Nosimpler

“But justice roll down like waters and righteousness like an ever-flowing stream” Amos 5:24 https://t.co/o89PSY1YBd

— James Comey (@Comey) December 1, 2017

Notwithstanding the usefulness of system dynamics in analyzing complex policy problems, policy design is far from straightforward and in many instances trial-and-error driven. To address this challenge, we propose to combine system dynamics with network controllability, an emerging field in network science, to facilitate the detection of effective leverage points in system dynamics models and thus to support the design of influential policies. We illustrate our approach by analyzing a classic system dynamics model: the World Dynamics model. We show that it is enough to control only 53% of the variables to steer the entire system to an arbitrary final state. We further rank all variables according to their importance in controlling the system and we validate our approach by showing that high ranked variables have a significantly larger impact on the system behavior compared to low ranked variables.

If and when you emerged from your happiness bubble to read the news, you’ll have seen (at least if you live in the US) that the cruel and reckless tax bill has passed the House of Representatives, and remains only to be reconciled with an equally-vicious Senate bill and then voted on by the Republican-controlled Senate.  The bill will add about $1.7 trillion to the national debt and raise taxes for about 47.5 million people, all in order to deliver a massive windfall to corporations, and to wealthy estates that already pay some of the lowest taxes in the developed world.

In a still-functioning democracy, those of us against such a policy would have an intellectual obligation to seek out the strongest arguments in favor of the policy and try to refute them.  By now, though, it seems to me that the Republicans hold the public in such contempt, and are so sure of the power of gerrymandering and voter restrictions to protect themselves from consequences, that they didn’t even bother to bring anything to the debate more substantive than the schoolyard bully’s “stop punching yourself.”  I guess some of them still repeat the fairytale about the purpose of tax cuts for the super-rich being to trickle down and help everyone else—but can even they advance that “theory” anymore without stifling giggles?  Mostly, as far as I can tell, they just brazenly deny that they’re doing what they obviously are doing: i.e., gleefully setting on fire anything that anyone, regardless of their ideology, could recognize as the national interest, in order to enrich a small core of supporters.

But none of that is what interests me in this post—because it’s “merely” as bad as, and no worse than, what one knew to expect when a coalition of thugs, kleptocrats, and white-nationalist demagogues seized control of Hamilton’s and Jefferson’s experiment.  My concern here is only with the “kill shot” that the Republicans have now aimed, with terrifying precision, at the system that’s kept American academic science the envy of the world in spite of the growing dysfunction all around it.

As you’ve probably heard, one of the ways Republicans intend to pay for their tax giveaway, is to change the tax code so that graduate students will now need to pay taxes on “tuition”—a large sum of money (as much as $50,000/year) that PhD students never actually see, that can easily exceed the stipends they do see, and that’s basically just an accounting trick that serves the internal needs of universities and granting agencies.  Again, to eliminate any chance of misunderstanding: PhD students, who are effectively low-wage employees, already pay taxes on their actual stipends.  The new proposal is that they’ll also have to pay taxes on a whopping, make-believe “X” on their payroll sheet that’s always exactly balanced out by “-X.”

For detailed analyses of the impacts, see, e.g. Luca Trevisan’s post or Inside Higher Ed or the Chronicle of Higher Ed or Vox or NPR.  Briefly, though, the proposal would raise taxes by a few thousand dollars per year, or in some cases as much as $10,000 per year (!), on PhD students who already live hand-to-mouth-to-ramen-bowl, with the largest impact falling on students in STEM fields.  For many students who aren’t independently wealthy, this could push a PhD beyond the realm of affordability, and cause them to leave academia or to do their graduate work in other countries.

“But isn’t there some workaround?”  Indeed, financial ignoramus that I am, my first reaction was to ask: if PhD tuition is basically an accounting fiction anyway, then why can’t the universities just declare that the tuition in question no longer exists, or is now zero dollars?  Feel free to explain further in the comments if you understand this stuff, but as far as I can tell, the answer is: because PhD tuition is used to calculate how much “tax” the universities can take from professors’ grant money.  If universities could no longer take that tax, and they had no other way to make up for it, then except for the richest few universities, they’d have to scale back research and teaching pretty drastically.  To avoid that outcome, the universities would be relying on the granting agencies to let them keep taking the overhead they needed to operate, even though the “PhD tuition” no longer existed.  But the granting agencies aren’t set up for this: you can’t just throw a bomb into one part of a complicated bureaucratic machine built up over decades, and expect the machine to continue working with no disruption to science.

But more ominously: as my friend Daniel Harlow and many others pointed out, it’s hard to look at the indefensible, laser-specific meanness of this policy, without suspecting that for many in Congress, the destruction of American higher education isn’t a regrettable byproduct, but the goal—just another piece of red meat to throw to the base.  If so, then we’d expect Congress to direct federal granting agencies not to loosen their rules about overhead, thereby forcing the students to pay the tax, and achieving the desired destruction.  (Note that the Trump administration has already made tightening overhead rules—i.e., doing the exact opposite of what would be needed to counteract the new tax—a central focus of its attempt to cut federal research funding.)

OK, two concluding thoughts:

1. When Republicans in Congress defended Trump’s travel ban, they at least had the craven excuse that they were only following the lead of the populist strongman who’d taken over their party.  Here they don’t even have that.  As far as I know, this targeted destruction of American higher education was Congress’s initiative, not Trump’s—which to me, underscores again the feather-thinness of any moral distinction between the Vichy GOP leadership and the administration with which it collaborates.  Trump didn’t emerge from nowhere.  It took decades of effort—George W. Bush, Sarah Palin, Karl Rove, Rush Limbaugh, Mitch McConnell, and all the rest—to transform the GOP into the pure seething cauldron of anti-intellectual resentment and hatred that we know today.
2. Given the existential risk to American higher education, why didn’t I blog about this earlier?  The answer is embarrassing to admit, and reflects no credit on me.  It’s simply that I didn’t believe it—even given all the other stuff that could “never happen in the US,” until it happened this past year.  I didn’t believe it, not because it was too far from me but because it was too close—because if true, it would mean the crippling of the research world in which I’ve spent most of my life since age 15, so therefore it couldn’t be true.  Surely even the House Republicans would realize they’d screwed up this time, and would take out this crazy provision before the full bill was voted on?  Or surely there’s some workaround that makes the whole thing less awful than it sounds?  There has to be … right?

Anyway, what else is there to say, except to call your representative, if you’re American and still have the faith in the system that such an act implies.

Since returning from a vacation partly spent isolated from the internet, I’ve been catching up and noticed that some of the most prominent sources of funding for math and physics research have been making the news:

• The New York Times and other sources have extensive reports based on leaked records from an offshore law firm that specializes in helping you avoid inconvenient US tax and reporting requirements. The story starts out with the example of Jim Simons, who has become the largest non-governmental funder of math and physics research. His Simons Foundation has been doing an excellent job of providing such funding. They have about \$3 billion in assets, annual income of around \$500 million. The Times reports that Simons (with a net worth of about \$18.5 billion) has an offshore version of the Foundation, the Simons Foundation International, with assets of \$8 billion, dwarfing the onshore version.
• The assets of these Foundations are presumably largely invested in the secretive and extremely successful Renaissance Technologies hedge fund, which also is the employer of quite a few physicists and mathematicians. I’ve asked many people over the years, but have never found anyone who knows (or will admit to knowing) what it is that RenTech does that is so successful. A peculiar aspect of the coming age of private math/physics research funding is that no one getting this funding really knows where the money comes from.

  In other news while I was away the CEO of RenTech, Robert Mercer, was finally induced to leave. Mercer had drawn a lot of attention recently since he in recent years has been taking the opposite tack to Simons, funding institutions devoted to promoting untruth over truth (e.g. Breitbart News), achieving fantastic success last year. He also has branched out from doing whatever secretive things RenTech does to make mountains of money using computers and data, starting up a firm called Cambridge Analytica, a firm involved in secretively using computers and data to undermine democracy in the US and elsewhere. I had been wondering for quite a while what Simons thought of Mercer’s activities. My understanding of highly-paid finance jobs was that your employer pays you a lot of money in return for having your full attention and devotion to not having negative stories about them come to public attention, so Mercer’s continued employment was surprising. It seems that Simons finally had enough, after realizing how much damage Mercer was doing to his firm, in particular by creating a situation that would discourage many people from wanting to work there (there also was a campaign underway to get institutions to divest from investments with RenTech).

• Another high profile source of funding for math and physics, in this case for cash prizes to mathematicians and physicists, has been venture capitalist Yuri Milner, with his Breakthrough Prize organization. New prizes will be announced in three weeks at a December 3 prize ceremony (I also believe there will be an associated Breakthrough Prize symposium held at Stanford shortly thereafter). It has always been well-known that much of Milner’s wealth derived from investments in Facebook and Twitter. Less well-known and recently revealed was that a major source of the funds for these investments was Russian state organizations closely tied to Vladimir Putin.
• Turning to sources of public funding, there’s not very positive news about a possible ILC collider in Japan, with reports of a cutback of the proposal from a 500 GeV to a 250 GeV machine (which would still cost about $7 billion).
• Foreign policy magazine has an article discussing the proposal for a huge new collider in China (discussed here). The point of view of the article is quite critical of the idea of locating a huge new project in a country with an increasingly authoritarian regime:
  

  China’s next-generation supercollider will unlock secrets of the universe — and destroy the ideals of the scientists running it.

  Luckily, for another more local prominent large country with an increasingly authoritarian and xenophobic regime, the issue of a possible problem with locating an international collider project there isn’t likely to come up since its leaders have no interest in funding such projects.

Scientists strive to understand how functionalities, such as conservation laws, emerge in complex systems. Living complex systems in particular create high-ordered functionalities by pairing up low-ordered complementary processes, e.g., one process to build and the other to correct. We propose a network mechanism that demonstrates how collective statistical laws can...

Humans seem to decide for themselves what to do, and when to do it. This distinctive capacity may emerge from an ability, shared with other animals, to make decisions for action that are related to future goals, or at least free from the constraints of immediate environmental inputs. Studying such volitional acts proves a major challenge for neuroscience. This review highlights key mechanisms in the generation of voluntary, as opposed to stimulus-driven actions, and highlights three issues. The first part focuses on the apparent spontaneity of voluntary action. The second part focuses on one of the most distinctive, but elusive, features of volition, namely, its link to conscious experience, and reviews stimulation and patient studies of the cortical basis of conscious volition down to the single-neuron level. Finally, we consider the goal-directedness of voluntary action, and discuss how internal generation of action can be linked to goals and reasons.

In preparation for the Applied Category Theory special session at U.C. Riverside this weekend, my crew dropped three papers on the arXiv.

My student Adam Yassine has been working on Hamiltonian and Lagrangian mechanics from an ‘open systems’ point of view:

• Adam Yassine, Open systems in classical mechanics.

  Abstract. Using the framework of category theory, we formalize the heuristic principles that physicists employ in constructing the Hamiltonians for open classical systems as sums of Hamiltonians of subsystems. First we construct a category where the objects are symplectic manifolds and the morphisms are spans whose legs are surjective Poisson maps. Using a slight variant of Fong’s theory of decorated cospans, we then decorate the apices of our spans with Hamiltonians. This gives a category where morphisms are open classical systems, and composition allows us to build these systems from smaller pieces.

He also gets a functor from a category of Lagrangian open systems to this category of Hamiltonian systems.

Kenny Courser and I have been continuing my work with Blake Pollard and Brendan Fong on open Markov processes, bringing 2-morphisms into the game. It seems easiest to use a double category:

• John Baez and Kenny Courser, Coarse-graining open Markov processes.

Abstract. Coarse-graining is a standard method of extracting a simple Markov process from a more complicated one by identifying states. Here we extend coarse-graining to open Markov processes. An ‘open’ Markov process is one where probability can flow in or out of certain states called ‘inputs’ and ‘outputs’. One can build up an ordinary Markov process from smaller open pieces in two basic ways: composition, where we identify the outputs of one open Markov process with the inputs of another, and tensoring, where we set two open Markov processes side by side. In previous work, Fong, Pollard and the first author showed that these constructions make open Markov processes into the morphisms of a symmetric monoidal category. Here we go further by constructing a symmetric monoidal double category where the 2-morphisms are ways of coarse-graining open Markov processes. We also extend the already known ‘black-boxing’ functor from the category of open Markov processes to our double category. Black-boxing sends any open Markov process to the linear relation between input and output data that holds in steady states, including nonequilibrium steady states where there is a nonzero flow of probability through the process. To extend black-boxing to a functor between double categories, we need to prove that black-boxing is compatible with coarse-graining.

Finally, the Complex Adaptive Systems Composition and Design Environment project with John Foley of Metron Scientific Solutions and my students Joseph Moeller and Blake Pollard has finally given birth to a paper! I hope this is just the first; it starts laying down the theoretical groundwork for designing networked systems. John is here now and we’re coming up with a bunch of new ideas:

• John Baez, John Foley, Joseph Moeller and Blake Pollard, Network models.

Abstract. Networks can be combined in many ways, such as overlaying one on top of another or setting two side by side. We introduce network models to encode these ways of combining networks. Different network models describe different kinds of networks. We show that each network model gives rise to an operad, whose operations are ways of assembling a network of the given kind from smaller parts. Such operads, and their algebras, can serve as tools for designing networks. Technically, a network model is a lax symmetric monoidal functor from the free symmetric monoidal category on some set to Cat, and the construction of the corresponding operad proceeds via a symmetric monoidal version of the Grothendieck construction.

I blogged about this last one here:

• Complex adaptive systems (part 6), Azimuth.

Network control principles predict neuron function in the Caenorhabditis elegans connectome

Nature 550, 7677 (2017). doi:10.1038/nature24056

Authors: Gang Yan, Petra E. Vértes, Emma K. Towlson, Yee Lian Chew, Denise S. Walker, William R. Schafer & Albert-László Barabási

Recent studies on the controllability of complex systems offer a powerful mathematical framework to systematically explore the structure–function relationship in biological, social, and technological networks. Despite theoretical advances, we lack direct experimental proof of the validity of these widely used control principles. Here we fill this gap by applying a control framework to the connectome of the nematode Caenorhabditis elegans, allowing us to predict the involvement of each C. elegans neuron in locomotor behaviours. We predict that control of the muscles or motor neurons requires 12 neuronal classes, which include neuronal groups previously implicated in locomotion by laser ablation, as well as one previously uncharacterized neuron, PDB. We validate this prediction experimentally, finding that the ablation of PDB leads to a significant loss of dorsoventral polarity in large body bends. Importantly, control principles also allow us to investigate the involvement of individual neurons within each neuronal class. For example, we predict that, within the class of DD motor neurons, only three (DD04, DD05, or DD06) should affect locomotion when ablated individually. This prediction is also confirmed; single cell ablations of DD04 or DD05 specifically affect posterior body movements, whereas ablations of DD02 or DD03 do not. Our predictions are robust to deletions of weak connections, missing connections, and rewired connections in the current connectome, indicating the potential applicability of this analytical framework to larger and less well-characterized connectomes.

The mammalian neocortex contains many cell types, but whether they organize into repeated structures has been unclear. We discovered that major cell types in neocortical layer 5 form a lattice structure in many brain areas. Large-scale three-dimensional imaging revealed that distinct types of excitatory and inhibitory neurons form cell type–specific radial clusters termed microcolumns. Thousands of microcolumns, in turn, are patterned into a hexagonal mosaic tessellating diverse regions of the neocortex. Microcolumn neurons demonstrate synchronized in vivo activity and visual responses with similar orientation preference and ocular dominance. In early postnatal development, microcolumns are coupled by cell type–specific gap junctions and later serve as hubs for convergent synaptic inputs. Thus, layer 5 neurons organize into a brainwide modular system, providing a template for cortical processing.

The post There is no great startup stagnation crow average is over appeared first on Marginal REVOLUTION.

I’ve been slacking off on writing this series of posts… but for a good reason: I’ve been busy writing a paper on the same topic! In the process I caught a couple of mistakes in what I’ve said so far. But more importantly, there’s a version out now, that you can read:

• John Baez, John Foley, Blake Pollard and Joseph Moeller, Network models.

There will be two talks about this at the AMS special session on Applied Category Theory this weekend at U. C. Riverside: one by John Foley of Metron Inc., and one by my grad student Joseph Moeller. I’ll try to get their talk slides someday. But for now, here’s the basic idea.

Our goal is to build operads suited for designing networks. These could be networks where the vertices represent fixed or moving agents and the edges represent communication channels. More generally, they could be networks where the vertices represent entities of various types, and the edges represent relationships between these entities—for example, that one agent is committed to take some action involving the other. This paper arose from an example where the vertices represent planes, boats and drones involved in a search and rescue mission in the Caribbean. However, even for this one example, we wanted a flexible formalism that can handle networks of many kinds, described at a level of detail that the user is free to adjust.

To achieve this flexibility, we introduced a general concept of ‘network model’. Simply put, a network model is a kind of network. Any network model gives an operad whose operations are ways to build larger networks of this kind by gluing smaller ones. This operad has a ‘canonical’ algebra where the operations act to assemble networks of the given kind. But it also has other algebras, where it acts to assemble networks of this kind equipped with extra structure and properties. This flexibility is important in applications.

What exactly is a ‘kind of network’? That’s the question we had to answer. We started with some examples, At the crudest level, we can model networks as simple graphs. If the vertices are agents of some sort and the edges represent communication channels, this means we allow at most one channel between any pair of agents.

However, simple graphs are too restrictive for many applications. If we allow multiple communication channels between a pair of agents, we should replace simple graphs with ‘multigraphs’. Alternatively, we may wish to allow directed channels, where the sender and receiver have different capabilities: for example, signals may only be able to flow in one direction. This requires replacing simple graphs with ‘directed graphs’. To combine these features we could use ‘directed multigraphs’.

But none of these are sufficiently general. It’s also important to consider graphs with colored vertices, to specify different types of agents, and colored edges, to specify different types of channels. This leads us to ‘colored directed multigraphs’.

All these are examples of what we mean by a ‘kind of network’, but none is sufficiently general. More complicated kinds, such as hypergraphs or Petri nets, are likely to become important as we proceed.

Thus, instead of separately studying all these kinds of networks, we introduced a unified notion that subsumes all these variants: a ‘network model’. Namely, given a set of ‘vertex colors’, a network model is a lax symmetric monoidal functor

where is the free strict symmetric monoidal category on and is the category of small categories.

Unpacking this somewhat terrifying definition takes a little work. It simplifies in the special case where takes values in the category of monoids. It simplifies further when is a singleton, since then is the groupoid where objects are natural numbers and morphisms from to are bijections

If we impose both these simplifying assumptions, we have what we call a one-colored network model: a lax symmetric monoidal functor

As we shall see, the network model of simple graphs is a one-colored network model, and so are many other motivating examples. If you like André Joyal’s theory of ‘species’, then one-colored network models should be pretty fun, since they’re species with some extra bells and whistles.

But if you don’t, there’s still no reason to panic. In relatively down-to-earth terms, a one-colored network model amounts to roughly this. If we call elements of ‘networks with vertices’, then:

• Since is a monoid, we can overlay two networks with the same number of vertices and get a new one. We call this operation

• Since is a functor, the symmetric group acts on the monoid Thus, for each , we have a monoid automorphism that we call simply

• Since is lax monoidal, we also have an operation

We call this operation the disjoint union of networks. In examples like simple graphs, it looks just like what it sounds like.

Unpacking the abstract definition further, we see that these operations obey some equations, which we list in Theorem 11 of our paper. They’re all obvious if you draw pictures of examples… and don’t worry, our paper has a few pictures. (We plan to add more.) For example, the ‘interchange law’

holds whenever and This is a nice relationship between overlaying networks and taking their disjoint union.

In Section 2 of our apper we study one-colored network models, and give lots of examples. In Section 3 we describe a systematic procedure for getting one-colored network models from monoids. In Section 4 we study general network models and give examples of these. In Section 5 we describe a category of network models, and show that the procedure for getting network models from monoids is functorial. We also make into a symmetric monoidal category, and give examples of how to build new networks models by tensoring old ones.

Our main result is that any network model gives a typed operad, also known as a ‘colored operad’. This operad has operations that describe how to stick networks of the given kind together to form larger networks of this kind. This operad has a ‘canonical algebra’, where it acts on networks of the given kind—but the real point is that it has lots of other algebra, where it acts on networks of the given kind equipped with extra structure and properties.

The technical heart of our paper is Section 6, mainly written by Joseph Moeller. This provides the machinery to construct operads from network models in a functorial way. Category theorists should find this section interesting, because because it describes enhancements of the well-known ‘Grothendieck construction’ of the category of elements of a functor

where is any small category. For example, if is symmetric monoidal and is lax symmetric monoidal, then we show is symmetric monoidal. Moreover, we show that the construction sending the lax symmetric monoidal functor to the symmetric monoidal category is functorial.

In Section 7 we apply this machinery to build operads from network models. In Section 8 we describe some algebras of these operads, including an algebra whose elements are networks of range-limited communication channels. In future work we plan to give many more detailed examples, and to explain how these algebras, and the homomorphisms between them, can be used to design and optimize networks.

I want to explain all this in more detail—this is a pretty hasty summary, since I’m busy this week. But for now you can read the paper!

Complex networks are the subject of fundamental interest from the scientific community at large. Several metrics have been introduced to characterize the structure of these networks, such as the degree distribution, degree correlation, path length, clustering coefficient, centrality measures etc. Another important feature is the presence of network symmetries. In particular, the effect of these symmetries has been studied in the context of network synchronization, where they have been used to predict the emergence and stability of cluster synchronous states. Here we provide theoretical, numerical, and experimental evidence that network symmetries play a role in a substantially broader class of dynamical models on networks, including epidemics, game theory, communication, and coupled excitable systems. Namely, we see that in all these models, nodes that are related by a symmetry relation show the same time-averaged dynamical properties. This discovery leads us to propose reduction techniques for exact, yet minimal, simulation of complex networks dynamics, which we show are effective in order to optimize the use of computational resources, such as computation time and memory.

We present an approach to generate chimera dynamics (localized frequency synchrony) in oscillator networks with two populations of (at least) two elements using a general method based on delayed interactions with linear and quadratic terms. The coupling design yields robust chimeras through a phase-model-based design of the delay and the ratio of linear and quadratic components of the interactions. We demonstrate the method in the Brusselator model and experiments with electrochemical oscillators. The technique opens the way to directly bridge chimera dynamics in phase models and real-world oscillator networks.

In network theory, a question of prime importance is how to assess network vulnerability in a fast and reliable manner. With this issue in mind, we investigate the response to parameter changes of coupled dynamical systems on complex networks. We find that for specific, non-averaged perturbations, the response of synchronous states critically depends on the overlap between the perturbation vector and the eigenmodes of the stability matrix of the unperturbed dynamics. Once averaged over properly defined ensembles of such perturbations, the response is given by new graph topological indices, which we introduce as generalized Kirchhoff indices. These findings allow for a fast and reliable method for assessing the specific or average vulnerability of a network against changing operational conditions, faults or external attacks.

Galaxy slime. [Img Src] Physicists have gathered evidence that space-time can behave like a fluid. Mathematical evidence, that is, but still evidence. If this relation isn’t a coincidence, then space-time – like a fluid – may have substructure. We shouldn’t speak of space and time as if the two were distant cousins. We have known at least since Einstein that space and time are inseparable,

We present a simple and computationally efficient coarse-grained and solvent-free model for simulating lipid bilayer membranes. In order to be used in concert with particle-based reaction-diffusion simulations, the model is purely based on interacting and reacting particles, each representing a coarse patch of a lipid monolayer. Particle interactions include nearest-neighbor bond-stretching and angle-bending, and are parameterized so as to reproduce the local membrane mechanics given by the Helfrich energy density over a range of relevant curvatures. In-plane fluidity is implemented with Monte Carlo bond-flipping moves. The physical accuracy of the model is verified by five tests: (i) Power spectrum analysis of equilibrium thermal undulations is used to verify that the particle-based representation correctly captures the dynamics predicted by the continuum model of fluid membranes. (ii) It is verified that the input bending stiffness, against which the potential parameters are optimized, is accurately recovered. (iii) Isothermal area compressibility modulus of the membrane is calculated and is shown to be tunable to reproduce available values for different lipid bilayers, independent of the bending rigidity. (iv) Simulation of two-dimensional shear flow under a gravity force is employed to measure the effective in-plane viscosity of the membrane model, and show the possibility of modeling membranes with specified viscosities. (v) Interaction of the bilayer membrane with a spherical nanoparticle is modeled as a test case for large membrane deformations and budding involved in cellular processes such as endocytosis...

Social behaviour shapes hypothalamic neural ensemble representations of conspecific sex

Nature 550, 7676 (2017). doi:10.1038/nature23885

Authors: Ryan Remedios, Ann Kennedy, Moriel Zelikowsky, Benjamin F. Grewe, Mark J. Schnitzer & David J. Anderson

All animals possess a repertoire of innate (or instinctive) behaviours, which can be performed without training. Whether such behaviours are mediated by anatomically distinct and/or genetically specified neural pathways remains unknown. Here we report that neural representations within the mouse hypothalamus, that underlie innate social behaviours, are shaped by social experience. Oestrogen receptor 1-expressing (Esr1+) neurons in the ventrolateral subdivision of the ventromedial hypothalamus (VMHvl) control mating and fighting in rodents. We used microendoscopy to image Esr1+ neuronal activity in the VMHvl of male mice engaged in these social behaviours. In sexually and socially experienced adult males, divergent and characteristic neural ensembles represented male versus female conspecifics. However, in inexperienced adult males, male and female intruders activated overlapping neuronal populations. Sex-specific neuronal ensembles gradually separated as the mice acquired social and sexual experience. In mice permitted to investigate but not to mount or attack conspecifics, ensemble divergence did not occur. However, 30 minutes of sexual experience with a female was sufficient to promote the separation of male and female ensembles and to induce an attack response 24 h later. These observations uncover an unexpected social experience-dependent component to the formation of hypothalamic neural assemblies controlling innate social behaviours. More generally, they reveal plasticity and dynamic coding in an evolutionarily ancient deep subcortical structure that is traditionally viewed as a ‘hard-wired’ system.

Predictive coding is becoming a popular theory in neuroscience (see for example Clark 2013). In a nutshell, the general idea is that brains encode predictions of their sensory inputs. This is an appealing idea because superficially, it makes a lot of sense: functionally, the only reason why you would want to process sensory information is if it might impact your future, so it makes sense to try to predict your sensory inputs.

There are substantial problems in the details of predictive coding theories, for example with the arbitrariness of the metric by which you judge that your prediction matches sensory inputs (what is important?), or the fact that predictive coding schemes encode both noise and signal. But I want to focus on the more fundamental problems. One has to with “coding”, the other with “predictive”.

It makes sense that brains anticipate. But does it make sense that brains code? Coding is a metaphor of a communication channel, and this is generally not a great metaphor for what the brain might do, unless you fully embrace dualism. I discuss this at length in a recent paper (Is coding a relevant metaphor for the brain?) so I won’t repeat the entire argument here. Predictive coding is a branch of efficient coding, so the same fallacy underlies its logic: 1) neurons encode sensory inputs; 2) living organisms are efficient; => brains must encode efficiently. (1) is trivially true in the sense that one can define a mapping from sensory inputs to neural activity. (2) is probably true to some extent (evolutionary arguments). So the conclusion follows. Critiques of efficient coding have focused on the “efficient” part: maybe the brain is not that efficient after all. But the error is elsewhere: living organisms are certainly efficient, but it doesn’t follow that they are efficient at coding. They might be efficient at surviving and reproducing, and it is not obvious that it entails coding efficiency (see the last part of the abovementioned paper for a counter-example). So the real strong assumption is there: the main function of the brain is to represent sensory inputs.

The second problem has to with “predictive”. It makes sense that an important function of brains, or in fact of any living organism, is to anticipate (see the great Anticipatory Systems by Robert Rosen). But to what extent do predictive coding schemes actually anticipate? First, in practice, those are generally not prediction schemes but compression schemes, in the sense that they do not tell us what will happen next but what happens now. This is at least the case of the classical Rao & Ballard (1999). Neurons encode the difference between expected input and actual input: this is compression, not prediction. It uses a sort of prediction in order to compress: other neurons (in higher layers) produce predictions of the inputs to those neurons, but the term prediction is used in the sense that the inputs are not known to the higher layer neurons, not that the “prediction” occurs before the inputs. Thus the term “predictive” is misleading because it is not used in a temporal sense.

However, it is relatively easy to imagine how predictive coding might be about temporal predictions, although the neural implementation is not straightforward (delays etc). So I want to make a deeper criticism. I started by claiming that it is useful to predict sensory inputs. I am taking this back (I can because I said it was superficial reasoning). It is not useful to know what will happen. What is useful is to know what might happen, depending on what you do. If there is nothing you can do about the future, what is the functional use of predicting it? So what is useful is to predict the future conditionally to a different set of potential actions. This is about manipulating models of the world, not representing the present.

Membrane protein transporters alternate their substrate-binding sites between the extracellular and cytosolic side of the membrane according to the alternating access mechanism. Inspired by this intriguing mechanism devised by nature, we study particle transport through a channel coupled with an energy well that oscillates its position between the two entrances of the channel. We optimize particle transport across the channel by adjusting the oscillation frequency. At the optimal oscillation frequency, the translocation rate through the channel is a hundred times higher with respect to free diffusion across the channel. Our findings reveal the effect of time dependent potentials on particle transport across a channel and will be relevant for membrane transport and microfluidics application.

Vladimir Voevodsky died last week. He won the Fields Medal in 2002 for proving the Milnor conjecture in a branch of algebra known as algebraic K-theory. He continued to work on this subject until he helped prove the more general Bloch–Kato conjecture in 2010.

Proving these results—which are too technical to easily describe to nonmathematicians!—required him to develop a dream of Grothendieck: the theory of motives. Very roughly, this is a way of taking the space of solutions of some polynomial equations and chopping it apart into building blocks. But this process of ‘chopping’ and also these building blocks, called ‘motives’, are very abstract—nothing easy to visualize.

It’s a bit like how a proton is made of quarks. You never actually see a quark in isolation, so you have to think very hard to realize they are there at all. But once you know this, a lot of things become clear.

This is wonderful, profound mathematics. But in the process of proving the Bloch-Kato conjecture, Voevodsky became tired of this stuff. He wanted to do something more useful… and more ambitious. He later said:

It was very difficult. In fact, it was 10 years of technical work on a topic that did not interest me during the last 5 of these 10 years. Everything was done only through willpower.

Since the autumn of 1997, I already understood that my main contribution to the theory of motives and motivic cohomology was made. Since that time I have been very conscious and actively looking for. I was looking for a topic that I would deal with after I fulfilled my obligations related to the Bloch-Kato hypothesis.

I quickly realized that if I wanted to do something really serious, then I should make the most of my accumulated knowledge and skills in mathematics. On the other hand, seeing the trends in the development of mathematics as a science, I realized that the time is coming when the proof of yet another conjecture won’t have much of an effect. I realized that mathematics is on the verge of a crisis, or rather, two crises.

The first is connected with the separation of “pure” and applied mathematics. It is clear that sooner or later there will be a question about why society should pay money to people who are engaged in things that do not have any practical applications.

The second, less obvious, is connected with the complication of pure mathematics, which leads to the fact that, sooner or later, the articles will become too complicated for detailed verification and the process of accumulating undetected errors will begin. And since mathematics is a very deep science, in the sense that the results of one article usually depend on the results of many and many previous articles, this accumulation of errors for mathematics is very dangerous.

So, I decided, you need to try to do something that will help prevent these crises. For the first crisis, this meant that it was necessary to find an applied problem that required for its solution the methods of pure mathematics developed in recent years or even decades.

He looked for such a problem. He studied biology and found an interesting candidate. He worked on it very hard, but then decided he’d gone down a wrong path:

Since childhood I have been interested in natural sciences (physics, chemistry, biology), as well as in the theory of computer languages, and since 1997, I have read a lot on these topics, and even took several student and post-graduate courses. In fact, I “updated” and deepened the knowledge that had to a very large extent. All this time I was looking for that I recognized open problems that would be of interest to me and to which I could apply modern mathematics.

As a result, I chose, I now understand incorrectly, the problem of recovering the history of populations from their modern genetic composition. I took on this task for a total of about two years, and in the end, already by 2009, I realized that what I was inventing was useless. In my life, so far, it was, perhaps, the greatest scientific failure. A lot of work was invested in the project, which completely failed. Of course, there was some benefit, of course—I learned a lot of probability theory, which I knew badly, and also learned a lot about demography and demographic history.

But he bounced back! He came up with a new approach to the foundations of mathematics, and helped organize a team at the Institute of Advanced Studies at Princeton to develop it further. This approach is now called homotopy type theory or univalent foundations. It’s fundamentally different from set theory. It treats the fundamental concept of equality in a brand new way! And it’s designed to be done with the help of computers.

It seems he started down this new road when the mathematician Carlos Simpson pointed out a serious mistake in a paper he’d written.

I think it was at this moment that I largely stopped doing what is called “curiosity-driven research” and started to think seriously about the future. I didn’t have the tools to explore the areas where curiosity was leading me and the areas that I considered to be of value and of interest and of beauty.

So I started to look into what I could do to create such tools. And it soon became clear that the only long-term solution was somehow to make it possible for me to use computers to verify my abstract, logical, and mathematical constructions. The software for doing this has been in development since the sixties. At the time, when I started to look for a practical proof assistant around 2000, I could not find any. There were several groups developing such systems, but none of them was in any way appropriate for the kind of mathematics for which I needed a system.

When I first started to explore the possibility, computer proof verification was almost a forbidden subject among mathematicians. A conversation about the need for computer proof assistants would invariably drift to Gödel’s incompleteness theorem (which has nothing to do with the actual problem) or to one or two cases of verification of already existing proofs, which were used only to demonstrate how impractical the whole idea was. Among the very few mathematicians who persisted in trying to advance the field of computer verification in mathematics during this time were Tom Hales and Carlos Simpson. Today, only a few years later, computer verification of proofs and of mathematical reasoning in general looks completely practical to many people who work on univalent foundations and homotopy type theory.

The primary challenge that needed to be addressed was that the foundations of mathematics were unprepared for the requirements of the task. Formulating mathematical reasoning in a language precise enough for a computer to follow meant using a foundational system of mathematics not as a standard of consistency to establish a few fundamental theorems, but as a tool that can be employed in ­everyday mathematical work. There were two main problems with the existing foundational systems, which made them inadequate. Firstly, existing foundations of mathematics were based on the languages of predicate logic and languages of this class are too limited. Secondly, existing foundations could not be used to directly express statements about such objects as, for example, the ones in my work on 2-theories.

Still, it is extremely difficult to accept that mathematics is in need of a completely new foundation. Even many of the people who are directly connected with the advances in homotopy type theory are struggling with this idea. There is a good reason: the existing foundations of mathematics—ZFC and category theory—have been very successful. Overcoming the appeal of category theory as a candidate for new foundations of mathematics was for me personally the most challenging.

Homotopy type theory is now a vital and exciting area of mathematics. It’s far from done, and to make it live up to Voevodsky’s dreams will require brand new ideas—not just incremental improvements, but actual sparks of genius. For some of the open problems, see Mike Shulman’s comment on the n-Category Café, and some replies to that.

I only met him a few times, but as far as I can tell Voevodsky was a completely unpretentious person. You can see that in the picture here.

He was also a very complex person. For example, you might not guess that he took great wildlife photos:

You also might not guess at this side of him:

In 2006-2007 a lot of external and internal events happened to me, after which my point of view on the questions of the “supernatural” changed significantly. What happened to me during these years, perhaps, can be compared most closely to what happened to Karl Jung in 1913-14. Jung called it “confrontation with the unconscious”. I do not know what to call it, but I can describe it in a few words. Remaining more or less normal, apart from the fact that I was trying to discuss what was happening to me with people whom I should not have discussed it with, I had in a few months acquired a very considerable experience of visions, voices, periods when parts of my body did not obey me, and a lot of incredible accidents. The most intense period was in mid-April 2007 when I spent 9 days (7 of them in the Mormon capital of Salt Lake City), never falling asleep for all these days.

Almost from the very beginning, I found that many of these phenomena (voices, visions, various sensory hallucinations), I could control. So I was not scared and did not feel sick, but perceived everything as something very interesting, actively trying to interact with those “beings” in the auditorial, visual and then tactile spaces that appeared around me (by themselves or by invoking them). I must say, probably, to avoid possible speculations on this subject, that I did not use any drugs during this period, tried to eat and sleep a lot, and drank diluted white wine.

Another comment: when I say “beings”, naturally I mean what in modern terminology are called complex hallucinations. The word “beings” emphasizes that these hallucinations themselves “behaved”, possessed a memory independent of my memory, and reacted to attempts at communication. In addition, they were often perceived in concert in various sensory modalities. For example, I played several times with a (hallucinated) ball with a (hallucinated) girl—and I saw this ball, and felt it with my palm when I threw it.

Despite the fact that all this was very interesting, it was very difficult. It happened for several periods, the longest of which lasted from September 2007 to February 2008 without breaks. There were days when I could not read, and days when coordination of movements was broken to such an extent that it was difficult to walk.

I managed to get out of this state due to the fact that I forced myself to start math again. By the middle of spring 2008 I could already function more or less normally and even went to Salt Lake City to look at the places where I wandered, not knowing where I was, in the spring of 2007.

In short, he was a genius akin to Cantor or Grothendieck, at times teetering on the brink of sanity, yet gripped by an immense desire for beauty and clarity, engaging in struggles that gripped his whole soul. From the fires of this volcano, truly original ideas emerge.

This last quote, and the first few quotes, are from some interviews in Russian, done by Roman Mikhailov, which Mike Stay pointed out to me. I used Google Translate and polished the results a bit:

• Интервью Владимира Воеводского (часть 1), 1 July 2012. English version via Google Translate: Interview with Vladimir Voevodsky (Part 1).

• Интервью Владимира Воеводского (часть 2), 5 July 2012. English version via Google Translate: Interview with Vladimir Voevodsky (Part 2).

The quote about the origins of ‘univalent foundations’ comes from his nice essay here:

• Vladimir Voevodsky, The origins and motivations of univalent foundations, 2014.

There’s also a good obituary of Voevodsky explaining its relation to Grothendieck’s idea in simple terms:

• Institute for Advanced Studies, Vladimir Voevodsky 1966–2017, 4 October 2017.

The photograph of Voevodsky is from Andrej Bauer’s website:

• Andrej Bauer, Photos of mathematicians.

To learn homotopy type theory, try this great book:

• Homotopy Type Theory: Univalent Foundations of Mathematics, The Univalent Foundations Program, Institute for Advanced Study.

We introduce the Hierarchically Interacting Particle Neural Network (HIP-NN) to model molecular properties from datasets of quantum calculations. Inspired by a many-body expansion, HIP-NN decomposes properties, such as energy, as a sum over hierarchical terms. These terms are generated from a neural network--a composition of many nonlinear transformations--acting on a representation of the molecule. HIP-NN achieves state-of-the-art performance on a dataset of 131k ground state organic molecules, and predicts energies with 0.26 kcal/mol mean absolute error. With minimal tuning, our model is also competitive on a dataset of molecular dynamics trajectories. In addition to enabling accurate energy predictions, the hierarchical structure of HIP-NN helps to identify regions of model uncertainty.

Spiking networks that perform probabilistic inference have been proposed both as models of cortical computation and as candidates for solving problems in machine learning. However, the evidence for spike-based computation being in any way superior to non-spiking alternatives remains scarce. We propose that short-term plasticity can provide spiking networks with distinct computational advantages compared to their classical counterparts. In this work, we use networks of leaky integrate-and-fire neurons that are trained to perform both discriminative and generative tasks in their forward and backward information processing paths, respectively. During training, the energy landscape associated with their dynamics becomes highly diverse, with deep attractor basins separated by high barriers. Classical algorithms solve this problem by employing various tempering techniques, which are both computationally demanding and require global state updates. We demonstrate how similar results can be achieved in spiking networks endowed with local short-term synaptic plasticity. Additionally, we discuss how these networks can even outperform tempering-based approaches when the training data is imbalanced. We thereby show how biologically inspired, local, spike-triggered synaptic dynamics based simply on a limited pool of synaptic resources can allow spiking networks to outperform their non-spiking relatives.

The materialistic transparency of culture 
has not made it more honest, only more vulgar 
(Th. Adorno)

From Bill Readings, The University in ruins:

I am attracted to Robert Young’s suggestion that the University, both inside and outside the market economy, should “function as a surplus that the economy cannot comprehend’’. The binary opposition is there, and the University will deconstruct it by being neither simply useful nor simply useless. All very good, and very much what Humboldt wanted: indirect utility, direct uselessness for the state.”

This sentence makes for a perfect answer to all those people who are trying to sex up the University courses by bringing more industry into education. Anyway, the book by Readings is good and dense with concepts that its’s almost impossible to choose which excerpts I like most, I’d have to re-write it all on my blog… Here’s how the academia threats the arts by normalizing them:

“Rather than posing a threat, the analyses of Cultural Studies risk providing new marketing opportunities for the system. Practices such as punk music and dress styles are offered their self-consciousness in academic essays, but the dignity they acquire is not that of authenticity but of marketability, be it in the cinema, on MTV, or as a site of tourist interest for visitor to London. […] To put it bluntly, the shock value of punk is not lasting in a cultural sense, since it soon becomes possible to be “excellently punk”.

Here academia works as an advanced tool to extrapolate culture out of the context where it was born as a social practice, to make into a product of “culture” for a community of rich and educated people who have never been punk and never wished to be.

The transition from wakefulness to general anesthesia is widely attributed to suppressive actions of anesthetic molecules distributed by the systemic circulation to the cerebral cortex (for amnesia and loss of consciousness) and to the spinal cord (for atonia and antinociception). An alternative hypothesis proposes that anesthetics act on one or more brainstem or diencephalic nuclei, with suppression of cortex and spinal cord mediated by dedicated axonal pathways. Previously, we documented induction of an anesthesia-like state in rats by microinjection of small amounts of GABA_A-receptor agonists into an upper brainstem region named the mesopontine tegmental anesthesia area (MPTA). Correspondingly, lesioning this area rendered animals resistant to systemically delivered anesthetics. Here, using rats of both sexes, we applied a modified microinjection method that permitted localization of the anesthetic-sensitive neurons with much improved spatial resolution. Microinjected at the MPTA hotspot identified, exposure of 1900 or fewer neurons to muscimol was sufficient to sustain whole-body general anesthesia; microinjection as little as 0.5 mm off-target did not. The GABAergic anesthetics pentobarbital and propofol were also effective. The GABA-sensitive cell cluster is centered on a tegmental (reticular) field traversed by fibers of the superior cerebellar peduncle. It has no specific nuclear designation and has not previously been implicated in brain-state transitions.

SIGNIFICANCE STATEMENT General anesthesia permits pain-free surgery. Furthermore, because anesthetic agents have the unique ability to reversibly switch the brain from wakefulness to a state of unconsciousness, knowing how and where they work is a potential route to unraveling the neural mechanisms that underlie awareness itself. Using a novel method, we have located a small, and apparently one of a kind, cluster of neurons in the mesopontine tegmentum that are capable of effecting brain-state switching when exposed to GABA_A-receptor agonists. This action appears to be mediated by a network of dedicated axonal pathways that project directly and/or indirectly to nearby arousal nuclei of the brainstem and to more distant targets in the forebrain and spinal cord.

His name almost, but not quite, movie villain scary.

Washington (CNN)US investigators wiretapped former Trump campaign chairman Paul Manafort under secret court orders before and after the election, sources tell CNN, an extraordinary step involving a high-ranking campaign official now at the center of the Russia meddling probe.

The government snooping continued into early this year, including a period when Manafort was known to talk to President Donald Trump.

Life comes at us fast.

WASHINGTON — Paul J. Manafort was in bed early one morning in July when federal agents bearing a search warrant picked the lock on his front door and raided his Virginia home. They took binders stuffed with documents and copied his computer files, looking for evidence that Mr. Manafort, President Trump’s former campaign chairman, set up secret offshore bank accounts. They even photographed the expensive suits in his closet.

The special counsel, Robert S. Mueller III, then followed the house search with a warning: His prosecutors told Mr. Manafort they planned to indict him, said two people close to the investigation.

Also from the first article:

The conversations between Manafort and Trump continued after the President took office, long after the FBI investigation into Manafort was publicly known, the sources told CNN. They went on until lawyers for the President and Manafort insisted that they stop, according to the sources.

Dumbest crooks on the planet.

Two critical challenges in the design and synthesis of molecular robots are modularity and algorithm simplicity. We demonstrate three modular building blocks for a DNA robot that performs cargo sorting at the molecular level. A simple algorithm encoding recognition between cargos and their destinations allows for a simple robot design: a single-stranded DNA with one leg and two foot domains for walking, and one arm and one hand domain for picking up and dropping off cargos. The robot explores a two-dimensional testing ground on the surface of DNA origami, picks up multiple cargos of two types that are initially at unordered locations, and delivers them to specified destinations until all molecules are sorted into two distinct piles. The robot is designed to perform a random walk without any energy supply. Exploiting this feature, a single robot can repeatedly sort multiple cargos. Localization on DNA origami allows for distinct cargo-sorting tasks to take place simultaneously in one test tube or for multiple robots to collectively perform the same task.

Temporal codes are theoretically powerful encoding schemes, but their precise form in the neocortex remains unknown in part because of the large number of possible codes and the difficulty in disambiguating informative spikes from statistical noise. A biologically plausible and computationally powerful temporal coding scheme is the Hebbian assembly phase sequence (APS), which predicts reliable propagation of spikes between functionally related assemblies of neurons. Here, we sought to measure the inherent capacity of neocortical networks to produce reliable sequences of spikes, as would be predicted by an APS code. To record microcircuit activity, the scale at which computation is implemented, we used two-photon calcium imaging to densely sample spontaneous activity in murine neocortical networks ex vivo. We show that the population spike histogram is sufficient to produce a spatiotemporal progression of activity across the population. To more comprehensively evaluate the capacity for sequential spiking that cannot be explained by the overall population spiking, we identify statistically significant spike sequences. We found a large repertoire of sequence spikes that collectively comprise the majority of spiking in the circuit. Sequences manifest probabilistically and share neuron membership, resulting in unique ensembles of interwoven sequences characterizing individual spatiotemporal progressions of activity. Distillation of population dynamics into its constituent sequences provides a way to capture trial-to-trial variability and may prove to be a powerful decoding substrate in vivo. Informed by these data, we suggest that the Hebbian APS be reformulated as interwoven sequences with flexible assembly membership due to shared overlapping neurons.

NEW & NOTEWORTHY Neocortical computation occurs largely within microcircuits comprised of individual neurons and their connections within small volumes (<500 μm³). We found evidence for a long-postulated temporal code, the Hebbian assembly phase sequence, by identifying repeated and co-occurring sequences of spikes. Variance in population activity across trials was explained in part by the ensemble of active sequences. The presence of interwoven sequences suggests that neuronal assembly structure can be variable and is determined by previous activity.

Molecular machines open cell membranes

Nature 548, 7669 (2017). doi:10.1038/nature23657

Authors: Víctor García-López, Fang Chen, Lizanne G. Nilewski, Guillaume Duret, Amir Aliyan, Anatoly B. Kolomeisky, Jacob T. Robinson, Gufeng Wang, Robert Pal & James M. Tour

Beyond the more common chemical delivery strategies, several physical techniques are used to open the lipid bilayers of cellular membranes. These include using electric and magnetic fields, temperature, ultrasound or light to introduce compounds into cells, to release molecular species from cells or to selectively induce programmed cell death (apoptosis) or uncontrolled cell death (necrosis). More recently, molecular motors and switches that can change their conformation in a controlled manner in response to external stimuli have been used to produce mechanical actions on tissue for biomedical applications. Here we show that molecular machines can drill through cellular bilayers using their molecular-scale actuation, specifically nanomechanical action. Upon physical adsorption of the molecular motors onto lipid bilayers and subsequent activation of the motors using ultraviolet light, holes are drilled in the cell membranes. We designed molecular motors and complementary experimental protocols that use nanomechanical action to induce the diffusion of chemical species out of synthetic vesicles, to enhance the diffusion of traceable molecular machines into and within live cells, to induce necrosis and to introduce chemical species into live cells. We also show that, by using molecular machines that bear short peptide addends, nanomechanical action can selectively target specific cell-surface recognition sites. Beyond the in vitro applications demonstrated here, we expect that molecular machines could also be used in vivo, especially as their design progresses to allow two-photon, near-infrared and radio-frequency activation.

There’s a new college-level textbook out, Cosmology for the Curious, targeted at physics courses designed to explain basics of cosmology to non-physics majors. The authors are Delia Perlov and Alex Vilenkin. Back in 2006 Vilenkin published a popular book promoting the multiverse, Many Worlds in One, which I wrote about at the time, making the obvious comment that there was nothing like a testable experimental prediction to be found in the book. It seemed to me then that the physics community would never take seriously an inherently untestable theory, recognizing such a thing as pseudo-science. I thought that the only reason claims like those of Vilenkin were getting any attention was that they had some novelty. Surely after a few more years of attempts to extract a prediction of some sort led to nothing, the emptiness of this sort of idea would become clear to all and everyone would lose interest.

Eleven years later I’m as baffled by what has happened to the field of fundamental physics as I’m baffled by what has happened to democracy in the US. As all attempts to extract a testable prediction from the multiverse have failed, instead of going away, pseudo-science has become ever more dominant, with a hugely successful publicity campaign (including a lot of “Fake Physics”) overcoming scientific failure. Now this sort of thing is moving from speculative pop science to getting the status of accepted science, taught as such to undergraduates.

Many are worried about the status of science in our society, as it faces new challenges. I don’t see how the physics community is going to continue to have any credibility with the rest of society if it sits back and allows multiverse mania to enter the canon. Non-scientists taking science classes need to be taught about the importance of always asking: what would it take to show that this theory is wrong? how do I know this is science not ideology?

Any student who reads this textbook and looks for answers to these questions in it will find just two “tests” of the multiverse proposed:

• Look for evidence of bubble collisions.
• Believe this paper, and then if you find a black hole population with a certain kind of mass spectrum, that would be evidence for the multiverse.

Of course there is no evidence for bubble collisions or such a black hole population, but these are no-lose “tests”: no matter what you observe or don’t observe, the multiverse “theory” can only win, it can never lose. Is it really a good idea to teach courses telling college students that this is how science works?