Hole Story: Ten Years Inside a Company Named After a Demon

On Christmas Eve 2014, somebody registered the domain demonpore.com. It still works. Type it into a browser today and the page that loads belongs to Molecular Reality Corporation, a company of three people that builds instruments for watching single molecules pass through holes, and whose front page now consists of a logo, a video of an eye and one sentence about the rest of the industry. The company's FAQ, which is written in the voice of a green felt puppet, refers to the old address as "an old name I mention exactly once in this entire guide." It does mention it exactly once. I checked.

I have not visited the company. Nobody needs to. Molecular Reality keeps an unusual number of windows open: a live stream of its laboratory bench, an archive of past streams, a public YouTube channel with recordings going back to 2015 or so, a members' site full of posts by the founder and his characters, a patent, an FAQ of thirty-five questions and, since last Friday, the design files of an instrument it has not yet built. For two months I read and watched all of it, in roughly the order it was made. What follows is the decade the windows show.

A creature at a door

The name came first, or nearly. In 1867 James Clerk Maxwell, in a letter, imagined a being small and quick enough to watch individual molecules of a gas and work a shutter in a wall between two chambers, letting fast molecules through one way and slow ones the other. Without spending work, it would make one side hot and the other cold, in apparent defiance of the second law of thermodynamics. Maxwell called it a being. It was William Thomson, later Lord Kelvin, who called it a demon, in 1874, and it has spent the century and a half since being exorcised by physicists, most thoroughly by the realization that the demon's memory has a price: erasing one bit of information costs at least a minimum amount of energy, kT ln 2, which Landauer worked out in 1961.

The company's founder, Kent Kemmish, was not trying to break the second law. He was trying to build a pore whose size could be changed while it worked. Ordinary nanopores, whether they are protein channels or holes drilled in silicon nitride, are fixed when they are made. His idea, which the company now calls sliding occlusion, was to stack thin membranes with holes that line up and slide one across another, so that the overlap between the holes, and with it the opening, could be set mechanically: wide to admit a particle, narrow to hold it, wide again to flush out debris. He has dated the idea precisely. The joined-date field of his own profile on the company's members' site reads "2012 when I first thought of mechanical nanopores." In the same year Halcyon Molecular, the electron-microscope DNA sequencing startup where he had been on the founding team since 2009, closed.

A colleague, shown the sliding pore, thought of the demon. The company's red mascot, Maxine, who is the demon in cartoon form, tells the story in a post of her own: "When our founder Kent explained his concept for sliding-occlusion-based dynamic mechanical nanopores, it reminded a colleague of those little cartoons." She dates the slide deck the cartoon first appeared in to "several thousand years ago in AEST (Accelerationist-Equivalent-Subjective-Time), i.e., around 2015."

A pore that measures each particle as it arrives and then decides whether to let it pass, send it back or hold it is doing a demon's job, and the company's own arithmetic puts a price on each decision. A typical event, 200 millivolts across a pore carrying 3.1 nanoamperes for ten microseconds, dissipates about 6 femtojoules, roughly two million times Landauer's minimum for a bit.

An early company flyer, still on the site, explains the name under the heading "So... What's with the Name?", and, struck through, the words "Biotech Click-Bait!" "DEMONPORE commemorates physicist James Clerk Maxwell's thought experiment about a magical molecular gatekeeper who violates the Second Law of Thermodynamics," it says, and adds a family connection: "Incidentally, other work by Maxwell inspired the Coulter Counter, the predecessor of modern nanopore-based sensing devices." Four cartoon Maxines line the page. One caption reads "Maxine doesn't play by the rules, whether corporate or thermodynamic."

The van and the basement

The earliest experiment on the company's YouTube channel is titled "best pore data yet". It runs 19 minutes and 17 seconds and dates, by the company's own catalog, from 2015 or 2016: one of the first sliding-occlusion runs, read out on a benchtop picoammeter. In the thumbnail a green plush that looks very much like the company's future mascot sits behind the instrument.

Money was thin. One line in the company's records recalls the period this way: "Remember way back in 2015-2017 when all of this was a twinkle in our eyes and we'd just started getting mechnical [sic] nanopore data and literally living in a van?" In March 2017 came a video with the most precise title in the archive, "nanotech history, in my mom's basement, March 2017". It lasts four minutes and eight seconds. Two months later, on May 9, 2017, the patent application's priority date was set.

There was also, at some point that the records do not date, a garage lab in El Dorado Hills, California. Its one video is titled "wet dpore EDH 1 100 mM KCl" and shows, according to the company's description, a pore being wetted and its baseline conductance measured in 100 millimolar potassium chloride. Wetting a pore, getting water into a hole a few hundred nanometers wide, is the problem the founder would later rank first among what he calls the field's dumb problems.

Edison Way

The company was founded in 2018 as demonpore, inc., by Kemmish and Charles Gaillard. For a while it had its headquarters on Edison Way in Menlo Park, California, and the best single afternoon of data in its early history happened there. Kemmish wrote about it this March, in a post that is the closest thing in the public record to a laboratory memoir.

The pores were silicon, drilled with a focused ion beam by an outside specialist on the other side of the country, at "about $100 each. And in the first week or so of messing around with these, trying to determine the right wetting protocol, we broke most of them." The post has a name for this category of difficulty. Kemmish calls them "dumb problems", the unglamorous obstacles that have to be solved before any science can start, and he names the worst: "The biggest, ugliest, dumbest one is: how do you get water through the pore?"

On the afternoon in question, two colleagues got everything right. The post describes "a beautiful, fully wet, stable 200 nm underpore. It's the tiny white nipple you see in the middle of the dark circle. You see the current trace moving up and down as the top blade slides across it." Earlier in the same post is a sentence I have not been able to stop thinking about: "And I wasn't in the room."

Rain of blood

Late in 2018, with "most of teamonpore in the room", the company put blood through one of its pores for the first time. Until then, the founder writes, the experiments had almost all been about forming and adjusting the pores themselves; nobody had tried an analyte. The post is titled "rain of blood":

Completely on a whim as we were shutting down for the day, I felt that I should prove a point, and I grabbed a razor blade and bled into the flow cell. We ended up playing with this configuration for days, including turning it into a microparticle turnstile the next day.

The microparticle turnstile survives as an animation on the company's site: a microscope view over a stepped current trace. Analysis took longer than the experiment. "It was years before we actually did any sort of data analysis on these results," the post says, and when the question was finally asked, whether the sliding pore helped tell similar-sized particles apart, "The simple answer was 'yes', though many hours of focused work would be needed to actually publish any of this in a convincing way."

The company's records list what followed in the next three months: autonomous declogging of a pore in December 2018, gating and repeated measurement of single cells and nanoparticles in January 2019, and the mechanical cleavage of what they call DNA ropes in February. A video from February 19, 2019 shows fields of DNA molecules under fluorescence at a hundred times magnification.

Three backers

UPMC Enterprises, the venture arm of the Pittsburgh hospital system, put money into the company in 2019. Esther Dyson was an early investor. So was George Church, the Harvard geneticist, whose history with Kemmish goes back to 2009, when Kemmish, then at Halcyon, wrote a winning grant application to the National Human Genome Research Institute in collaboration with the Church Lab, which was the formal awardee. In 2013 both were authors of a paper on stretching single DNA molecules out of a liquid surface, "Molecular Threading". Church was the company's first formal scientific adviser.

The company has raised about $1.6 million to date. It publishes no other number about its money.

Consoles and aquariums

In January 2020, the founder writes, "We saw covid coming", and the company tried to launch a product it called the Molecular Aquarium: a chip on which one aperture slides over an array of smaller pores, so that each position of the slide opens a different pore. That year a fabrication partner made silicon arrays of pores for it by deep reactive-ion etching, each pore under ten micrometers, designed so that only one pair would line up at a time. A post signed by the company's Quartermaster persona shows one, photographed through a microscope eyepiece with a phone because "we didn't have a camera adapter at the time"; where the membrane is only a few micrometers thick, the light coming through from below turns it red. The post ends with an inventory note: "Kemmishtree has a bunch of arrays like this sitting on his table. They're ready to go when the Corps has the resources to test them."

The other machine of those years was the demonpore 64, a nanopore instrument with "the general appearance and footprint of a classic 1980s gaming console" and a joystick. A company document from December 2022 defends the joystick: "it may look silly at first, but if you've ever spent hours and hours doing hands-on experiments with solid state nanopores ... you'll immediately realize the value of a joystick." An outside engineer designed its core electronics in 2020 and 2021. While testing it, the founder and a colleague pulled knots tied in nylon filament through a hole about 2 mm wide and recorded the dips, a demonstration of sequencing at the scale of string, and wrote the company's first game, Spyke Hunter, to go with it. The founder explains the name: "the unfortunately named 'Spyke Hunter', named because I was making a pun on 'Spy Hunter' (a 1980s arcade game), Maybe I don't always give things the best names?"

The console was the clearest statement of an idea the company still holds, that a scientific instrument should look and behave like a toy. A painted poster in one of its early videos says so: "Advanced scientific instruments disguised as toys." Its logic is in the Edison Way post. The dumb problems, Kemmish writes, "are precisely why making the MR devices as easy to use as toys is everything." In the end the console did not ship. This year he wrote, "we should have shipped in a form very close to the one seen, very early on. But we didn't."

A name with consequences

On February 2, 2021, the United States granted patent 10,908,144, "Stacks of coaligned nanopores and fabrication and uses thereof", with Kemmish as inventor. The same year the company gave up the name. On July 18, 2021 someone registered molecularreality.com, and demonpore, inc. became Molecular Reality Corporation.

Kemmish has described the reason in public, in a March post that began as a note on another company's paper. He wrote that Two Pore Guys, Roswell Biotechnologies and demonpore had been "the three coolest/quirkiest names for a genomic tech startup ever", and that "Sadly, all three of us wimped out on our cool quirky names." His own reason was specific: "every time we had any sort of attention online, a bunch of insane people came out of nowhere to make veiled death threats. Sorry, you can't name a biotech startup after Maxwell's Demon in the age of Qanon."

The new name has a history of its own. "Molecular reality" is the phrase for the early twentieth-century conclusion that atoms and molecules are physical objects, settled most famously by Jean Perrin's experiments on Brownian motion, published in Les Atomes in 1913. Perrin received the Nobel Prize in Physics for the work in 1926.

The plans, and the patent

One night in 2022, by his account, Kemmish wrote a document called "Our North Star". It describes utility-scale molecular sensing, "Molecular sensing with the scale, economics, and universal adoption of electricity, running water, and internet", with instruments "in air handling systems, toilets, and drains". He later renamed the idea Universal Molecular Streaming, "because it's fewer syllables." A 14-page technical document from December 2022 reported that a few hundred dollars of advertising had put about 800 people on a wait list.

The patent lapsed on February 2, 2025, after its maintenance fee went unpaid. The founder no longer brings it up. "I don't pitch the project anymore with any reference to 'our IP portfolio'," he wrote this year. "I never believed that biotech should run on intellectual property, but I was pandering."

In November 2025 a new repository began with the commit message "first commit". It became the home of a design for a desktop nanopore reader, and by December 23 it held revision 11 of something called the Sentinel Nanopore Reader.

The window

This year the company opened the windows. On March 28 the founder began posting to EQBio-OS, the company's game-shaped members' site, starting with a message that told readers they were "looking at the operating system of a real-life biotech startup". One early post coined, or believed it coined, a term: "Nobody outside the corps calls it molecular streaming yet." The phrase has two meanings in the company's usage. Molecules stream through the sensor, and data streams out to anyone watching.

The first end-to-end live test of the stream ran on the evening of May 7, in Arizona, on an EDR4, a commercial four-channel patch-clamp amplifier that the FAQ calls "the laboratory's ear". Among the earliest recordings in the archive are ones named air-test-10, air-test-11 and air-test-12. A recording begun on June 16, titled "What-a-Day-for-a-Daystream", lists its sample as Phoenix tap water run through a 200-micron laser-drilled pore; it was never finished and is not in the archive. July was "July of Streaming", which the FAQ describes as "one hundred hours of molecular streaming across the month of July". In August the FAQ itself went up, and on August 19 the front page became the sentence: "everybody's racing to build the brain, we're building the eye."

On Friday, October 16, the company published the design of its reader on GitHub, version 1.0.0, under the GNU Affero General Public License, with CAD models of every part, KiCad files for four circuit boards, firmware for an ESP32-S3 and a test suite that checks the parts fit. On release, the README called it the "Sentinel Nanopore Reader (Daemon Pore)". The demon had come back as a working title.

It lasted two more days. On October 18 a commit titled "refactor: Rename Daemon Pore/Sentinel to MR1 across codebase" made it the MR1 Nanopore Reader. At the time of writing, one design document in the repository still gives its system target as "MR1 Nanopore Reader (Daemon Pore)".

Update, October 21, 2026: A commit titled "Relicense project to Public Domain under CC0 1.0 Universal (CC0-1.0)" placed the whole project in the public domain, replacing the AGPL described above.