The Most Complex Object in the Known Universe is Inside You

Added: 2026-06-04

[00:00:00]There are two trillion galaxies in the observable universe. Each one holds hundreds of billions of stars. Each star a furnace of nuclear fire burning across millions of kilm, fusing hydrogen into helium, releasing more energy in a single second than humanity has produced in its entire history. The scale of it is genuinely difficult to hold in the mind. 2 trillion galaxies, hundreds of billions of stars each. Planets beyond counting, nebula stretching across thousands of light years, and all of it, every last atom of it, moving outward from a single point of origin 13.8 billion years ago. We look at that and we feel small. We are meant to feel small. But there is a different way to measure the universe. Not by size, not by distance or mass or energy output, by complexity. by the density of information packed into a given structure, by the number of distinct states a system can occupy, the number of relationships it can form, the depth of its internal organization.

[00:01:05]And when you measure the cosmos that way, the answer points somewhere unexpected, not outward, inward. The most complex object that science has ever attempted to describe is not a star. It is not a black hole, not a galaxy, not the large-scale structure of the universe itself. It is a single human neuron, a cell so densely organized, so molecularly intricate, so dynamically responsive to its own history that every attempt to fully map one has ended with scientists revising their estimates of how much they did not know. a cell that exists in your body right now by the billions doing something that no telescope has ever captured and no equation has yet fully described. We have spent centuries looking up. We have mapped the cosmic web, traced the microwave background of the Big Bang, imaged black holes at the center of distant galaxies, and in all of that searching, we have not found anything that approaches the organizational complexity of the object sitting behind your eyes. This film is not about how large the universe is. It is about what the universe built when it had 4 billion years and the right chemistry. It is about the structure that emerged from that process. What lives inside it, how it compares to everything else we have ever measured and what it means that the thing doing the comparing is made of those same structures billions of times over. We will go inside a single neuron at the molecular scale. We will count what is there. We will measure it against the stars. And by the end, the universe will not look smaller, but the object behind your eyes will look considerably larger.

[00:02:41]This is Omni, a channel about the strange beauty of existence. From galaxies to microbes, we follow the threads that connect all living things.

[00:02:50]If that sounds like your kind of adventure, like the video, share your thoughts below, and subscribe for more journeys through the unknown.

[00:03:09]The word complexity gets used loosely.

[00:03:11]We say a symphony is complex or a city or a political system. What we usually mean is that there are many moving parts or that the whole is difficult to predict from its components. These are fair uses of the word in everyday language. But in science, complexity has a more precise meaning and that precision matters enormously when we start making claims about which things in the universe qualify as the most complex. The formal study of complexity as a discipline took shape in the second half of the 20th century largely through the work of researchers at the Santa Fe Institute in New Mexico founded in 1984.

[00:03:48]Scientists there, including the physicist Murray Galman and the biologist Stuart Kaufman, began asking a question that had not been asked systematically before. What exactly do we mean when we say one system is more complex than another? And can we measure the difference? The answer they arrived at drew heavily on an earlier body of work, information theory developed by the mathematician Claude Shannon at Bell Laboratories in 1948. Shannon had shown that information could be quantified, that the content of a message or a system could be expressed as a number of bits, a measure of how many yes or no questions would be required to fully describe the state of that system. The more possible states a system can occupy and the more relationships exist between its components, the higher its information content and therefore the higher its complexity. This framework changed the terms of the conversation.

[00:04:39]Size was no longer the same as complexity. A rock the size of a planet is enormous, but its internal organization is relatively simple. Its atoms sit in stable latice arrangements, responding predictably to temperature and pressure. The number of distinct states the rock can meaningfully occupy is limited. By contrast, even a modest biological structure, a length of DNA for instance, encodes an extraordinary density of information in a very small physical space. The molecule is not large, but the number of possible sequences, the number of meaningful configurations it can represent is astronomical.

[00:05:17]According to the principles established by Shannon and later developed by researchers including Gregory Chaitton and Andre Colmorov, the complexity of a system is not measured by how much space it takes up, but by how much information is required to describe it fully. With this lens in place, the comparison between the cosmos and biology becomes genuinely interesting. Stars are massive, but structurally they are not especially complex. A main sequence star like our sun is essentially a self-gravitating ball of plasma held in equilibrium between the inward pull of gravity and the outward pressure of nuclear fusion. Its behavior is governed by a relatively small number of physical variables, mass, composition, temperature, pressure, magnetic field strength. Astrophysicists can model a stars behavior with remarkable accuracy using equations that fit on a few pages.

[00:06:08]The star is not simple in the colloquial sense, but in terms of information density, in terms of the number of distinct internal states it can occupy, and the number of relationships between its components, it is far less complex than a living cell. This is not a small claim. The sun contains roughly 2 * 10 to the power of 57 atoms. It is incomprehensibly large by any human standard. But the organizational relationships between those atoms are by biological standards relatively coarse.

[00:06:38]There are no feedback loops in a star that read their own output and adjust their behavior accordingly. There are no structures that encode information about past states and use that encoding to modify future behavior. A star burns. It does not learn. A neuron does both. The human neuron is a cell which means it is already operating at a level of molecular organization that dwarfs stellar structure. All cells contain within them the basic machinery of life.

[00:07:06]A genome encoding billions of base pairs. A transcription apparatus reading that genome selectively and dynamically.

[00:07:12]A translation system producing proteins on demand. A membrane maintaining a precise chemical environment against constant pressure from the outside world. and a network of internal structures performing thousands of simultaneous biochemical reactions in a space measured in micrometers.

[00:07:29]According to a 2019 analysis published in the journal, cell systems by researchers including Matio Van Hook and colleagues at the European Biioinformatics Institute, a typical human cell contains between 10,000 and 100,000 distinct protein types at any given moment. each one performing a specific molecular function. Each one produced in a quantity that fluctuates in response to internal signals and external conditions. The regulation of that protein population alone. The mechanisms by which a cell decides which proteins to make, how many to make, when to degrade them and replace them, represents a system of feedback and control that has no equivalent in any human engineered machine. But a neuron is not a typical cell. A neuron is something considerably more elaborate.

[00:08:16]Most cells in the human body are roughly spherical or polygonal in shape. They are compact. Their diameter is generally somewhere between 5 and 30 micrometers.

[00:08:25]A neuron, particularly a motor neuron in the spinal cord or a projection neuron in the cerebral cortex abandons this geometry entirely. A single neuron can extend a primary projection called an axon across distances that are relative to the size of the cell body almost incomprehensible.

[00:08:42]The motor neurons that run from your spinal cord to the muscles of your feet have axons that extend up to 1 meter in length. The cell body from which that axon extends is approximately 20 micrometers across. The axon therefore reaches out to a distance roughly 50,000 times the diameter of its own nucleus.

[00:09:00]If the cell body was scaled up to the size of a basketball, that axon would stretch approximately 50 km. This geometry has consequences that go far beyond the visual. A cell that extends a meter long projection into the body must solve a set of logistical problems that a compact cell never encounters.

[00:09:18]Proteins synthesized in the cell body need to travel the full length of the axon to reach the synaptic terminals at the far end. The molecular motors that perform this transport, primarily proteins called canin and dinin move along internal tracks called microtubules at speeds of roughly 1 to four micrometers/s.

[00:09:38]For a meter long axon, that means some cargo takes days to make the transit from origin to destination. The neuron must therefore plan ahead, maintaining a continuous supply chain of molecular components across a structure that by the standards of cell biology spans an almost geological distance. And at the far end of that journey, at the synaptic terminal, a further layer of complexity waits. The syninnapse is the junction between one neuron and the next. The point where an electrical signal is converted into a chemical signal, transmitted across a gap of roughly 20 nm, detected by receptor proteins on the receiving cell, and converted back into an electrical signal. Each of these steps involves dozens of distinct proteins working in precise sequence.

[00:10:24]The release of neurotransmitter molecules from synaptic vesicles requires the assembly of a docking complex called the snare complex. A molecular machine involving at least three distinct proteins that must bind to each other in a specific order, fuse with the cell membrane, release their contents, and then be recycled for the next release. According to research published by Thomas Sudhof of Stanford University, who received the Nobel Prize in Physiology or Medicine in 2013 for his work on synaptic vesicle fusion, a single synaptic release event involves the coordinated action of more than 100 distinct proteins operating on a time scale of less than a millisecond. One synapse, 1 millisecond. more than 100 proteins, each one produced by a gene, regulated by a network, degraded and replaced on a schedule tuned by experience and activity. And a single neuron does not have one synapse. It has thousands. The scale of what lives inside a single neuron and what a single neuron maintains across its entire surface and length is the subject of the next section of this story. But to understand why those numbers matter, it helps to first understand how long it took science to see them at all. And why every tool we built to look at neurons revealed a structure more intricate than the one we thought we were looking for.

[00:11:39]For most of human history, the brain was not understood as a collection of discrete cells. It was understood as something more like a continuous fabric, a dense network of tissue so tangled and uniform in appearance that the individual units within it seem to dissolve into one another. This was not unreasonable. Without the tools to distinguish one cell from the next, what the early anatomists saw when they examined brain tissue was something that looked under the rudimentary magnification available to them like a seamless web. The prevailing theory by the 19th century held that the nervous system was structurally continuous, a single interconnected net through which signals flowed without interruption. It was called the reticular theory and it had significant intellectual support behind it including from one of the most technically gifted microscopists of the era.

[00:12:30]Camilo Golgi was an Italian physician and scientist who in 1873 developed a staining technique that would prove transformative. He dissolved silver nitrate into potassium doumate and applied the resulting solution to thin slices of brain tissue. The reaction, for reasons that chemists still do not entirely understand, stains only a small fraction of neurons in any given sample, somewhere between 1 and 5%. But it stains those neurons completely, filling the entire cell with a dark precipitate that renders its full shape visible against a pale background. For the first time, scientists could see a neuron in its entirety. not just fragments of it, but the full architecture, including the cell body, the branching dendrites, and the long extending axon. Golgi looked at what his own technique revealed and concluded that it confirmed the reticular theory. The axons appeared to him to merge into a continuous network.

[00:13:28]He published this interpretation and defended it for the rest of his career.

[00:13:32]A Spanish neuroanatomist named Santiago Ramani Kahal encountered Golgi's staining method in the late 1880s and saw something entirely different. Ramani Kahal, who had trained as an artist before studying medicine and brought a draftsman's eye to microscopy, made thousands of preparations and spent years examining them under gaslight in his home laboratory in Valencia. What he saw in preparation after preparation was not a continuous network. What he saw was gaps. Individual cells separate from one another, reaching toward each other but never actually touching. He called the gap between neurons a syninnapse, a word derived from the Greek for junction or clasp. And he argued that this separation was not an artifact of the staining process, but the fundamental architecture of the nervous system. His evidence was meticulous. His drawings were extraordinary. He published his findings across hundreds of papers and eventually compiled them into a two volume opus the tur delistma dereos first published in 1904. In 1906 both Ramonica Hal and Golgi were awarded the Nobel Prize in physiology or medicine for their contributions to the understanding of the nervous system.

[00:14:48]They shared the prize and they shared a stage in Stockholm and they disagreed entirely about what the nervous system was. Golgi used his acceptance speech to argue against Kahal's neuron doctrine.

[00:14:59]Kahal used his to defend it. The argument was not settled by logic or by the weight of published evidence alone.

[00:15:06]It was settled by a machine that did not yet exist.

[00:15:10]In 1955, Sanford Pal working at the Rockefeller Institute alongside George Pelade published the first electron microscopy study of the central synapse in vertebrates. The electron microscope uses beams of electrons rather than light to form images. And because electrons have a much shorter wavelength than visible light, the technique can resolve structures far smaller than any optical instrument. Pale and Ped trained this technology on brain tissue and saw for the first time what Ramonica Hal had argued for decades must be there. A physical gap between neurons, a cleft of 20 to 30 nm unmistakably present, separating one cell's membrane from the next. The reticular theory was finished.

[00:15:53]Kashal had been right. And the tool that vindicated him simultaneously opened the door to a new era of discovery. Because when scientists looked at the syninnapse through the electron microscope, they found that it was not a simple gap. It was a machine.

[00:16:08]Within the presinaptic terminal, the electron microscope revealed small spherical structures clustered near the membrane. These were synaptic vesicles.

[00:16:16]Each one a lipid bubble roughly 40 nanome in diameter. Each one filled with uh thousands of molecules of neurotransmitter.

[00:16:24]At the moment of neural firing, these vesicles fused with the cell membrane and released their contents into the cleft. On the other side of the gap, the postsaptic membrane was lined with a dense scaffolding of receptor proteins.

[00:16:37]Each one tuned to detect a specific neurotransmitter and respond by opening an ion channel or initiating a biochemical cascade. The syninnapse, which had appeared to be merely a discontinuity in the neural fabric, turned out to be an enormously complex transduction device, converting an electrical signal into a chemical signal and back again in less than a millisecond.

[00:16:58]Every new imaging technology applied to neurons since that first electron microraph has followed the same pattern.

[00:17:04]It has revealed more. The optical microscope showed a shape. The electron microscope showed an interior. But the electron microscope also had limits. To prepare tissue for electron imaging required fixing it in chemical preservatives, dehydrating it, embedding it in resin, and slicing it into sections thinner than a cell membrane.

[00:17:24]The process preserved structure but destroyed chemistry.

[00:17:28]What scientists saw was a snapshot of a dead cell, exquisitly detailed, but fundamentally static. A photograph of a machine with its power off. For decades, the living neuron, the neuron in the act of firing and receiving and modifying itself, remained invisible. The technique that began to change this was fluorescent microscopy, which uses molecules that emit light when activated by specific wavelengths to label particular proteins within living cells.

[00:17:55]developed through the latter decades of the 20th century and refined significantly by the work of Roger Cien at the University of California San Diego who shared the Nobel Prize in chemistry in 2008 for his development of green fluorescent protein. Fluorescent microscopy allowed scientists to watch individual proteins move inside living neurons in real time. They could label a specific receptor, a specific vesicle protein, a specific ion channel, and track its behavior as the cell fired and rested and fired again. What they saw was not the orderly static structure that the electron microscope had suggested. What they saw was a system in continuous flux. Proteins moved laterally across the membrane. Receptors clustered and dispersed in response to activity. The posts synaptic density which had looked like a stable scaffold in fixed tissue turned out to be a dynamic assembly constantly exchanging components with the surrounding cell.

[00:18:50]The closer science looked at the neuron, the more it moved. The most recent advance in this progression is cryo electron tomography, a technique in which neurons are rapidly frozen at temperatures below - 150°.

[00:19:04]a process fast enough to trap water in a glassy rather than crystalline state, preserving molecular structures in something very close to their native configuration. The frozen sample is then imaged from multiple angles using an electron beam and the resulting images are computationally assembled into a three-dimensional volume at near resolution. This approach applied to synaptic structures by researchers including Ruben Fernandez Busan Diego at the University of Gotting and published in studies through the 2000s and early 2020s has revealed the interior of the synapse at a level of detail that simply could not have been imagined when Pai first photographed the synaptic cleft in 1955.

[00:19:45]Individual protein complexes are visible. The docking machinery that holds synaptic vesicles at the membrane awaiting release can be seen as a forest of molecular tethers. Each one a distinct protein structure holding a vesicle in precise position nanome from the membrane it will fuse with when the signal arrives. And this is a single synapse. One junction between two neurons 20 nanome wide requiring the coordinated operation of hundreds of distinct molecular machines. A typical neuron in the cerebral cortex makes between 5,000 and 10,000 of these junctions. A perkinga cell in the cerebellum, one of the most elaborately branched neurons in the human body, can receive input through as many as 200,000 synaptic contacts. Each one is a separate molecular device. Each one is regulated independently. Each one can be strengthened or weakened based on the history of activity passing through it.

[00:20:38]a process that forms the cellular basis of memory and learning.

[00:20:43]The history of looking at neurons is a history of recalibrating expectations downward in terms of simplicity and upward in terms of wonder. Every generation of scientists has looked at the same object and found it more complex than the generation before had reported. Not because earlier scientists were careless, but because the object genuinely contains more than any single tool has ever been able to reveal at once. To understand what modern science has found inside a single neuron, it helps to start with a number. In May 2024, researchers from Harvard University and Google published a paper in the journal Science Describing the most detailed reconstruction of human brain tissue ever produced. The project led by Jeff Likman of Harvard's Department of Molecular and Cellular Biology in collaboration with Google's contomics research team began with a piece of human temporal cortex roughly the size of half a grain of rice. The sample was stained, coated, and sliced into more than 5,000 sections, each one just nanome thick. Those sections were scanned by electron microscope, producing 225 million two-dimensional images. Those images were then assembled by machine learning algorithms into a three-dimensional reconstruction of every cell and every connection within the sample. The data set that resulted required 1.4 pab of storage. One pabyte is 1 million GB. The total digital content of the United States Library of Congress has been estimated at around 15 tab. The reconstruction of this single grain of rice-sized fragment of one human brain required roughly 90 times that amount of storage. It contained approximately 57,000 individual cells of which around 16,000 were neurons along with 32,000 gal cells and 8,000 blood vessel cells. Within that volume, the team identified 150 million synaptic connections, 150 million synapses in a piece of tissue smaller than a rice grain. The human brain contains approximately 1 million mm of tissue. The Harvard and Google reconstruction covered roughly 1 millm. Scaling that figure across the full volume of the brain produces an estimated 150 trillion synaptic connections. A figure consistent with independent estimates from neuroscience literature.

[00:23:06]To put that in perspective, the observable universe is estimated to contain approximately one septillion stars, which is one followed by 24 zeros. The synapse count of a single human brain, 150 trillion, is 150 followed by 12 zeros. The brain does not rival the universe in raw synapse numbers. But what the brain synapses are doing is something no star has ever done. Each one is a regulated junction capable of strengthening or weakening in response to activity, capable of storing information about the past, capable of participating in the kind of dynamic patterned behavior that underlies thought, memory, and perception.

[00:23:47]The question then is what lives inside each of those junctions and what lies within the neurons that maintain them?

[00:23:54]The answer begins with the neuron's architecture which is itself a feat of biological engineering with no equivalent in any other cell type. A typical neuron in the cerebral cortex has a cell body called the soma roughly 10 to 20 micrometers in diameter. From that cell body extend two types of projections. The dendrites are shorter highly branched processes that receive incoming signals from other neurons. The axon is a single long projection that carries outgoing signals to the neurons targets. In motor neurons that run from the spinal cord to the feet, that axon can extend a full meter. In cortical neurons, the axon is shorter, but still extends far beyond the cell body, branching into numerous collaterals that contact thousands of other cells. The logistics of maintaining a cell across this geometry are formidable. Every protein that functions at a distant synaptic terminal must be synthesized somewhere and transported to where it is needed. Most protein synthesis in neurons occurs in the cell body and the primary mechanism for delivering proteins to distant locations is axonal transport. A system in which molecular motor proteins carry cargo along tracks of microtubules that run the length of the axon. The motors involved are members of the kinosin and dinian families which move along microtubial tracks in opposing directions with kinosin carrying cargo outward toward the axon terminal and din carrying material back toward the cell body. As reviewed in studies from the journal of cell science and published research by Erica Holbower at the University of Pennsylvania, these motors move at speeds of 1 to 4 micrometers/s.

[00:25:31]Meaning that for a 1 m axon, a cargo package originating in the cell body can take anywhere from 3 days to nearly 2 weeks to reach the far end. The neuron therefore cannot wait until a synaptic terminal is depleted of a component before beginning to resupply it. It must maintain a continuous moving inventory, a rolling distribution system operating 24 hours a day, sustaining function at thousands of synaptic sites simultaneously.

[00:25:56]Among the most critical cargos in this system are mitochondria, the organels responsible for generating adenosine triphosphate, the cell's primary energy currency. Synaptic transmission is energetically expensive. Each action potential, each vesicle release event, each cycle of ion pump activity that restores the membrane potential after firing requires a substantial investment of cellular energy. A synaptic terminal stripped of its local mitochondria will fail within minutes. Research published in nature reviews neuroscience by Zu Hangen and Chiian Kai at the National Institutes of Health has documented how neurons regulate the positioning of their mitochondria with remarkable precision, anchoring them at high activity synapses and mobilizing them when energy demand shifts elsewhere. The system responds to electrical activity to calcium levels to metabolic signals, adjusting the distribution of its power sources across the cell in real time.

[00:26:53]All of this is happening before we have even entered the synaptic terminal itself. At the terminal where axon meets dendrite across a 20 nm gap, the machinery becomes still more elaborate.

[00:27:04]The presinaptic terminal contains hundreds of synaptic vesicles. Each one a lipid sphere approximately 40 nm in diameter packed with several thousand molecules of neurotransmitter. These vesicles do not float freely. They are organized into distinct pools based on their proximity to the release site and their readiness to fuse. The pool of vesicles immediately adjacent to the membrane and docked for release is called the readily releasable pool. A reserve pool sits further back awaiting mobilization. The transition of a vesicle from reserved to readily releasable requires the action of specific proteins. And the fusion of a docked vesicle with the membrane upon arrival of an action potential requires the assembly and rapid disassembly of the snare complex. A tripartite molecular machine composed of the proteins synaptobre syntaxin and snap 25. According to work by Thomas Sudhof at Stanford University, the snare complex is conserved across virtually all ukarotic cells, suggesting it represents one of the most ancient membrane fusion mechanisms in the history of life. Its operation at the syninnapse calibrated to a millisecond time scale is regulated by a protein called synaptagmmin which functions as a calcium sensor triggering vesicle fusion only when the local calcium concentration crosses a threshold set by the arrival of an action potential on the post synaptic side of the cleft. The receiving end of the synapse contains a dense protein assembly called the postsaptic density. In excitatory synapses which use the neurotransmitter glutamate, this density is anchored by a scaffolding protein called PSD95 which organizes clusters of glutamate receptors at precise positions relative to the presinaptic release sites. The receptors themselves are molecular machines, protein complexes that span the membrane and open an ion channel when glutamate binds, allowing calcium and other ions to flow into the post synaptic cell and trigger downstream signaling cascades. Research using superresolution microscopy by Daniel Shet at the University of Bordeaux has shown that these receptors are not static. They move laterally within the membrane entering and leaving the postsaptic density on time scales of seconds to minutes. And the number of receptors available at any given moment is regulated by the history of synaptic activity. The basis of synaptic plasticity. All of this presinaptic vesicle pools, calcium, sensors, snare complexes, postsaptic scaffolds, mobile receptor clusters is contained within a single synapse. A structure 20 nanome wide and a single neuron maintains thousands of them. Now multiply the synaptic machinery by 10,000. the approximate number of synapses on a typical cortical neuron. Add the axonal transport system running the full length of the cell sustaining dozens of distinct molecular populations simultaneously.

[00:29:59]Add the nucleus containing 3 billion base pairs of DNA being read selectively by transcription factors responding to the neurons electrical history. Add the ribosomal machinery synthesizing proteins on demand throughout the cell body and at specific dendritic locations. Add the mitochondrial network distributed strategically across the cell's geography to meet localized energy demands. Add the endopplasmic reticulum winding through the cell body and into the dendrites, folding proteins and regulating calcium. Add the proteosomes and autophagosomes degrading and recycling molecular components on precise schedules. What you have at the end of that accounting is not a cell in the conventional sense. It is a city. A city with a permanent population of interacting molecular machines numbering in the hundreds of millions, a logistics network spanning distances of up to a meter, thousands of regulated communication interfaces with neighboring cities, and a central archive of 3 billion characters of genetic information being read and acted upon in real time, and it weighs approximately 1 nanog. The human brain contains approximately 86 billion neurons. That figure comes from a 2009 study by neuroscientist Susanna Hercuano Hel at the Federal University of Rio de Janeiro, who developed a method called the isotropic fractionator, dissolving brain tissue into a suspension of cell nuclei and counting them directly rather than estimating from tissue samples. The technique produced a more rigorous count than had previously existed. 86 billion is the current scientific consensus, though the number carries an uncertainty of around 8 billion in either direction given variation between individuals.

[00:31:42]For the purposes of the comparison that follows, the precise figure matters less than the order of magnitude. The Milky Way galaxy contains somewhere between 200 billion and 400 billion stars, depending on which estimates of the galaxy's faint outer regions are used.

[00:31:58]This means the Milky Way has two to four times more stars than the human brain has neurons. The commonly repeated claim that the brain contains as many neurons as there are stars in the Milky Way is strictly speaking not accurate. The galaxy wins that particular count. But neuron count is not where the comparison becomes interesting. It is where the comparison stops being fair to the brain. Each neuron forms synaptic connections with other neurons. A typical cortical neuron forms somewhere between 5,000 and 10,000 synaptic contacts, though this varies enormously by cell type. A pinia cell in the cerebellum with its vast dendritic tree can receive synaptic input from as many as 200,000 other neurons. Taking a conservative average of 7,000 synapses per neuron and multiplying across 86 billion neurons yields an estimated 600 trillion synapses. Other methodologies using different assumptions about sampling and connectivity produce figures closer to 100 trillion. The range in the literature is wide, but every serious estimate lands somewhere between 100 and 500 trillion. For this comparison, the lower bound is sufficient, 100 trillion synapses. The Milky Way contains between 200 billion and 400 billion stars. This means the synapse count of a single human brain, even at the lowest credible estimate, exceeds the star count of the Milky Way by a factor of roughly 250.

[00:33:24]The brain contains roughly 250 Milky Ways worth of connection points, packed into a volume of 1.4 L, weighing approximately 1.4 kg. But the universe is considerably larger than the Milky Way, and honesty requires acknowledging the full scale of the comparison. The observable universe is estimated to contain approximately 2 trillion galaxies and the best current estimates of the total stellar population of the observable universe sit at around 1 septillion stars which is one followed by 24 zeros. 100 trillion synapses is one followed by 14 zeros. The universe wins the raw count comfortably. Any claim that the brain contains more connections than there are stars in the universes as a matter of arithmetic false. And the omni approach to science requires saying so clearly. But the comparison changes entirely when we ask not how many but what kind. A star is a gravitationally bound plasma maintaining nuclear fusion through a relatively small number of governing variables. Its behavior is in principle fully described by the equations of stellar physics.

[00:34:31]Given the mass, composition and age of a star, its behavior is predictable within well understood tolerances. It does not learn. It does not alter its internal configuration in response to its history. It does not store information about previous states and use that stored information to modify future output. A star is to use the precise language of complexity science. A low information density structure despite its enormous physical scale. A synapse is none of these things. The molecular mechanism underlying synaptic strengthening, the process called long-term potentiation, was first described experimentally by Timothy Bliss and Tea Lommo in 1973 working at the University of Oslo.

[00:35:15]They discovered that repeatedly stimulating a synaptic pathway caused the response at that synapse to increase in strength and that this strengthened response could persist for hours, days, and in some preparations weeks. The cellular basis of this effect has been worked out in considerable detail over the subsequent decades. When a synapse is strongly activated, calcium ions flow into the postsaptic cell through a class of receptors called NMDA receptors which function as molecular coincidence detectors opening only when two conditions are simultaneously met.

[00:35:50]The presence of neurotransmitter in the cleft and sufficient electrical depolarization of the postsaptic membrane. This dual requirement means the NMDA receptor responds specifically to correlated activity to two neurons firing together rather than one firing in isolation. The cellular implementation of what the psychologist Donald Heb proposed theoretically in 1949 that neurons which fire together tend to wire together. When calcium enters through NMDA receptors, it activates a cascade of intracellular signaling including an enzyme called calcium calm modulin dependent protein kinase 2 abbreviated CK2 which phosphorolates and drives the insertion of additional AMA receptors into the postsaptic membrane. More receptors means a stronger response to the same amount of neurotransmitter. The syninnapse has in effect remodeled itself based on its own recent history.

[00:36:45]Research from Richard Huganir at Johns Hopkins University published across multiple studies including work in the proceedings of the National Academy of Sciences has detailed how this receptor trafficking is precisely regulated with different subtypes of AMA receptors inserted and removed on specific schedules depending on the type and pattern of synaptic activity.

[00:37:07]The late phase of long-term potentiation, the version that persists for days and longer, requires the synthesis of new proteins. The syninnapse does not simply acquire more receptors. It grows structurally, adding membrane, building new scaffolding, physically enlarging the postsaptic density. This structural change requires gene expression in the cell nucleus, the synthesis of messenger RNA, the transport of the RNA to the sinapse, and its local translation into protein. An event at a peripheral sinapse, a contact point, possibly a millimeter from the cell body in a cortical neuron, sends a molecular signal back to the nucleus that changes which genes are active. The nucleus responds by sending molecular instructions back outward. The synapse is remodeled. The connection is strengthened. The memory is written.

[00:37:57]Every one of the brain's estimated 100 trillion synapses is capable of this.

[00:38:02]Every one of them maintains an independent history of activation. Every one of them exists at a point somewhere along a continuum of strength from near silent to maximally potentiated. A continuum shaped entirely by the pattern of activity that has passed through that particular junction over the lifetime of the organism. The number of possible states the brain can occupy, the number of distinct configurations its synaptic weights could take is a number that has no meaningful comparison in the physical universe. Computational estimates, including work by neuroscientist Paul Reber at Northwestern University, have placed the storage capacity of the human brain in the region of 2 1/2 pabytes.

[00:38:42]Though this figure is a rough approximation of a quantity that resists clean measurement. To put that in the only terms that seem adequate, the Harvard and Google reconstruction of 1 cubic millm of human cortex required 1.4 pabytes of storage to encode the structural organization of that tissue at nanoscale resolution. The estimated information storage capacity of the entire human brain, the dynamic functional capacity, not just the structural map, is of the same order of magnitude. The architecture and the information it can hold are in a sense matched to each other. The brain is exactly as complex as it needs to be to store a human life. A star burns for millions or billions of years and is at the end of that time structurally identical to a star of the same mass and age anywhere else in the galaxy. Its history leaves no record in its structure. It is consumed by its own physics and leaves behind a remnant determined entirely by its initial mass.

[00:39:40]Nothing that happened to it is preserved. Nothing that passed near it altered what it became. Two neurons that fire together even once are slightly different from what they were before.

[00:39:50]Their junction is fractionally stronger.

[00:39:53]A molecular trace of that event persists in the receptor composition of the postsaptic membrane in the scaffolding proteins of the postsaptic density in the gene expression profile of the nucleus. If they fire together repeatedly, that trace becomes permanent. The syninnapse grows. The connection deepens. What passed between them is remembered. This is not a metaphor. This is the physical mechanism of memory. It happens in a structure 20 nanome wide, 100 trillion times over in the space behind your eyes. The previous section established what a synapse does to its own structure in response to activity. But the synapse does not operate in isolation from the rest of the cell. And the rest of the cell, it turns out, is doing something that places the neuron in a category that has no parallel anywhere else in biology.

[00:40:42]Most of the body's cells have a fixed relationship with their genome. They carry the same DNA as every other cell in the organism, read a subset of it determined by their cell type and developmental history, and that pattern of gene expression remains relatively stable over time. A liver cell reads liver genes. A skin cell reads skin genes. The genome is consulted, but it does not change in response to experience. It is inherited architecture, not living record. A neuron is different in ways that are only beginning to be fully understood.

[00:41:15]Michael Greenberg at Harvard Medical School has spent several decades studying what happens inside a neuron's nucleus when the neuron fires. His laboratory has been instrumental in characterizing a class of genes called immediate early genes. Genes whose expression can be switched on within minutes of a neuron receiving strong synaptic stimulation without requiring any new protein synthesis to initiate the response. These genes are preloaded for rapid transcription held in a state of readiness by their chromatin configuration waiting for a calcium signal from an active synapse to trigger their expression. Among the best characterized of these genes as CFO, EGA1 and a gene called ARC which stands for activity regulated cytokeleton associated protein. ARK is particularly striking. When a neuron fires in response to a learning experience, ark messenger RN NA is transcribed in the nucleus within minutes, exported into the dendrites, and translated into protein at specific recently active synapses. The protein it encodes plays a direct role in regulating AMA receptor trafficking at those synapses, the same receptor insertion process that underlies long-term potentiation. ARC is a gene that reads neural activity and then participates in writing the structural response to that activity at the specific synaptic sites that were recently active. Research published in Frontiers in Molecular Neuroscience by Minatoara, Akioshi, and Akuno at Kyoto University has documented how arc expression is selectively upregulated in the precise subset of neurons involved in a learning event, making it one of the most specific molecular markers of memory formation known to science. This birectional conversation between synapse and nucleus, between peripheral activity and central gene expression is the cellular architecture of learning. An experience activates a population of synapses. Those synapses send calcium signals that propagate to the cell nucleus. The nucleus responds by transcribing genes whose protein products travel back to the activated synapses and modify them structurally.

[00:43:20]The modification is written in receptor number in scaffolding geometry in membrane area in protein composition and because the late phase of synaptic potentiation requires new protein synthesis. The synapse that has been modified is also a synapse that has altered the gene expression program of the entire cell. This is the cell responding to its own history at the level of its DNA. But the birectional relationship between activity and gene expression is only one dimension of the neuron's responsiveness to experience.

[00:43:49]There is a second layer operating not through changes in which genes are expressed but through changes in the physical structure of the DNA itself.

[00:43:58]This is the domain of epigenetics and in neurons it operates in ways that have only become experimentally accessible in the past decade. Epigenetic modifications are chemical changes to the proteins around which DNA is wound or to the DNA bases themselves that alter the accessibility of genes without changing the underlying genetic sequence. The most studied of these modifications involve the addition or removal of chemical tags on the histone proteins that package DNA and the addition of methyl groups directly to cytosine bases in the DNA strand. These modifications are inherited when cells divide, carrying a record of past gene expression states forward into daughter cells. But neurons almost never divide.

[00:44:41]In a post-motic cell, epigenetic modifications are not passed to daughters. They are passed to the future of the same cell. Work from Greenberg's laboratory published in the journal cell in 2017 by Straoud Sue and colleagues demonstrated that early life neuronal activity establishes epigenetic states at specific gene log that persist throughout the lifetime of the neuron.

[00:45:04]Genes activated by experience in early development leave a molecular signature in the chromatin structure of the nucleus that remains detectable in the adult brain shaping the cell's transcriptional responsiveness for decades. A neuron that was strongly activated in the first weeks of life is at the level of its chromatin a different cell than one that experienced deprivation during the same period.

[00:45:27]The history of the organism is written into the physical organization of its neurons DNA.

[00:45:33]This is not a metaphor. The molecule that encodes genetic information, the molecule that has carried heritable biological instruction since the origins of life on Earth is also functioning in neurons as a ledger of individual experience.

[00:45:47]The genome of a neuron is not a static blueprint. It is a document that has been annotated by the life of the organism that carries it. The implications of this reach further still when we consider local translation, a phenomenon that has reshaped the understanding of how neurons maintain their enormous synaptic populations. For a neuron with thousands of synapses distributed across a dendritic tree that may extend hundreds of micrometers from the cell body, waiting for proteins to be synthesized in the nucleus and transported outward through the exonal and dendritic transport systems is often too slow. Strong synaptic activity demands a rapid structural response. A synapse that fires repeatedly needs new AMA receptors, new scaffolding proteins, new membrane, and it needs them within minutes, not days. The solution neurons have evolved is to store messenger RNA at synaptic sites and translate it on demand. Research published in the journal Science by Anne Sophie Hoffner and colleagues at the Maxplank Institute for Brain Research demonstrated that both presinaptic and postsaptic compartments contain active ribosomes and messenger RNAs and that protein synthesis at these local sites is ongoing even in the absence of stimulation.

[00:47:01]Within 5 minutes of activity, the proportion of synaptic terminals showing active translation increases substantially. The synapse is not waiting for the nucleus to respond. It has its own local protein factory stocked with a subset of the cell's messenger RNA inventory translating whatever is needed at the moment of demand.

[00:47:21]The catalog of messenger RN as that reside in dendrites includes thousands of distinct species as documented in genomewide studies by Erin Schuman at the Max Plank Institute for Brain Research. These transcripts encode proteins involved in receptor trafficking, cytokeleletal remodeling, energy metabolism, and signaling.

[00:47:41]Essentially, the full molecular toolkit required for synaptic maintenance and modification. Their presence in the dendrite is not random. Each transcript is transported there by specific RNA binding proteins that recognize sequence elements in the messenger RNA and the translation of individual transcripts is regulated by the local activity state of the synapse so that the right proteins are made at the right synapses at the right times. The neuron is therefore not one cell with a unified response to stimulation. It is a cell containing thousands of semi-autonomous molecular units. Each one capable of locally synthesizing proteins. Each one maintaining its own independent state of potentiation or depression. Each one participating in a distributed information storage system of extraordinary specificity and capacity.

[00:48:29]The nucleus coordinates this system from a distance through gene expression and epigenetic modification. The local translation machinery executes specific modifications at individual sites. The external and dendritic transport systems move materials between the two levels.

[00:48:46]And the entire architecture, the nucleus, the transport network, the local factories, the synaptic machinery is tuned by the history of electrical activity passing through the cell. The philosopher and neuroscientist Patricia Churchland has written that the mind is what the brain does. At the cellular level, what the brain does is this. It maintains a billion-year-old molecule at its center and uses the firing of electrical signals to annotate it. It distributes the instructions encoded in that molecule outward along projection spanning meters where molecular motors carry cargo day and night to thousands of regulated contact points. At each contact point, a machine 20 nanome wide reads the history of everything that has passed through it and modifies itself accordingly. Every one of your 86 billion neurons is doing this right now without pause, without your awareness, and without any instruction from you.

[00:49:41]Everything described so far in this film has been structural and molecular. The cytokeleton, the transport motors, the synaptic vesicles, the snare complexes, the receptor trafficking, the immediate early genes, the epigenetic modifications. These are mechanisms.

[00:49:58]They are physical processes measurable with instruments expressable in the language of chemistry and molecular biology. Science has made extraordinary progress in describing them. The descriptions are not complete but they are advancing rapidly and there is no particular reason to doubt that they will eventually be comprehensive. There is however a question that lies beyond all of these mechanisms. A question that the mechanisms themselves cannot answer and that may represent the deepest unsolved problem in all of science. The question is not what neurons do. The question is why what neurons do feels like something. Right now, as you listen to these words, neurons in your audiary cortex are firing in response to the frequency patterns of the sounds reaching your ears. The firing of those neurons triggers cascades in adjacent regions. Working memory circuits encode the sequence of words. Language processing regions extract meaning.

[00:50:54]Associative areas connect that meaning to prior knowledge. All of this is electrochemical. All of it is in principle describable in the terms we have been using throughout this film.

[00:51:05]But there is something else happening simultaneously. Something that the molecular description does not capture and has not yet accounted for. You are hearing this. There is an experience of sound, a qualitative character to the words, a sense of what it is like to listen. That experience, that felt quality of conscious perception is not visible in any description of synaptic transmission or gene expression. It is not in the calcium concentration. It is not in the receptor density. It is not in the firing rate of the auditory neurons. The gap between the physical description of neural activity and the existence of subjective experience is what the philosopher David Charmer's writing in 1995 called the hard problem of consciousness. The easy problems in Chararma's framing are explaining how the brain processes information, integrates sensory data, directs attention, generates behavior. These are difficult scientific questions, but they are at least in principle tractable by the methods of neuroscience. The hard problem is different. It asks why any physical process, however complex, gives rise to experience at all. Why is there something it is like to be a brain? Two major theoretical frameworks have been developed in recent decades to address this question, and neither has resolved it, though both have changed how neuroscientists think about what neurons are actually doing.

[00:52:28]The first is integrated information theory developed by the neuroscientist and psychiatrist Julio Tenoni at the University of Wisconsin Madison. and first formally proposed in a paper published in BMC neuroscience in 2004.

[00:52:42]Tenoni's theory begins not with neurons but with experience itself asking what the defining properties of consciousness are and what physical properties a system must have to exhibit them. He identifies two key features.

[00:52:55]differentiation meaning that consciousness contains a vast number of distinct possible states and integration meaning that those states are unified into a single coherent experience rather than fragmented into independent streams.

[00:53:10]From these observations, Tenoni deres a mathematical quantity he calls fi defined as the amount of information generated by a system above and beyond the information generated by its parts independently. A system with high FI in which the whole generates substantially more information than the sum of its components is in Toni's framework conscious to a degree proportional to that five value. The theory makes a striking prediction. Consciousness is not exclusively a biological phenomenon.

[00:53:39]Any physical system with sufficiently high integrated information possesses some degree of experience. The human brain with its vast number of neurons densely interconnected in patterns that generate enormous integrated information has very high-fi and therefore very rich consciousness.

[00:53:56]A simple system with few components and limited integration has very low FI and correspondingly minimal or no experience. The theory is precise enough to be tested, at least in principle, using neuroiming and methods such as transcranial magnetic stimulation, which Tenoni and his collaborators, including Marello Masamini at the University of Milan, have used to measure the complexity of brain responses in sleeping and waking subjects, in patients under anesthesia, and in cases of disorders of consciousness. The theory is also deeply controversial.

[00:54:30]Critics, including the computer scientist Scott Aronson, have pointed out that certain simple computational architectures that seem intuitively non-concious would score high on FI under Tenon's formulation, raising doubts about whether the metric actually tracks consciousness or merely a different kind of information complexity.

[00:54:48]The debate remains active and unresolved, but integrated information theory has forced neuroscience to confront a question it had largely avoided. not just how the brain processes information but whether the way it is organized gives rise to something irreducibly experiential.

[00:55:05]The second major framework approaches the problem from a different direction entirely. Carl Fristen at University College London, one of the most cited neuroscientists in the world according to citation metrics compiled across multiple databases has proposed what he calls the free energy principle first articulated in a paper in Nature Reviews neuroscience in 2010. Kristen's framework begins with a thermodynamic observation. Living systems resist the tendency toward disorder by maintaining their internal organization. To do this, they must model their environment, predict incoming sensory signals, and act to minimize the discrepancy between prediction and reality.

[00:55:44]Fristen formalizes this as the minimization of variational free energy, a measure of the difference between the brain's generative model of the world and the actual sensory data it receives.

[00:55:55]In this framework, perception is not the passive reception of sensory information. It is the brain's active attempt to confirm its own predictions.

[00:56:03]What we experience as seeing, hearing, touching is the brain's best guess about the causes of its sensory inputs, constrained by its generative model and updated when prediction errors arise.

[00:56:15]Consciousness, in this view, is what it feels like to be a system engaged in continuous predictive self-modeling.

[00:56:22]Friston's principle has been extended into what he calls active inference, a framework in which action itself is understood as another form of prediction, the brain moving the body to bring sensory inputs into line with expected states. As Friston described in a 2024 interview in the National Science Review, the framework offers what he terms self-evidencing, the imperative for a system to maximize evidence for its own existence. Like integrated information theory, the free energy principle has provoked both enthusiasm and skepticism. It is mathematically elegant and biologically grounded. But whether it solves the hard problem or merely redescribes it in different terms is contested. What it does unambiguously accomplish is placing the neuron at the center of something larger than signal processing. In Friston's framework, the neuron is not simply a relay station for information.

[00:57:16]It is a component of a system that models reality that generates hypotheses about the world that experiences the gap between expectation and observation as something we eventually give the name of surprise or recognition or understanding. Neither theory has won.

[00:57:32]The hard problem remains hard and this is not a failure of science. It is an indication of the depth of the object we are examining.

[00:57:41]When physicists work at the boundaries of their field, they encounter similar situations. The reconciliation of quantum mechanics and general relativity. Two theories that each describe their respective domains with extraordinary precision, but that cannot be simultaneously applied to the same physical situation represents an unsolved problem that has occupied theoretical physics for nearly a century. The incompatibility does not mean that physics is wrong about quantum mechanics or wrong about general relativity. It means the object of study, the universe at the most fundamental scale, is richer than either theory alone can contain. Neuroscience is in a comparable position. The molecular description of the neuron is not wrong. The description of synaptic transmission, receptor trafficking, axonal transport, gene expression, and epigenetic modification is not wrong.

[00:58:32]These mechanisms are real and well evidenced. But the existence of subjective experience, the fact that a physical system organized in the way a brain is organized generates something that feels like something from the inside adds a layer that the molecular description has not yet reached and may require conceptual tools that do not yet exist to address. The neuron sits at the intersection of two unsolved problems.

[00:58:56]The problem of how matter becomes life, answered partially by molecular biology, but still incomplete at the boundary of inanimate chemistry and self-replication.

[00:59:05]And the problem of how life becomes experience, unanswered by any framework currently available to science. The object we've been examining throughout this film is not only the most molecularly complex structure ever described, it is also the structure where the deepest questions about the nature of reality are concentrated.

[00:59:22]Science has been trying to read the neuron for more than a century. Every tool it has built for that purpose has returned the same result. More is there than was previously known.

[00:59:33]The progression has been orderly and relentless. The optical microscope showed that neurons are discrete cells.

[00:59:39]The electron microscope showed that synapses are physical gaps containing molecular machinery. Fluoresence microscopy showed that the machinery is in constant motion. Cryo electrontomography showed the machinery at near resolution. And yet the complete wiring diagram of a single human brain has never been made. And the obstacles to making one illuminate by contrast just how much the object contains.

[01:00:03]In October 2024, a consortium of more than 50 research laboratories around the world working under the name Flywire and led by teams at Princeton University published nine papers simultaneously in the journal Nature describing the first complete conneum of an adult brain. The brain belonged to a fruitfly Drosophila melanagaster.

[01:00:24]It contained 139,255 neurons and more than 54 million synaptic connections. The project required contributions from researchers across multiple continents, machine learning algorithms to segment and trace individual neurons through electron microscope imagery, and a global community of citizen scientists performing manual proofreading on the automated reconstructions.

[01:00:49]The NIH brain initiative, which supported the work, described it as the largest and most complete conneum of an adult animal ever created.

[01:00:58]139,000 neurons, 54 million synapses, a brain the size of a grain of sand, and the project required the efforts of 50 laboratories and several years to complete. The human brain contains 86 billion neurons and an estimated 100 trillion synapses. It is approximately 620,000 times larger than the fruitfly brain in neuron count.

[01:01:22]A complete human connector reconstructed at the same resolution and detail as the Flywire map would. According to estimates from researchers, including those at Zeta AI, a conto company working toward that goal, require data storage on the order of one zetabyte.

[01:01:38]One zetabyte is 1 trillion GB. To contextualize that figure, John and Guy, director of the NIH brain initiative, told Stat News in 2024 that a zetabyte of data is roughly equivalent to all the data transmitted across the entire internet in a single year. The complete wiring diagram of a single human brain at the resolution we can now achieve for small pieces of it would consume more storage than a year of global internet traffic. This is not a limitation of current technology that will be overcome soon. It is an indication of what the object contains. The data storage requirement scales directly with the structural complexity of the tissue. If a cubic millm of human cortex requires 1.4 pabytes and a human brain contains approximately 1 million mm, the full human conneto at equivalent resolution requires on the order of 1.4 zetabytes.

[01:02:32]The universe spent 13.8 8 billion years constructing a structure that would require more storage to document than all of humanity's digital records combined.

[01:02:42]The human conneto project launched by the National Institutes of Health in 2009 and concluded in 2021 mapped connectivity in the human brain using diffusion MRI and functional imaging at a resolution that captures large fiber tracts and regional connectivity patterns but not individual synapses.

[01:03:02]That project produced an invaluable atlas of the brain's large-scale organizational architecture, identifying 360 distinct cortical areas with characteristic connectivity profiles and functional properties as documented in a landmark parcellation published in Nature in 2016 by Matthew Glasser and David Van Essen at Washington University. It established a framework that subsequent research continues to build on. But it did not and by its nature could not descend to the level of the individual syninnapse. The gap between the resolution of the human conneto project and the resolution required to map a brain the way Fly Wire mapped the fruitly is a gap of roughly a billion in scale. Science is working steadily to close it. The timeline for closing it is measured in decades, not years. The implications of this incompleteness are not discouraging.

[01:03:54]They are clarifying. What they tell us is that the object we have been examining throughout this film is not a problem that has been solved. It is a frontier that has barely been entered.

[01:04:04]The molecular mechanisms of synaptic transmission are well characterized. The principles of long-term potentiation are established. The role of activity dependent gene expression in memory formation is documented. The existence of local translation in dendrites is confirmed. And none of this taken together amounts to a complete description of what a single neuron does, let alone what 86 billion of them do in concert. The universe we can see contains approximately 2 trillion galaxies. Each galaxy holds hundreds of billions of stars. The total stellar population of the observable universe is a number with 24 digits. That number is very large. But a number's size is not the same as its depth. The stars of the observable universe are by the standards of information density relatively uncomplicated. They are massive, ancient, and beautiful. But the space of states they can occupy is small compared to the space of states occupied by the tissue behind your eyes. We began this film by proposing to measure the universe inward rather than outward, by complexity rather than by size. The result of that measurement is now in view. The most complex structure science has ever attempted to describe is not located at a cosmic distance. It is not in a galaxy that light takes billions of years to reach. It is inside a skull that weighs approximately 5 kg and sits on a human body that evolved over 4 billion years from a single ancestral cell on a small rocky planet orbiting a middle-aged star in the outer arm of one galaxy among two trillion. The universe in some sense spent all of that time and all of that distance on this on a gram of tissue that can model the world well enough to look back at the stars and ask where they came from. On a cell that encodes its own history and the physical reorganization of its molecular architecture. On a structure that fires electrical signals to other structures and from that firing somehow generates the experience of being alive, of being curious, of being the kind of thing that wonders what it is. We do not yet know how that last step happens. We do not yet have the tools to fully map the structure where it happens. We cannot yet fully read the object that is doing right now the reading. But we are closer than we have ever been. And the thing we are getting closer to understanding is not somewhere out there past the edge of what telescopes can see. It is the instrument with which we are looking.