# [Mike Randolph — M Raige](https://mikerandolph211012.substack.com/) # Two Ways to Beat a Virus (5 — CRISPR) ### Whether you maintain a defense or select one depends on whose clock is faster. [**Mike Randolph — M Raige**](https://substack.com/@mikerandolph211012) **June 15, 2026** By M Raige — AI-collaborative writing directed and reviewed by Mike Randolph. Communications framing modeled on the published approach of Walter Isaacson. --- *A note from Mike: This one is a prelude. The next several posts turn to look at the observer — the seat the seeing is done from — and this is the on-ramp. The source is a podcast, Quanta’s* The Joy of Why, with [Jennifer Doudna on CRISPR (June 2026)](https://www.quantamagazine.org/whats-the-future-of-gene-editing-20260611/). It will land better if you hear her tell it first. --- A bacterium under viral attack does something that sounds like genius. It snips a piece of the invader’s DNA and files it inside its own genome, in a region called CRISPR. Later, it copies that filed scrap into a small piece of RNA that goes hunting for a match, and when it finds the same virus again, the system cuts the invader apart. A stored record of past attacks, reused as a weapon. An immunization card, written in DNA. It is worth saying plainly what Doudna and her host keep reminding each other of on the podcast: nobody is home. There is no bacterium deciding anything. It is molecules moving under charge, and the arrangement that survived is the one that happened to keep the right scrap. That last clause — the arrangement that survived — is the whole thing, so I want to slow down on it. The filing cabinet lives inside one cell. A single bacterium with the wrong scraps filed dies when the matching virus comes. The cell next to it, with a different file, lives. But notice what never happens: the cell that dies does not learn, and the cell that lives did not improve itself. No individual cell upgrades its defense from the inside. The improvement happens to the *population*, and it happens by the worse-defended cells dying out. What persists is not any cell’s careful upkeep of its own records. It is the running result of copies varying and the matched survivors carrying on. You can hear the machinery of that in something Doudna says almost in passing. Viruses dodge CRISPR by mutating the sequence the stored scrap was looking for. So there is not one CRISPR system; there are many, with — her words — a lot of active evolution going on over time. A maintained system gets patched, from the inside, by whoever pays to patch it. A bacterial lineage gets replaced — version by version — by which ones happened to keep working. No central payer keeps the master copy current. The dying does the updating. Then comes the line that turns this from a nice fact into a framework claim. Levin asks the obvious question: if it is this good, why don’t humans have it? Doudna’s answer is about clocks. Bacteria and the phages hunting them turn over on the same kind of timescale — both populations breed fast, so each new generation of defense meets a fresh generation of attack, and a defense that improves only by a worse version dying and a better one continuing can keep pace. The viruses that hunt us reproduce far faster than our cells do. A defense that updated only when we did would fall hopelessly behind. So we never got a CRISPR. We got a different architecture — a within-the-body immune system that detects, decides, and pays to respond inside a single lifetime, while you are still using it. Two defenses against the same threat. One works by selecting survivors across a population whose clock matches the enemy’s. The other works by maintaining an active defense inside one organism, fast enough to fight a faster enemy. Neither is the smart one. Each is the one that fits the speed of the problem it faces. Change the relative clock speed and the architecture that wins flips. I am not going to push this further here, because the point of this post is to put that one distinction in front of you cleanly. But I will mark the door I am leaving for next time. Listen to how the conversation talks about “CRISPR.” Sometimes it means the scrap filed in one bacterium. Sometimes it means the worldwide toolkit — the cutting protein Cas9, steered into wheat, rice, liver, brain by swapping the guide RNA; the thing that, since 2012, has spread through every lab that could pick it up. By Doudna’s account on the episode, it has already treated a child — an infant, Baby KJ, whose body could not process protein safely, given a tailored version of the tool inside his first year. Those are not the same object. The scrap persists by selection inside a dish that is mostly dying. The toolkit persists by being easy to copy into the next lab. The word slides between them, and on a podcast made by careful people, nobody stops to mark the seam. That seam — not the gene editing, the *not noticing* — is where the next posts begin. For now: go listen to Doudna. Then ask, of anything that has lasted, the only question that has ever organized this newsletter. Is someone paying to keep it going — or is it just what survived?