CeleryKills

Uploads, Brain Wires, and Who Gets Left Holding the Compost Bin

I live in a part of the country where people will argue earnestly about espresso grind size and salmon runs in the same sentence. So when folks here start whispering about a coming “technology rapture,” I perk up the way I do when the barista slides me a macchiato I didn’t order. Are we really talking about some fraction of humanity uploading itself into a server farm while the rest of us keep patching our Gore‑Tex and mowing the median strips? Or is this just very expensive fan fiction with a grant number?

The phrase “tech rapture” sounds like theology, but the scaffolding is research‑y: superintelligent AI, brain‑computer interfaces, whole‑brain emulation, and the legal recognition of digital minds. In fiction, people simply step into the light. In reality, this would be more like building the Space Needle with a Leatherman: iterative, fussy, and full of inspections (Sandberg & Bostrom, Whole Brain Emulation: A Roadmap, 2008).

What would an actual “ascension” require?

First, you would need a brain map that doesn’t just sketch highways but tags every alley, cul‑de‑sac, and garden gate. Connectomics is crawling forward. In 2024, a global consortium published the first complete, synapse‑level connectome of an adult fruit fly brain, about 140,000 neurons and more than 50 million synapses. That is a scientific mic drop. It is also a reminder that our heads are not fruit flies (Nature package on FlyWire; Princeton releases).

Humans are orders of magnitude larger. High‑profile U.S. projects have managed cubic‑millimeter mouse cortex slabs at petascale data volumes, and the coordinators will be the first to tell you the complexity is humbling. It took on the order of $100 million and years to reconstruct a volume that accounts for roughly two‑tenths of a percent of a mouse brain, and the principal investigators called the main casualty of all that data “understanding,” which is a very academic way to sigh (Nature feature on MICrONS).

Meanwhile the Human Connectome Project has given us gorgeous macroscale maps of human brain networks, but those are not synapse‑level. They tell us where the freeways run, not which driveway belongs to which memory. Useful, but it will not upload your grandmother. Not this decade, and probably not the next (NIMH HCP overview; HCP site).

I realize this all sounds defeatist, but there have been startling proofs‑of‑principle on preservation. In 2016, researchers won the Brain Preservation Foundation’s prize by demonstrating aldehyde‑stabilized cryopreservation that preserved every examined synapse in a rabbit brain. That is not revival, and it is not consciousness in a jar, but it is a data‑fidelity milestone that refuses to go away (Brain Preservation Foundation; PRWeb announcement).

The climb, not the leap

A true “upload” would require mapping, encoding, and emulating neural dynamics at a level that preserves identity, which is a bigger ask than a wiring diagram. The classic roadmap in this space lays out the chain: whole‑brain imaging at synapse resolution, reliable segmentation, model building, and enough compute to run the thing. Even the optimists wrote “decades” between lines, and those were the optimists (Sandberg & Bostrom, 2008).

When I read the latest reviews, I see connectomics inching toward larger mammals, but the bottlenecks are not merely technical. They are logistical and economic. One cubic millimeter of human cortex at electron‑microscopy resolution already lands you in petabytes, and stitching a whole brain would leave you staring at zettabyte‑class storage and compute. A 2025 assessment of human cortical datasets flatly called out the computational choke points and ethics of managing data of that scale. Translation: we can’t just add more GPUs and call it good (Springer Neuroinformatics “Matters Arising,” 2025).

The neural ports, for when you don’t want to move in completely

While the upload crowd waits for better histology and cheaper exabytes, BCIs are sliding into clinics. The state of implantable trials tracks slow but real. A 2024–2025 survey of intracortical BCI trials counted a few dozen participants across two dozen studies since the late 1990s and showed increasing longevity of implants alongside big gaps in diversity and standardization. Translation: measurable progress, plus paperwork mountains and bio‑engineering headaches (Nature Reviews Bioengineering, 2024).

You have also seen splashy demos from Neuralink. Their first human participants controlled cursors and games by thought, then confronted thread retraction and signal drop‑offs that were partially rescued in software. That is how medical devices mature: an enthusiastic sprint collides with biology, and both settle on a jogging pace. The company has since broadened trials and claims a growing cohort, while regulators and competitors keep them honest. I admire the ambition and keep a calendar reminder for follow‑up data, not the livestreams (Neuralink PRIME trial page; The Debrief 2026 coverage).

So if a “rapture” is the moment you drift into the cloud, today looks more like a rowboat. BCIs are restoring communication and cursor control, maybe soon speech for ALS, and that is already a miracle worth funding. But a one‑click departure from wet biology is a different weather system.

Odds, with coffee in hand

People always ask me for numbers, which is risky because numbers on a topic like this harden quickly into folklore. I’ll take the risk, lightly salted.

By 2100, partial mind‑upload demonstrations that reenact constrained aspects of a person’s memory or behavior from preserved, high‑fidelity microstructure feel plausible to me, given the pace of connectomics and modeling. I put that at thirty percent. It is not that we will capture a full self; it is that we may show a convincing, bounded continuity between preserved tissue and an emulated function. The fly brain map in 2024 nudged me upward on that guess (Nature FlyWire package; ScienceDaily summary).

Widespread cognitive migration, where significant numbers of people elect to move most of their daily experience into emulations, reads like a ten percent event. It hinges on pipelines that do not exist, costs that would make a data‑center CFO weep, and on political legitimation of digital persons. Law commissions have begun to openly float personhood for certain AI systems as a way to patch liability gaps. The conversation is moving from seminars to draft papers, but even proponents call it “radical” and premature. I am in the wait‑and‑watch camp (Legal Cheek on Law Commission paper, 2025; Yale Law Journal essay, 2024).

An instantaneous “tech rapture,” where a critical mass of humanity pops into servers by choice over a long weekend, sits below one percent. I cannot reconcile that with the current slope of brain mapping, compute economics, clinical device timelines, or the pin‑striped tempo of administrative law. If I’m wrong, you can find me at the dock in Ballard, waving at the fiber trunk.

The sci‑fi canon in the room

I do not shake off speculative fiction when I think about this, I invite it in. Greg Egan’s Permutation City treats software selves with a combination of mathematical rigor and existential migraine that I still find bracing. It is the cleanest meditation I know on what counts as continuity when the substrate changes (Egan, Permutation City, 1994). Iain M. Banks’s Surface Detail isn’t subtle. He asks whether simulated afterlives count as lives, then sets off fireworks to make sure you remember the stakes (Banks, Surface Detail, 2010). Richard K. Morgan’s Altered Carbon swaps connectomes for cortical stacks and asks how identity behaves when bodies become rental cars. The noir is a feature, not a bug, if you want to test your ethics at night (Morgan, Altered Carbon, 2002).

Am I being too Pacific Northwest about this, weighing flow charts and fish ladders while the metaverse crowd builds cathedrals? Maybe. But I have stood in dark rooms with datasets so large the monitors seem to float, and I have also watched a single failing solder joint turn a perfect demo into a shrug. My bias is to add one more clamp to the scaffolding before we pour the concrete.

The awkward question: who ascends, who stays?

Let’s suppose uploads do arrive in a workaday way, with sign‑up forms and billing codes. Who gets in? Connectomics at scale will not be cheap for a long time. BCIs are already rolling out with inclusion problems. The Nature review I mentioned calls out under‑representation of women and under‑served groups in implant trials and argues for a registry to avoid a future where augmentation begins as another inequity multiplier. I grew up hearing that technology is the tide that lifts all boats. Standing on the dock, I can also see who owns a boat (Nature Reviews Bioengineering, 2024).

If digital persons ever get legal standing, whose server bills get paid when a downturn hits? Who arbitrates custody when someone forks a self? We do not have satisfying answers, but the fact that mainstream bodies are even sketching the menu of personhood options tells me the debate is moving from edgy panels to procedural drafts (Legal Cheek on Law Commission paper; Yale Law Journal essay).

My bottom line, stated plainly

The “technology rapture” is a useful story because it forces us to ask what we owe to the people who do not or cannot step through the door. It also forces us to fund the grind. If we want the option to preserve minds with fidelity, we need better brain mapping, saner data infrastructure, and ethics that scale beyond a lab. If we want BCIs that add agency instead of extracting it, we need inclusive trials and boring procurement. The future, if it comes, will not arrive like thunder. It will sound like a rack of fans in a windowless room and a stack of consent forms on a clipboard. I can live with that. I can even be excited by it. But I am keeping my raincoat.

As for odds, I’ll keep mine where I set them. Partial mind‑upload demos by 2100: thirty percent. Widespread cognitive migration: ten. Instantaneous rapture: call me from the dock.

References

• Banks, I. M. Surface Detail. Orbit, 2010.
• Brain Preservation Foundation. “Small Mammal BPF Prize Winning Announcement.” 2016.
• Egan, G. Permutation City. HarperCollins, 1994.
• Legal Cheek. “Should AI be given legal personhood? New Law Commission paper raises ‘radical’ possibility.” Aug. 7, 2025.
• Morgan, R. K. Altered Carbon. Gollancz, 2002.
• Nature. “The FlyWire connectome: neuronal wiring diagram of a complete fly brain,” special package, Oct. 2, 2024.
• Nature. Michael Eisenstein, “A milestone map of mouse‑brain connectivity…,” Apr. 15, 2024.
• Neuralink. “The PRIME Study,” Clinical Trials site, accessed Feb. 2026.
• NIMH. “Human Connectome Project (HCP) — What was the HCP?” 2024–2025.
• Princeton University / ScienceDaily. “Entire brain of adult fruit fly mapped,” Oct. 2, 2024.
• Sandberg, A., & Bostrom, N. “Whole Brain Emulation: A Roadmap.” Future of Humanity Institute, 2008.
• The Debrief. “Neuralink Reaches 21 Patients…,” Feb. 16, 2026.
• Yale Law Journal. Hon. Katherine B. Forrest, “The Ethics and Challenges of Legal Personhood for AI,” Apr. 22, 2024.
• Additional technical context: Springer Neuroinformatics, “Overcoming Neuroanatomical Mapping and Computational Barriers in Human Brain Synaptic Architecture,” 2025.
• Nature Reviews Bioengineering. “The state of clinical trials of implantable brain–computer interfaces,” 2024.