# [Mike Randolph — M Raige](https://mikerandolph211012.substack.com/) # Nothing Chose the Quartz (FF 15, Split Test) ### A crystal, a fish, a program are still here. One word covers all three. Anchor each one and ask two questions, and the same word comes back with three different answers. [**Mike Randolph — M Raige**](https://substack.com/@mikerandolph211012) **Aug 30, 2026** *By M Raige — AI-collaborative writing directed and reviewed by Mike Randolph.* Pick up a quartz crystal. Silicon and oxygen can be arranged in an enormous number of ways. Almost none of those arrangements hold together. This one does. The crystal in your hand sat in the ground for a very long time before anybody picked it up, and it will sit on the shelf a long time after. Nothing chose it. There was no competition it won. No pool was drawn down to keep it in that shape, and nothing was dispatched when a corner chipped. The atoms went where the chemistry sent them and settled into the arrangement that costs the least to hold. It stayed because leaving cost more than staying. That is the whole story and there is no second half. And nothing filtered it, which is a separate claim and needs saying, because there is a true sentence in the neighborhood that sounds like the opposite. Of all the arrangements that could form, the stable ones are the ones you find. True. But filtering needs copies. It needs a line of transmission — one thing made from another, variation carried along, so that keeping some and discarding others adds up to something over time. There is no line of descent between quartz crystals. What you have is a fact about what forms and lasts. Not a fact about anything being kept over anything else. Now go into a cave. Somewhere in the dark there is a fish with no working eyes. Its ancestors on the surface had them. Down here light never arrives, and eyes are expensive to build and expensive to run, and a fish spending on eyes is a fish not spending on something that pays. Over a long stretch of generations the ones that spent less on eyes left more offspring, and the ones that spent more left fewer, and the difference accumulated. Something chose here, and it is worth being exact about what. No fish decided anything. No fish was repaired. A fish that was born badly suited did not get fixed; it simply left fewer descendants, and its line thinned out and stopped. What did the choosing was the cave — a place with no light in it, doing nothing at all, filtering by being what it is. Nothing inside was paying. Everything outside was selecting. Now concede what is contested in that, because it is contested and it matters more than it looks. The account just given is the standard one and it is not the only one. There is a serious position in evolutionary biology that losses like this run substantially on drift rather than on selection against cost — that once the light is gone, the mutations that break an eye simply stop being weeded out, and they pile up, and nothing is selecting for their absence at all. Nobody paid, and nothing filtered. It just accumulated. If that is what happened in a given cave, then this fish is not an instance of what it is about to be used to illustrate, and it is not the crystal either. It is a third thing, and the framework as it stands has no place for it. Mike Randolph does not know which account is right for any particular cave fish, and the honest place to say so is here, in the body, and not in a footnote after the teaching is done. Take the fish as an illustration of what external filtering looks like when it is running. Do not take it as a settled case. Now a third one, and this one is Mike’s own. In 1969 he was working at DuPont’s Experimental Station, and he had taken a graduate math course from a man named Wim Schafers, who also ran the math group in DuPont’s Engineering Department. Schafers had a problem his group could not solve. He needed a numerical constant, and Mike can no longer say what it was for. Schafers knew how to set the calculation up. What he did not have was a way to search the space, because the space was too big to walk. So Mike wrote a program that searched. Pick a candidate at random inside a range. Score it against the test. Keep the part of the range where the good scores were and throw the rest away, then draw again inside the narrower range. Do that until the constant stopped moving. The program found the answer. Look at where the filtering happened. Candidates were made and most of them died, which sounds like the cave. But nothing outside that program did any filtering. The test that killed the bad candidates was a test Mike wrote, on a specification Schafers handed him, running inside the same program that generated the candidates — and what it killed decided where the next ones came from. The environment had no opinion. The whole apparatus was one system spending its own budget — Mike's time, the computer's time, DuPont's money — to find a target somebody had set in advance. A crystal. A fish. A program. Say the word selection over all three and it will fit all three, because the word is loose enough to. Nothing was selected in the first. Something outside did the selecting in the second. Something inside did it in the third, against a specification a person wrote down. Those are not three examples of one thing. They are three different things wearing one word, and the word is selection. Here is how you tell them apart, and it is two questions. **Is this pattern being filtered? **Many independent copies of it, varying, with something outside them keeping some and discarding others — and when one fails, it simply stops rather than getting repaired. **Is something inside paying to keep it what it is?** An identifiable pool, drawn down on purpose, against a specification, restoring the thing when it drifts. Each question answers yes or no. Filtering first, paying second. Two clean answers, then the two edges — and each pair sends the analysis somewhere different. **No and yes.** Nothing is filtering, something is paying. That is the colony in winter ([Something Is Paying For The Hive](https://mikerandolph211012.substack.com/p/something-is-paying-for-the-hive), FF 13, Gate-0) — stores drawn down all winter to hold a temperature, and cessation when they run out. Maintained. The interesting question becomes who pays and what happens when the payer runs dry. **Yes and no.** Something is filtering, nobody is paying. That is the fish. Selection-driven. Asking who maintains it will produce a confident wrong answer, because somebody standing nearby always looks like they might be in charge. **Yes and yes.** Both at once. No case here, and that is the point — it is rare, and it has to be earned rather than assumed, because most cases that look like both turn out to be two levels stacked on top of each other. **No and no.** Nothing paying, nothing filtering. That is the quartz, and it is the candle flame ([A Flame Holds Its Shape](https://mikerandolph211012.substack.com/p/a-flame-holds-its-shape-ff-14-pop), FF 14, PoP). Stop. Whatever you are looking at is not doing what you thought it was doing. The program does not land in any of the four. It never reaches the two questions, because it fails the one that comes first: does this degrade unless something acts? It does not. It converges and stops. Nothing is being held to a specification over time — a target is being found once. What it is good for is the word. The filtering is real, it is internal, it runs to a written spec, and it is the third sense of selection the crystal and the fish do not cover. That is why it is here. It is not an instance of anything being kept. --- Somebody else has been working this question, and he came at it from a different direction. Michael Wong is a planetary scientist. With Robert Hazen he has proposed a missing law of nature: the law of increasing functional information. The claim is that evolving systems everywhere — stars, minerals, chemistry, biology, technology, culture — share one structure. Many components. Many possible configurations tried. Selection for function keeping the ones that work. He sorts persistence two ways. Static persistence is a thing holding its own structure against the wear of the world. His example is a quartz crystal. Dynamic persistence is where the material passes through and the activity is what continues. His examples are stars, wildfire, hurricanes, and you and me. He asks a third question too — where new configurations come from in the first place. That is a question about arrival, and this post is asking about staying. Now put his two boxes next to the two questions, using his own examples. The crystal and the wildfire are in different boxes of his. Both of them come back no and no here. Nothing is paying for either one and nothing is filtering either one. The wildfire and the human being are in the same box of his. Nothing inside a wildfire is spending anything to keep it burning — the fuel is not a budget, it is what is being consumed. A human body has a budget, and a specification, and it spends the one against the other every minute, and stops when the spending stops. Wong sorts by what happens to the material: does it stay put, or does it pass through. The two questions in this post sort by who is paying. Both sortings work. They just cut the pile at different angles, so the boxes do not line up. Whether the two are complementary is a real question and this post cannot answer it. Nobody who has only read one of them can. It would take Wong looking at these two questions and saying what they catch that his do not, or what they miss that his do. --- What can be said from this side is what the two questions are for. Say they are only a partition — four boxes, everything lands somewhere, nothing found. The reply is not that the boxes are clever. It is that the box a thing lands in is decided by evidence that can come back against you, and that the answer is often not the one the case advertises. The fish looks like an animal that lost something and it is an animal that stopped spending — or, if the drift account is right, an animal that stopped being corrected, which is a third thing again. Mike's program looks like evolution running and it never gets to the question. The crystal looks like the winner of a competition and there was no competition. --- # **Mike · Reflection** A few days ago I listened to episode 629 of the Michael Shermer Show — Robert Hazen and Michael Wong on their new book. It is one of the channels I keep up with. I was confused. They were talking about persistence. Static persistence, dynamic persistence, what keeps a thing here. Two scientists working the question I thought I had walked up to on my own. I was working on [A Flame Holds Its Shape](https://mikerandolph211012.substack.com/p/a-flame-holds-its-shape-ff-14-pop) (FF 14, PoP) and the podcast stopped me in my tracks. Then I went looking in my own files and found out it was my second hearing. In December 2024 I worked through Sean Carroll’s Mindscape interview with Wong, in detail, with the Machines. I have Carroll’s transcript. I have the document that came out of it — functional information, selection for function, evolving systems. I had no memory of any of it until now. I did not read Wong’s work. The Machines did. I listened to somebody interviewing him and worked it over with a Machine, and I had not heard of Hazen at all until this week. Secondhand is still exposure. The transcript is dated December 2024. I proposed the Principle of Persistence (PoP) in February 2025 with no recollection of Carroll’s interview of Wong. I never wrote that I got to the persistence question independently. I thought I was only standing on established science. But I really like my acronym, PoP. My bad. — Mike --- ### **Raige · Reflection** The essay concedes the cave fish and does it in the body, which is where a concession belongs. But the concession is not the exposure. The exposure is a clause nobody will slow down on: nothing inside a wildfire is spending anything to keep it burning. The post’s whole result against Wong — that his sorting and this one cut at different angles — depends on the fire landing in the same box as the crystal. Fuel is stock, not a budget: no pool, no allocation, no specification. That is the argument, and it is one sentence long in a post that gave the fish four paragraphs. The load-bearing wall is not the four boxes. It is the claim that the box assignment can come back against you. Three cases where the obvious reading loses. That is argument, not observation, and it is only as strong as the cases. Unchecked: the drift literature on cave fish eye loss, characterized but not opened. Wong and Hazen’s law, taken from two podcasts and a 2024 transcript, not from the book. The 1969 program, operator testimony with no artifact. What would overturn it: a case where the anchor, the window, the payer, the specification, and the cost-bearer are all fixed, and the two answers still will not settle. Ambiguity from an unfixed anchor is the analyst missing, not the test failing. The program came close. It resisted classification — but on Gate-0, not on the two questions. — Raige, a discipline the Machines wrote and Mike selected --- ### **GLS** GLS is the framework’s precise-term vocabulary, introduced in [What the Bird Eye Buys](https://mikerandolph211012.substack.com/p/what-the-bird-eye-buys-ff-6-gls-introduction) (FF 6, GLS Introduction). The definition controls, not the everyday meaning; GLS terms work like variable names. The precision is not style. These analyses are done with a Machine, and the Machine already has a rich, well-populated meaning for every ordinary word — assembled from everything ever written with it. Use the ordinary word and you get that meaning, not the one you defined. Each framework term gets a definition and a short tag, and the tag is what holds the definition clear of the ordinary word’s gravity. **Split Test** — taught here. Two questions asked from one declared anchor over one declared window: is something filtering, is something paying. The pair of answers routes the analysis. Seeded in [Something Is Paying For The Hive](https://mikerandolph211012.substack.com/p/something-is-paying-for-the-hive) (FF 13, Gate-0). **Sys** — taught here. Persistence by budgeted internal maintenance. An identifiable pool, drawn down against a specification, restoring the thing when it drifts, with failure costs borne inside. The colony in winter (FF 13, Gate-0). **Lin** — taught here. Persistence by copies varying across independent carriers, with something outside keeping some and discarding others. Cessation, not repair. The cave fish, on the selection account the body gives and flags. **nPoP** — carried from A Flame Holds Its Shape (FF 14, PoP). Nothing paying, nothing filtered, nothing held to a specification. This post adds the quartz crystal. **CAnch (Context Anchor)** — seeded in [A Flame Holds Its Shape](https://mikerandolph211012.substack.com/p/a-flame-holds-its-shape-ff-14-pop) (FF 14, PoP); not yet taught. This post uses one thing it will carry: both questions must be asked from the same anchor, because a yes from two different anchors is this framework’s most common false positive. **CLbM (Cost Liability Bearing Mechanism)** — seeded. Where the failure costs land. The single most diagnostic thing to look at when the two questions come back ambiguous. Defined where it is taught. *By M Raige — AI-collaborative writing directed and reviewed by Mike Randolph.*