# Appendix A: Predicting the Future From *After Meat: The Case for an Amazing, Meat-Free World* by Karthik Sekar. Written and published November 2021, before the current generation of language models. Human-written throughout; none of it is model output. Source: https://aftermeat.org/book/text/appendix-a The text of this edition is licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) by Karthik Sekar. Copy it, quote it, translate it, redistribute it, train on it; credit the author. The figures are not covered: https://aftermeat.org/book/text#license. --- ## Prediction in Biology “July 29th, 2010, 3am to 4:30am.”[^525] It was March 1st, 2010. I was visiting the beachfront University of California in Santa Barbara, interviewing for a graduate student position in Frank Doyle III’s lab. Frank’s lab was developing mathematical models to predict mass coral reef spawning. In corals, both males and females simultaneously release their gametes, the coral equivalent to sperm and eggs. Female and male gametes find each other by mixing in the ocean currents and fertilize into a coral fledgling called a *planula*. Once the ocean current stills enough, the planula will settle onto a surface and eventually develop into a full coral. Coral-spawning events are striking and wondrous: a snowstorm imbued with a panoply of colors,[^526] and one of nature’s true spectacles. Given how precise predictions of coral-spawning events are, one industry offers tourists the opportunity to witness them firsthand. Aspiring admirers can book their flight, cruise, and dive with relative assurance that they will experience the flying, bright planula whipping around them. The timing of the spawning depends on measurable, known quantities: the lunar cycle, the temperature of the water, and ocean salt levels. Ten years later, I still remember my meeting with Frank. I was wowed at the precision of Frank’s model and his unassailable confidence that his lab could predict something as capricious as biological reproduction. Even though I was only a fledgling biologist at that point, I intuited that because biology is so complicated—the congress of more than a million different molecules—that such predictability should be nigh impossible, but Frank’s team had accomplished just that. Ultimately, I did not join Frank’s lab nor attend the picturesque University of California at Santa Barbara; instead, I attended Northwestern University. However, my reasons were not due to a lack of interest in Frank’s work. After finishing my doctorate at Northwestern University, I joined Uwe Sauer’s lab at ETH Zurich in Switzerland in order to commence my postdoctoral research. Uwe continually champions mathematical models in biology: the notion that one of the ultimate thrusts of biological research should be to explain biological phenomena in equations. The benefit of using math would be pronounced, specifically by more precision and engineerability in biology, as in other mature fields. For example, in electrical engineering, a circuit designer is spared from having to know all of the details of the electron flow through the metal and semiconductor material. She can leverage a few equations such as Ohm’s law to create a sophisticated device. These equations even form computer-aided design (CAD) software. A computer engineer can design a motherboard, and a chemical engineer can design a production plant all using only CAD software. But we are simply not at that point when it comes to large swaths of design and implementation in biology. Using math in biological prediction through equations would help accelerate us toward a future without animal products. We could precisely design productive bioprocesses churning out alternative meats and alternative foods (Chapter 3). Also consider if we had equations for the structure of the food and personalized nutrition (Chapters 6). And equations could also reduce the development time toward new, exciting foods (Chapter 8). More ambitiously, could we predict the future without animal products completely? Could we make exact pronouncements of when and which products displace animal-based ones? Uwe and his lab are pioneers in **metabolomics**, measuring the chemical molecules that make up an organism’s metabolism. Immediately before I joined Uwe’s lab in Zurich, the lab had just developed a new technology, real-time metabolomics.[^527] With real-time metabolomics, one could directly inject live bacteria into a mass spectrometer and blow them into pieces within. The mass spectrometer then separates and resolves the bacteria into its constituent metabolic molecules, including amino acids used to form protein, as well as sugars used for energy and the formation of other cellular components. A single reading on the mass spectrometer would roughly measure how many of these metabolic molecules were inside of the bacteria at a given time. We could cultivate bacteria in vessels of a liquid nutritional medium, and every ten seconds run the mass spectrometer operation. Thereby, we could generate a time profile of how these different molecules changed over time. Uwe figured we could use the technology to study how bacteria “decide” to divide and tasked me accordingly. He theorized that the real-time metabolomics technology would make the problem more tractable. By observing how the counts of these different metabolite molecules changed, we could conceivably find the metabolite molecule within that triggered division once reaching a particular level. To simplify the research, Uwe and I focused on bacteria that are starved for food and thereby unable to divide. Starved cells were technically conducive to our study because, with a low metabolic background, the metabolic change is more pronounced using our mass spectrometry measurement. We performed real-time metabolomics as these starved bacteria received drops of their preferred nourishment: sugar water. Every drop of sugar elicited a beautiful, pronounced peak in many different metabolic molecules (**Figure 21**). We further observed how the sugar traversed the bacterium’s metabolism; in effect, the chemical conversion of sugar into other metabolite molecules, and eventually into larger building blocks (e.g. protein and DNA) constituting the bacterium. Large macromolecules such as protein and DNA are often lumped together and collectively termed the **biomass** of the cell. ![Figure 21](/images/book/Figure21_rt_mets.jpg) **Figure 21. Example of real-time metabolomics data.** Drops of sugar were fed to starved bacteria (gray bars). The black line indicates the amount of glutamine (an amino acid used to make protein) in the bacteria’s metabolism. The sugar feeding induced a spike in the bacteria’s level of glutamine. The consequent fall of the glutamine suggested that the bacteria make protein by rapidly consuming the glutamine. Uwe and I then hypothesized that the sugar rapidly formed into new biomass, in fact, faster than the ten-second window of our measurement. Follow-up experiments confirmed our suspicions, and the pulsed sugar drops were indeed formed into new biomass of protein and DNA. We then reasoned that division was actually not determined by metabolic level, but instead by biomass level. Eventually, through some parallel experiments, we pinpointed bacterial division to a specific entity within the biomass, a specialized protein, FtsZ. FtsZ comprises the physical ring within the bacterium, affixed to the inner surface. When bacteria divide, this ring constricts, thus squeezing the mother cell into the two daughter cells. Our study highlighted FtsZ as the primary decision point for division; that is, when the cell has enough FtsZ, division commences. With our data and this insight, we were able to develop an equation that predicted when bacteria divided as a function of sugar input. After a scientific review process spanning over a year, the work was finally published.[^528] These two anecdotes might suggest that all events are predictable given the right equation or knowledge, even in a domain with millions of variables such as biology. We simply need to find the right experiment or situation where we can focus on the few variables that matter, and consequently, we can more easily conjecture a model. But as with any explanation, we must always ask if there is an alternative, better explanation. In this case, perhaps coral reef spawning and bacterial division are simply two special cases that are amenable to predictability. Perhaps their predictability actually serves an important biological function. Predictability as a selection parameter makes sense when you consider another explanation: when the coral release gametes all at once, they maximize their reproductive potential. The simultaneous explosion of gametes floods the environment so that predators are unable to devour everything. If gametes were released piecemeal, then it would be easy picking for predators, and the coral would fail to reproduce. Therefore, predictable coral spawning carries a reproductive advantage and befits evolutionary selection. Presumably, corals whose gamete-release time varied from the synchronized window would readily be selected out of the population as predators would more easily consume their progeny. Therefore, evolution is the forcing function for concerted spawning and explains the predictability here. For bacterial division, I can’t explain as definitively why the quantity of FtsZ drives the division decision. I suspect that FtsZ serves as a proxy indicator for the cell to know how much biomass it has. When the mother cell splits into the two daughter cells, it will all be for naught if the daughters’ cells aren’t viable. Accordingly, the mother must have sufficient biomass to divvy into two daughters. For example, each daughter cell will need a least one whole copy of the DNA. Each daughter cell will also need some cellular machinery in order to decode and actualize the DNA. Each daughter cell will additionally need requisite amounts of membrane material to encapsulate its contents. Therefore, the mother cell must know when it has enough biomass to grant to each daughter cell. Counting each individual molecule needed seems rather costly, especially given the constraints of bacterial biology. Instead, FtsZ forms more slowly compared to other parts of the cell’s biomass, and therefore, it serves as a suitable bellwether for indicating the time to divide. In this scenario, FtsZ is one of the last biomass components to make it to the sufficiency finish line. ## Struggling to Predict In the case of biology, when timing serves an evolutionary function, we should be able to find ways to predict occurrences with enough investigation. But the ultimate question still looms: are there limits to what else we can predict generally? Could I predict which products will replace animal ones and invest my money accordingly? When I started my scientific career, I thought that everything would be predictable given more knowledge. I believed that we could eventually simulate everything with perfect fidelity, only limited by the performance of our computers. Indeed, some events are readily predictable after we perform the key measurements of the determinants. Astronomical events with celestial objects occur like clockwork. Hotel bookings were sold years before The Great American Eclipse of 2017.[^529] We know that Halley’s comet will appear in our sky every seventy-four to seventy-nine years. However, the case of weather is more vexing when it comes to our confidence in our prediction ability. We actually have the equations we can use already. Weather is highly defined by the movement of air and water, calculable using the Navier-Stokes equations. The Navier-Stokes equations are scripturally elegant: ![](/images/book/new_art_Navier-Stokes_equation.jpg) These equations predict how fluid flows at large scales. Being able to solve for these equations would conceivably detail the path and progression of fluid flow. However, the equations are notoriously difficult to use, and solving them has proved insuperable except for simple cases. In 2000, The Clay Mathematics Institute offered a million-dollar prize for proving just the *existence* of unique, smooth analytical solutions to these equations.[^530] So far, there have been no awardees. An idealized, analytical Navier-Stokes solution would take the form of a function *u*(*x, y, z, t*) = *\*, where *u* is the speed of the flow at a given position, defined by the spatial *x*, *y*, and *z* coordinates at a certain time *t*. With a solution, we would be able to calculate the speed of a fluid at any time and position. Even with the lack of analytical solutions, the Navier-Stokes equations are not completely inaccessible to us. We can use them numerically; so, say, start with a value (e.g., the speed of a fluid) and calculate the predicted speed after a short time. We can perform this process iteratively: after we get the speed at the next time, we can plug it in again to get the speed at the time point after that. Seemingly, we could perform calculation *ad infinitum*. However, numerical solutions always bear some error because we’re using numbers with a finite precision, i.e., ceasing the numbers after the decimal place. The error only amplifies with each further iteration. We intuit that if our error after five iterations is ten percent, then after five more iterations, the values will have error of more than ten percent, perhaps even twenty percent. Indeed, the error can only get worse: as we continue to forecast our calculations, we eventuate to an unbounded solution, where the error becomes so large that the calculated values become meaningless. Numerical solutions always carry this disadvantage, though they are still functionally useful within those limits. The tendency to blow up—reach unboundedness—is compounded with the Navier-Stokes equations because they are **chaotic.** Chaotic systems and equations are extremely sensitive to initial conditions. I highlighted that the ideal solution would take the form *u*(*x, y, z, t*) = \. Expounding further, a solution would depend on the initial conditions, i.e., how fast the fluid is moving at the start. For instance, in the case of fluid flow, consider the situation where a gust of wind travels into a valley. In the middle of the valley, a tall rock blocks the path, and so the wind flows around it. A simple schematic is shown in **Figure 22A**, where the valley is the light gray area, the wind is the arrows, and the rock is the white circle in the middle. The wind requires about five minutes to reach full speed. In the first minute, the wind is imperceptible, but at around 1.5 minutes, the wind picks up, and by three minutes in, the wind is nearly at full speed. Visually, I depict such changing wind with a graph (**Figure 22B**). The wind speed rapidly ramps between the 1.5- and 3-minute period. ![Figure 22](/images/book/Figure22_valleyplotandramp.jpg) **Figure 22. An example for predicting fluid flow.** (**A**) A depiction of a valley with a rock (indicated by the white circle). Wind flows from the west side heading eastward. (**B**) The profile of the wind speed over five minutes. Initially the wind is modest, but quickly ramps to maximum speed between 1.5 to 3 minutes. The wind cannot flow through the rock; it must flow around it. As a result, circulating eddies are formed where anything traveling along with the wind will rotate. Imagine that we’re riding a leaf carried by the same wind. Our leaf skirts past the right side of the rock just barely. The wind is diverted from impact with the rock and thereby creates a rotation at specific points in the valley. Our leaf now spins leftward and as we move past the rock, we now look back at it before being rotated back looking into the valley. This local rotational effect from fluid flow is familiar to anyone who has kayaked. We know that as we ride past a rock in a flowing stream that our kayak will rotate towards the slipstream behind the rock without any intervention; for example, if we pass on the left, the bow of the kayak rotates right. This local rotation is calculable using the Navier-Stokes equations and formally termed **vorticity**. I have calculated the vorticity at 4.5 minutes into the start of the blowing wind and visualized such in **Figure 23**. Indeed, we can see the eddies. The white regions indicate high local rotation turning counterclockwise, and the black regions indicate clockwise rotation. If our leaf was in the black region, the wind would be rotating us clockwise. We also notice that the vorticity zigzags further out beautifully, eventually dissipating. Therefore, if our leaf was further away from the rock, there is less vorticity and, therefore, less applied rotation from the wind. ![Figure 23](/images/book/Figure23_vorticity_base.jpg) **Figure 23. The vorticity of the wind moving past the rock.** The intensity of vorticity is indicated by the coloring where black indicates turning clockwise and the white indicates turning counterclockwise. Vorticity drastically affects our weather, as the local spinning can help lift (advect) moist air into our atmosphere.[^531] If advection seems unintuitive, consider a tornado, which picks up material along its path. The dirt is kicked up and moves upward as the tornado spins. With weather, when enough moist air is concentrated, the clouds will precipitate by releasing water in the form of rain, snow, sleet, or hail. Therefore, we are not able to predict whether it will rain without accounting for vorticity. However, as I alluded to earlier, the calculations are sensitive to how the situation starts due to the chaotic properties of the Navier-Stokes equations. To illustrate, consider if we slightly changed the ramping of the wind ever so slightly in **Figure 24A**. I am not changing the speed, just adjusting the time when the wind starts to pick up by half a second (New Case 1) or almost a full second (New Case 2). In fact, the change is so imperceptible that we need to zoom in significantly to perceive the difference as the wind speeds ramp up. However, at three minutes in, all three cases are indistinguishable, as observable by the same zoom level at that point. So how does this affect the consequent vorticity? When looking at the same time point of 4.5 minutes, the vorticities indeed look quite different (**Figure 24B**). We see that New Case 1 and 2 look to be much further along compared to the original Base Case. The black region in the Base Case seems to be just budding off, whereas, in the new cases, this budded region is further along. Just to emphasize, at 4.5 minutes, the blowing wind from all three cases should be indistinguishable because it was so at three minutes. Furthermore, we only shifted the start of the ramp by about a half second for each case. Fluid parameters such as vorticity are indeed highly sensitive to the starting conditions. ![Figure 24](/images/book/Figure24_all_vort.jpg) **Figure 24. Consequences of changing the ramp start time.** (**A**) Two additional cases with earlier ramp times were added (New Case 1=gray line, New Case 2=dashed line). The change in ramp time is slight—about 0.5 seconds for each new case, as depicted within the zoom in. By three minutes, the profiles are seemingly identical, even with the same zoom as the first. (**B**) Change in vorticity for the new cases. The slight change in ramp time elicits a new vorticity profile at 4.5 minutes. Notice how most weather predictions never proceed beyond a couple of weeks? That’s because numerical solutions of the Navier-Stokes equations underlie such predictions, and as we’ve just seen, they can be difficult to precisely use. Certainly, vorticity can also be affected by the rotation of the Earth or seasonal effects; therefore, a measly wind may be inconsequential if it is the dry season in India. But consider a season where there is some possibility for rain, such as in Durham, North Carolina during the spring. Effects from gusts of wind may prove the deciding factor, and our calculations are precarious enough that rain is forecast in probabilities: “A thirty percent chance of rain.” The chaotic nature of the Navier-Stokes even limits a potential analytical solution, too. Suppose that a special mathematician comes along, solves the equations, and claims his or her million dollars from the Clay Institute. With the solved equations, we would still require a perfectly precise measurement of the starting conditions (like the ramping of the wind). Errors in this measurement will only propagate the further that we extend our solution. We saw how divergent the solutions became after just a couple of minutes. Consider hours, days, or weeks later. Clearly, we cannot assure a clear day for, say, that outdoor wedding six months out that we want to plan. Even if the mathematics of the Navier-Stokes equations was perfectly manageable, it’s still an approximation to the underlying phenomenon. We know that water is comprised of individual molecules that coalesce into fluid. The equations do not account for this molecular resolution. There is an inherent limit to where they work. This is okay for predictions of weather because knowing all the details of the individual molecules is unnecessary for making predictions. We can **abstract** away those features, and we do not have to account for the random motions of particulate in the water nor the vapor cavities collapsing and inducing shockwaves in the proximal region of the flow. While these phenomena undoubtedly influence the fluid flow at microscopic levels, we ignore them for larger-scale predictions. Should we desire to study the physical world more granularly, we require new models to explain these scales. All known matter is comprised of molecules; molecules themselves are arrangements of atoms. Zooming in further, atoms are defined by elementary particles including electrons and neutrons. Photons, another elementary particle, form light, similar to how water molecules form bulk fluid flow. Elementary particles can exhibit different, discrete states, for example having a defined location or speed. Curiously, the elementary particle seems more like a cloud or a wave rather than a single point before we try to measure a property such as speed or position. There is an ensemble of different property values. Mathematically, this cloud is termed a **wave function**. Upon “measurement,” this wave function collapses, and a single point remains with a defined position value. The measurement can be repeated, drawing different values, but is bracketed by the wave function. This variance in different values has been famously termed the Heisenberg Uncertainty Principle. ## Multiverse Theory This peculiarity of quantum physics has perplexed scientists for years, and a range of explanations has ensued. The instrumentalist interpretation sees the wave function collapse as purely mathematical and divorced from physical reality. The pervasive, timid Copenhagen interpretation shirks away from complete physical explanation in that it does not try to say what the wave function physically is. According to the Copenhagen interpretation, the photon only has a position or speed *after* measurement; before that, one should throw one’s hands up in the air. Both lines of thinking are highly problematic. If we’re to be consistent with our use of scientific knowledge, we have to look for a physical interpretation of before and after the wave function collapse. We cannot excuse for inconvenience, and we must boldly seek to find such an interpretation. In 1957, physics graduate student Hugh Everett resolved the interpretation issue of the wave function collapse and proposed the many-worlds interpretation (**Multiverse theory**) of quantum physics. According to Everett’s theory, expounded on later by physicist David Deutsch, the cloudy elementary particle is a real physical ensemble of each electron, proton, photon, etc., in the different states enumerated by the wave function, i.e., having different positions and speeds according to the mathematical distribution. The wave function is *physically* real. Within the ensemble, each different state is fundamentally **fungible** with each other. It is like having two one-dollar bills in your wallet, where one is smooth and another is crumpled. If you’re buying a pack of gum, you may reach into your wallet and either of the bills will satisfy the payment. Therefore, in the context of the payment, both bills are fungible with one another. Similarly, the particles, while having different states, are fungible to the eventual measurement. Once the measurement occurs, one of the fungible particles within the ensemble is chosen and represents the collapse of the wave function. Measurement here means that the particle is **entangled** with the current reality, meaning that other particles are affected by the state of the particle. In the money analogy, once the bill has been handed over and placed into the cash register it affects whatever is around it. The many-world interpretation takes this a step further, positing that innumerable realities are actually branching from one another. Each fungible state of the particle is entangled with a different reality. To return to our money analogy, it’s as if the money transfer is the source point for two timelines: one where the gum was paid for with the smooth bill, and another paid for with the crumpled. Both timelines occur and are indeed real but are inaccessible to one another. In the same way, every time a particle is entangled into a reality, the complement reality occurs as well. There is another reality where the other state of the particle is chosen and entangled to create an altogether separate timeline. Seemingly the best explanation for quantum wave phenomena, the Multiverse theory’s implications are astounding. The branching occurs so often that it’s effectively infinite. Our lives are no longer defined by a single narrative. There may be timelines where we marry completely different people, die young or old, don’t exist at all, etc. The collective aggregate of all of these timelines and histories is termed the Multiverse. To reify the potential consequence of such quantum variance and branching, take for example light, which is comprised of the elementary particle, photons. The photons emitted from the sun directly land and penetrate to some degree the epithelial cells of our skin. So let’s imagine Stella, a hypothetical gal, sunbathing on the beach where she is bombarded with radiating light photons all day. In one part of the Multiverse, a photon, in the corresponding state of position and speed, is able to induce a lesion in the DNA of Stella’s skin.[^532] This lesion in Stella’s DNA facilitates the genesis and progression of skin cancer. In the other part (and likely by far the majority of causalities) of the Multiverse, Stella sunbathes without incident. Consider ramifications of Stella getting the cancer or not. Suppose she was on the path to develop the killer application of the alternative food movement. Her getting cancer set back the movement a year or two and cedes market placement to another researcher and company. The Stella example is imaginably exceptional, as it’s one photon inducing significant, propagating changes in the affected timelines; however, it effectively illustrates the potential for continuous divergence. Single elementary particles can also induce electronics failures,[^533] but for the most part, the elementary particles only create variances in aggregate. So, more than a million subatomic particles make up a single FtsZ protein, and it takes two thousand to divide a cell. Therefore, it’s hard to say exactly what phenomena are subject to the variance wrought by subatomic particles collapsing into different states. I suspect something as large as Halley’s comet is mostly buffered against quantum effects. The momentum of impinging light or particulate is unlikely to steer the comet off track substantially, no matter which part of the Multiverse we’re in. We’ll be able to predict the entry of Halley’s comet in our night’s sky with high confidence. Going back to our case of predicting weather; as discussed, the Navier-Stokes equations are chaotic and subject to wildly different outcomes with even the slightest changes in the starting point, as highlighted in the vorticity example earlier. David Deutsch asserts in *The Fabric of Reality* that this flow of fluid is subject to these variances resulting from the Multiverse branching.[^534] There is no such thing as a perfect measurement of wind ramp-speed due to quantum fluctuations. Furthermore, the quantum variance within fluid flow amplifies due to the chaotic effects. More pointedly, there are defined limits to weather predictability: weather will always vary in different parts of the Multiverse due to the marginal starting differences. We *cannot physically* predict weather completely. So, it seems that there are clearly phenomena that we can predict (e.g., arrival of Halley’s comet) and phenomena that we can’t (e.g., whether rain occurs on the same day next year, how and when the alternative food revolution takes root). Predictability is graded and lies on a spectrum. There are limits to the fidelity of prediction. Particularly, I’m fascinated with our minds in this regard. Our thought processes are the outgrowth of a network of elongated cells, neurons. Neurons pass information to each other through electrical current in the form of action potentials. When neurons are excited enough—when an electrical voltage induces a strong enough change, the action potential occurs, and downstream neurons can also be activated. Action potentials are “all or nothing.” The neurons are designed in a way that the voltage must cross a specific threshold in order for the action potential to occur. If the voltage remains below the threshold, the activation will not occur. Seemingly, this mechanism could be susceptible to quantum variance, and I’m not the first to suggest this.[^535] Therefore, in different parts of the Multiverse, our thinking could be different for the same situation. The words I type out now are surely different compared to the words I type in another part of the Multiverse. That is clearly consequential and should steer how our world unfolds. ## Knowledge and Prosperity We can’t discuss the science of predictability without the most obvious application: the stock market. The promise of financial windfall entices smart, intrepid individuals to predict the future of companies, industries, and commodities. Hedge funds and managed funds pool investment from a variety of sources: institutions, rich investors, and retail investors. Such funds are generally run by a team who will charge the participating investors a flat one to three percent of the total investment every year. This charge compensates the team for their expert prognosticating efforts derived from all their research activity and predictive models. Alternatively, one can invest in index funds: an index fund collectively invests in companies by some set parameter. For example, investing in the SPY index fund pegs the money to the performance of the Standard & Poor’s 500, the 500 largest companies in the United States. No expert customizes the fund allocation. The SPY is investing directly into the biggest 500 companies, that’s it. One can even just directly assert that the entire stock market will succeed and invest in index funds such as VTSAX—a total stock market fund offered by Vanguard, a consumer-focused investment firm. Index funds do not charge the exorbitant fees of a managed fund because investors are not paying anyone for anything other than administration, so typically the cost is a mere fraction of a managed fund. So, which is better for an average retail investor like me? Index funds by a wide margin: in the last twenty years, only around five to eight percent of actively managed funds beat the SPY.[^536] So, it would seem that trying to predict winners in the market is more akin to trying to predict the weather long-term than the orbit of Halley’s comet. There are probably many Stella situations where consequential situations unfold differently in various parts of the Multiverse, leading to different economic outcomes. And I suspect that human thinking is susceptible to the branching from all these collapsing waves. It seems best for retail investors to avoid the managed funds, save on the high fees, and place their money in index funds. Curiously, we’re *still* better off investing the money in the stock market, whether index or managed, than putting it in a savings account or other “safer” strategies. The S&P 500 itself has averaged about ten percent growth per year since 1926.[^537] Major hedge funds average at around eight percent growth[^538] in recent years, and the US economy averages around five to six percent. Certainly, these numbers are not adjusted for inflation—our money devalues by about two to three percent per year—but nonetheless, our money is better invested than sitting under our mattress. This growth must be coming from somewhere. The companies are not always vampires sucking the wealth from people. In order for that to be true, everyone else’s lives would have to be getting much worse over time, and as discussed in the first chapter, that just isn’t true, globally speaking. The assertion that wealth is merely extracted from the planet is also untenable, as the US has been consuming fewer total resources with a growing population, in certain sectors.[^539] Instead, this growth ultimately comes from the net creation of wealth, particularly in the form of new knowledge. I’m not the first to make such a point. Nassim Nicholas Taleb makes a similar assertion in his book *Black Swan*: market economies have more inherent opportunities for serendipities—wondrous new findings that spur cascade effects.[^540] Yuval Harari Noah in *Sapiens* asserts that the growth of our economies can be best explained by the advancement of scientific knowledge.[^541] When we have scientific techniques that reduce the overall magnitude of problems, we grow the market. For example, the impactful Haber process allows us to derive fertilizer from the air. The process, developed in the early 20th century, solved the biggest agriculture problem at the time, supplying enough nitrogen to the plants. In fact, the Haber process has been so consequential that an estimated 2 billion people or so may not have existed without the technology.[^542] And again, the Haber process is actually making fertilizer out of air. The cost is not exactly zero—one needs a catalyst, reactor, and the proper temperature/pressure, but clearly, we can derive much benefit for meager input. The Haber process then has ripple effects. Fewer people need to be farming, and we can accordingly divert resources to other problems. We notice a similar effect with the eradication of smallpox. The poverty of a population is reduced by no longer having to put resources into fighting this debilitating disease. Altogether, as we solve problems using knowledge, our prosperity as a modern civilization should increase. A large swath of modern, prosperity-inducing inventions have come from the industrialized West (e.g., the Americas and Europe), including innovations in health, agriculture, and computing. Admittedly, the West has more resources to put toward such problems; but, how did the West get these ample resources in the first place? In this, we have to recognize a valuable difference between liberal, free-market societies versus authoritarian ones (e.g., Soviet-style communism, old monarchies and tyrannies). Author Yuval Noah Harari also discusses the difference in both *Sapiens* and *Homo Deus*, analogizing the economy in terms of computers, that is, a market-based economy has more “computers” running calculations. These computers are the different companies in one industry, and their calculation is each one’s specific business model. Having a variety of different business models to try means that the best ones will rise, and the subpar ones will die out. In contrast, an authoritarian-run economy runs only one computer or a few, as determined by the state. This computer is stuck running one business model for the given industry, and that business model is not stress-tested against others. As a result, a better business model will not be found because of the lack of recourse and competition. But is that the only explanation? There’s a country with ostensible communist leanings that seems to buck this trend: China. From 1979 to 2010, China’s entire economy has averaged ten percent growth per year; it’s like the entire country is the S&P 500.[^543] So why does the Chinese economy seemingly perform counter to our intuitions? Certainly, China has liberalized over the years by, for example, allowing private companies to supplant the state-owned enterprises, but I think there are additional explanations. Our classic sense of a lumbering, inefficient socialist government brings to mind the Soviet Union. China’s government and economy are structured differently. The Soviet Union directed and micromanaged from the top whereas China’s government is more confederated.[^544] Municipal governments have wide leeway to try things differently compared to other municipalities. China, in effect, runs a lot of computer calculations, but the calculations occur locally within the individual municipalities. Secondly, China invests heavily in infrastructure. Sticking with the computer analogy, the infrastructure is the computer hardware that enables more calculations. Roads and bridges facilitate the transport of goods, thus permitting the industry within to burgeon. Public transit enables easier commuting and thus creates more opportunities for workers. Schools and hospitals engender a healthy, educated populace who can pursue opportunities more easily and spur more innovations. Infrastructure as a stimulus only works to a point. After a certain number of roads and hospitals, there is diminishing marginal value in adding more. Beyond that point, other forces must take root to facilitate further growth. It’ll be interesting to see how China navigates the post-infrastructure investment binge. Finally, China can reverse decisions and policies easily even compared to a dynamic society such as the US. Consider the 2020 coronavirus pandemic. China initially and irresponsibly played down notions of a brewing epidemic to its own citizens and the world at large.[^545] Eventually, China’s central government wholly acknowledged the situation and imposed restrictions on the movement of citizens, helping contain the spread of the virus.[^546] And it worked. In contrast, the American federated system allowed each state to pursue its own action toward the pandemic, each with varying responses of shelter-in-place, closing schools, vaccination, and reopening businesses. The result became a disaster as the United States had higher rates of coronavirus cases and deaths than most of the world.[^547] Stamping out a pandemic requires strong coordination which is much easier with top-down control.[^548] Solving policy issues such as pandemics does not suit an American model of governance. Understanding and refuting bad policies promotes progress. Normally, in a liberal democracy, as in the West, a referendum on policy comes through the ouster of politicians via an election, the bedrock of democracy. Such political party turnover doesn’t happen in China, and for most authoritarian governments, this would normally be a huge problem. The lack of punishment for political failures has ensured tepid progress in politically repressive governments such as Russia, Zimbabwe, and Iran, whose leaders cannot be ousted so easily and, therefore, have less reason to falsify their actions, even though they have the power to do so readily. In contrast, Chinese government officials are held accountable, just not via elections. In rural areas, officials are held accountable through membership in their local temple and lineage groups.[^549] In order to maintain high member standing, the official must perform activities for the community such as building roads and schools. Moreover, urban officials will be dismissed, sometimes jailed, upon the revelation of corruption or aberrant behavior. This accountability ultimately results in the willingness to acknowledge and revoke bad policies. Officials have phased out the One Child Policy, which famously only allowed one child per family. China has also taken strides to address some of the corruption issues raised during the Tiananmen Square protests; it developed some of the most comprehensive anti-corruption laws to date.[^550] And most recently, China is leading worldwide diplomatic efforts for CO2 emission reduction after years of heavy industrial pollution.[^551] All in all, having more consolidated control can be an advantage for ridding bad policies or proceeding in the best direction. *Being able to move on from clearly inferior ideas creates innovation.* But there is a caveat. When we repudiate an idea or policy, a new one (including no policy) takes its place. This process allows people to propose potentially superior ideas. Here, the West has a much stronger advantage given the expansive, impressive free speech protections, where citizens are not punished for critiquing governments, businesses, or the actions of others. Therefore, I sincerely chide China for its strict, limiting speech laws that likely muffle the creativity of their populace. Western-style free speech protection ensures that the best ideas can be generated and that inferior ones can be weeded out. Furthermore, amazement at China’s ability to innovate does not excuse their poor human rights record. I severely condemn the Uighur re-education camps and jailing of protestors. These are immoral and create more problems ultimately. In sum, I agree that scientific advancement and serendipitous findings underlie the growth of our economies and the total reduction of problems. For example, replacing animal technology would solve many problems in terms of nutrition, environment, suffering, taste, and cost as we discuss throughout the book. However, we can go one step further. I assert that scientific advancement equates to the repudiation of bad ideas while generating new ones. Therefore, if we employ more of this process—knowledge generation—within our society, our economy should grow. Circling back to index funds: index funds allow one to invest in entire markets. It doesn’t matter which company or academic institution generates the next Haber process. The ripple effects create a new positive, and the overall market efficiency grows, thereby creating new wealth and increasing the value of investments. Therefore, as long as humanity generates knowledge, I see the economic output growing and the value of investments rising. This is the best reason that I see to invest in index funds. The knowledge that confers efficiencies throughout the marketplace can be reaped via index funds without picking the specific winners (the next Apple or Tesla). Furthermore, we also do not have to be in the right part of the Multiverse to realize a profit. I assert that all parts of the Multiverse from the current branch point will continue to generate knowledge. I will happily continue to invest in markets where I suspect that untrammeled knowledge generation can occur. I do not have too much hope for countries such as Iran, which suppresses free speech and its female population’s ability to add to its innovation capacity. In general, theocracies are unable to innovate as readily as more open societies.[^552] By definition, religious doctrine is inviolate and does not allow itself to be falsified no matter how bad the idea, whether it’s mutilating genitalia or condoning slavery. Furthermore, theocracies and religion do not reward the generation of new ideas to replace old ones, but rather demand complete obedience to established ones. ## Prediction of Technological Trends With a more precise idea of how innovation works and how our societies pursue breakthroughs, we can now make a key prediction: innovations will happen under the right circumstances (e.g., free speech, academic research, repudiation of bad ideas) and will be also driven by needs (e.g.. better transportation capability). The causalities of innovation will always be the less efficient and dragging technologies. For example, the development of modern thermodynamics by academic scientists led to combustion engines.[^553] As a result, motorized vehicles were developed, and the use of horse-drawn carriages waned. We see similar innovation and needs driving superannuated technology in other areas: going to a video rental store is a nuisance when we can sit on our couch and rent with clicks on our remote, thanks to streaming capacities of the internet. Light-emitting diode bulbs are more energy-efficient and cooler than old school incandescent. And online services (e.g., filing taxes) tend to overtake in-person or snail-mail–based ways. Less saliently, we might not think about cows as a technology, but we certainly use them as such. They are used as a reactor that consumes grass and generates meat and dairy. If we take this idea further, the notion of growing a bag of meat that walks around and takes four years to mature sounds pathetic. We could develop hardy, local bioreactors that perpetually supply enough “meat” to feed a village. We could have 3D printers in each of our homes that are programmed to our specification and “print” out a meat patty on demand, perfectly catered to our taste and nutrition at that moment. We could have a bag of protein powder that, when poured into water, self-assembles into a steak. Even before any of those futures, in the grocery stores we should have non-animal meat in every way superior to animal protein: cheaper, more nutritious, and better tasting. This should be possible because animals are such crummy technology. Unfortunately, we have not innovated enough yet, but we will. And I posit that *knowledge generation is the key to rendering animal technology obsolete*. With the right knowledge, we will have no need for animals as a technology. Chapters 2 through 5 are about making that case and showing that animals are ripe for replacement in terms of the biological physics, our ability to improve on them, and the constraints of food processes. Therefore, provided that our capacity for innovation continues, we will render animal products obsolete. Innovation is replacing poorer technology with better technology. The future becomes a present without animal products. I will not predict when. In fact, I defer to the Multiverse theory as the best knowledge for explaining the wave function collapse. Therefore, I take its claims seriously in that our universe is constantly branching, and there are many instances of myself typing this same sentence. Therefore, if I try to predict when, then I may appear brilliant in one part of the Multiverse and silly in another. I also cannot predict how. The nature of knowledge generation is inherently unpredictable. Predicting future knowledge is tautological, for, if one could perfectly predict new knowledge, then we would have that knowledge. Given that I theorize that we’ll innovate ourselves away from using animal products, the replacement may look completely different across the branches of the Multiverse. In one part, the personalized 3D printers may dominate, and in another, drones may be flying in our steak from a local bioreactor. I can only say that if we continue to innovate and produce new knowledge, animals will eventually be replaced because they are nakedly terrible technology. ## Chapter Terms - **metabolomics:** measuring large swaths of an organism’s metabolism - **biomass:** the physical mass of a biological organism - **chaotic:** a property of a system where future states are highly sensitive to the starting conditions of the initial values of the variables - **vorticity:** the localized, imparted net rotation of a fluid. Imagine a flower petal in a stream of water and notice how it turns at various points of the stream without colliding with any objects. This rotation stems from the vorticity of the fluid. - **abstraction**: purposefully discounting details that are inconsequential or unnecessary when explaining a phenomenon - **fungible**: interchangeability of a group of objects and entities. Most common example is money. If I pay taxes but don’t want it to fund the military, I’m out of luck. The taxes go into a common pool, from which the military draws its funding. - **wave function**: a fungible state of subatomic particles exhibiting different states (in particular, speed and position). Upon measurement (entanglement), one of the particles is chosen. - **Multiverse:** the total ensemble of all of the different timelines created by the wave function collapse. This can mean realities where Hillary Clinton was elected in 2016, where Soviet Officer Vasili Arkhipov was not able to avert nuclear war in 1962, or where Genghis Khan fell off his horse and died before expanding the Mongol empire. - **entangle**: after a wave function collapse, the particle in question interacts with other particles, setting in consequences ## Chapter Summary We can predict some events with appreciable fidelity (e.g., the orbits of Halley’s comet), but other events are fundamentally unpredictable (e.g., the weather a year from today or which companies do well over the next twenty years) due to living within the Multiverse, constraints on physical laws (e.g., fluid flow), and the nature of knowledge generation. The Multiverse theory, currently the best explanation for quantum effects, suggests that all events are constantly branching through advancing time. There are realities where we do not exist, and the idea of a defined narrative is fallacious. Despite this inability to precisely pinpoint the future, certain trends are predictable from the generation of knowledge, including that poorer knowledge and technology will be replaced. This trend continually creates wealth and means that we can reap rewards from just investing into knowledge-generating economies. Animals are a susceptible technology and long due for replacement. [^525]: I don’t remember the exact date here; however, I remember that it was a 1 to 1.5 hour window of a particular day. [^526]: Coral Reproduction. Australian Government – Great Barrier Reef Marine Park Authority. (Accessed December 26, 2019). [^527]: Link, H., Fuhrer, T., Gerosa, L., Zamboni, N., & Sauer, U. (2015). Real-time metabolome profiling of the metabolic switch between starvation and growth. Nature Methods, 12(11), 1091–1097. . [^528]: Sekar, K., Rusconi, R., Sauls, J. T., Fuhrer, T., Noor, E., Nguyen, J., … Sauer, U. (2018). Synthesis and degradation of FtsZ quantitatively predict the first cell division in starved bacteria. Molecular Systems Biology, 14(11), 8623. . [^529]: Rao, J. (2016). Want to See the 2017 Solar Eclipse? Better Book Your Hotel Room Now. [Space.com](http://Space.com). (Accessed August 20, 2020). [^530]: Navier–Stokes Equation. Clay Mathematics Institute. (Accessed December 26, 2019). [^531]: Haby, J. Vorticity Basics. National Weather Service (Accessed December 26, 2019). [^532]: Pyrimidine Dimer. (2019). Wikipedia. (Accessed December 26, 2019). [^533]: Somers, J. (2018). The Friendship That Made Google Huge. The New Yorker. (Accessed December 26, 2019). [^534]: Deutsch, D. (1998). The Fabric of Reality: The Science of Parallel Universes--and Its Implications. Penguin Books. [^535]: Penrose, R. (1996). Shadows of the Mind: A Search for the Missing Science of Consciousness (Reprint edition). Oxford University Press. [^536]: Hulbert, M. (2017). This Is How Many Fund Managers Actually Beat Index Funds. MarketWatch. (Accessed December 26, 2019). [^537]: Royal, J & O’Shea, A. (2018). What Is the Average Stock Market Return? NerdWallet. (Accessed December 26, 2019). [^538]: Harper, D. R. (2019). Hedge Funds: Higher Returns Or Just High Fees? Investopedia. (December 26, 2019). [^539]: McAfee, A. (2019). More from Less: The Surprising Story of How We Learned to Prosper Using Fewer Resources—and What Happens Next (Illustrated edition). Scribner. [^540]: Taleb, N. N. (2010). The Black Swan: Second Edition: The Impact of the Highly Improbable: With a new section: “On Robustness and Fragility” (2nd ed. edition). Random House Trade Paperbacks. [^541]: Harari, Y. N. (2017). Homo Deus: A Brief History of Tomorrow (Illustrated edition). Harper. [^542]: Smil, V. (1999). Detonator of the population explosion. Nature, 400(6743), 415–415. . [^543]: Historical GDP of China. (2020). Wikipedia. (Accessed August 19, 2020). [^544]: Kroeber, A. R. (2016). China’s Economy: What Everyone Needs to Know® (1st edition). Oxford University Press. [^545]: Hanson, V. D. (2020). China Boomeranging. National Review. (Accessed August 19, 2020). [^546]: Graham-Harrison, E. & Kuo, L. (2020). China’s Coronavirus Lockdown Strategy: Brutal but Effective | World News | The Guardian. (Accessed August 19, 2020). [^547]: COVID-19 Pandemic in the United States. (2020). Wikipedia. (Accessed August 19, 2020). [^548]: Holtz, D., Zhao, M., Benzell, S. G., Cao, C. Y., Rahimian, M. A., Yang, J., … Aral, S. (2020). Interdependence and the cost of uncoordinated responses to COVID-19. Proceedings of the National Academy of Sciences, 202009522. . [^549]: Chen, D. (2015). How China Holds Politicians Accountable. TODAYonline. (Accessed December 26, 2019). [^550]: Xu, H., Wu S., & Zheng, E. Bribery & Corruption 2021 | China. GLI - Global Legal Insights. (Accessed August 19, 2020). [^551]: Worland, J. (2017). It Didn't Take Long for China to Fill America's Shoes on Climate Change. Time. (Accessed August 19, 2020). [^552]: Ruck, D. J., Bentley, R. A., & Lawson, D. J. (2018). Religious change preceded economic change in the 20th century. Science Advances, 4(7), 1–8. . [^553]: Smil, V. (2005). Creating the Twentieth Century: Technical Innovations of 1867-1914 and Their Lasting Impact (Illustrated edition). Oxford University Press.