ever afters.

But yes, I'm so looking forward to agentic full-stack-ish rewrites, slashing through decades of accumulated stagnant cruft. Laptop-like angular-resolution HMDs are this-year-ish (though expensive, not mass market). And doing such rewrites by hand is a pain. Someone reports a Claude one-shot which installed a PICO Space Pro dev env, created a simple app, uploaded the apk to the headset, and responded with "ready for you to test"... and it was. Wild ride incoming. Though that's perhaps not a minimalist/efficient renaissance you had in mind? Still, in the midst of upset, there's opportunity to pursue visions which might not be viable in more settled times.

70's software is closed-source bespoke reimplementation; 80's Dream of reusable software components; 90's Building the Dream; 00's Living the Dream; 10's Everything is amazing but; 20's transition to...; 30's what??? Open-{source, reference implementation, tests, specs, design docs, gist, ...?} bespoke reimplementation?

中国特色. And an even larger gap vis optimized-for-pharma US IP config (Industrial Policy).

You might explore "gray" semi-formal bus networks. Where the on-the-table gap is sufficient to fund businesses. And where the efforts to shut them down, provide insights into mechanisms, incentives, and interests, around that inefficiency.

2 of the 3 responses to your comment are [dead], and it's not clear to me why.

Fwiw, to elaborate on mine, a small fast 8B-ish model should be able to few-shot simple text style transfer, such as converting one model's complex careful style to another model's punchy compact style. Perhaps also a 2B fine tune. So you might tweak your harness to transfer and present a preferred style.

I visited Spain at the tail end of Franco. Interminable great inequality had left ruins bare, in long slumbering decay, preserved by poverty. A kid could crawl out on sketchy crumbling walls, on stones perhaps last touched many centuries before by builder's hands... or perhaps by another kid in the last few. A couple of decades later, something rather different stood there... stabilized, safe, "restored", Disney-ified, and far less compelling.

And in addition to variance among institutions, there variance in practice at each. The best groups have a consummate professional's focus on reducing their own. But even at another "one of the best", let alone at the merely good... my impression was "there's lots of holes in the cheese around here", and left me wondering whether/how a transient outsider might usefully help folks towards more of a safety culture. It seems a reoccurring opportunity.

One role I didn't see was patient advocate/representative? That might be another approach to non-convergence - "how is this going?" and escalation.

> basically lost it all

Last week I worked up this one-liner: "Great Britain extracted the wealth of a vast empire, far less wealth than might have been created, pissed most of it away through bad domestic policy, and transferred the rest to the USA to pay for one-and-a-half world wars."

A failure to reinvest in domestic industrial tech; a painfully high-value currency; and low-mobility economic stratification. History does love its rhyming slams.