blevin

Sounds fascinating. What coordinate systems are used for treatment planning? Given how much bodies can change over time, and the difficulty of re-achieving a specific pose, I'm curious if there are interesting ways to correlate measurements over time. Certainly, medical training involves learning lots of prepositional anatomy words like "antecubital" but is there anything more precise, a GPS system for bodies? This seems very challenging for e.g. the gastrointestinal tract -- but I could imagine something using lots of relative reference points, the way I assume surgeons orient themselves.

Due to the designed complexity of our tax code the IRS cannot compute it both correctly and efficiently across the taxpayer base. Instead, they put the burden on the taxpayers, with stochastic validation and penalties (aka audits).

For my part, I hope someone out there is managing to claim those Ottoman Turkish Empire Settlement payments. All I know is that the Empire hasn't existed in 100 years but is something my tax code asks me about every year.

As self driving cars become more common I'm curious if we will see an opportunity to shift to more globally efficient cooperative behavior, or if the need to interop with human-driven cars means the same incentives and behavior patterns will persist.

Very much agree on these points. In my experience the pragmatic path for larger projects has been a hybrid of keeping visual programming graphs self-contained -- roughly within the "asset" boundary -- and then using a hierarchical format with textual representation to assemble those assets and drive their interface parameters. This is roughly how animation studios use USD (openusd.org), for example. Arguably this is just a strategy of combining a couple domain specific languages, with edges where the optimal tradeoffs of each domain flip. It's a very interesting question whether these limitations of visual programming are essential complements of their benefits, or if there is some good way to provide better abstraction, re-use, etc. Certainly Houdini, Nuke, Katana, etc. all provide limited forms of those things (ex: Houdini OTL's; scripts that version-bump nodes and try to auto-upgrade the parameters), and they do see lots of use in industry.

Knew someone who worked at PG&E. Her department's task was to support calculation of whether it was financially optimal to obey the law or risk paying fines, across PG&E's service footprint.

JB Straubel's Redwood Materials is another interesting piece of that picture.

Some do, including penicillin:

“Antibiotics that cross the BBB tend to have certain physical characteristics including: small molecular size, being lipophilic, binding to fatty molecules on cell membranes, low plasma protein binding and other specialized characteristics“

http://lymemd.blogspot.com/2013/03/blood-brain-barrier.html?...

Same with sugar in cigarettes, even. Sugar is a major ingredient, and makes them easier to inhale compared to cigars, which partly accounts for their global popularity.

"Humans are designed not to be happy, but to pursue happiness -- including wealth and status. Unfortunately, such a pursuit partly promotes unhappiness. Discomfort with our current lot gives us the drive for absolute & relative success -- an evolutionarily useful trait." from a footnote in Anti Ilmanen, _Expected Returns_

Rather than adding fixed channels, many cinemas moved to Dolby Atmos, which has 128 audio streams with spatial placement metadata. Those are rendered to whatever speaker array is available.

https://www.dolby.com/us/en/technologies/cinema/dolby-atmos....

Philosophical Economics has a thoughtful essay on this: https://www.philosophicaleconomics.com/2016/05/passive/

I’m going to argue that the trend towards passive management is not only sustainable, but that it actually increases the accuracy of market prices. It does so by preferentially removing lower-skilled investors from the market fray, thus increasing the average skill level of those investors that remain. It also makes economies more efficient, because it reduces the labor and capital input used in the process of price discovery, without appreciably impairing the price signal.

Sugar is also a major ingredient in cigarettes. In some brands it is the #2 ingredient by weight.

http://cebp.aacrjournals.org/content/27/3/357.1

Know any good writing that introduces the game item market to outsiders? I am curious about the thought process of people who spend money in this way, and their perception of value.

I imagine a taxonomy of motivations, like the one Larry Harris has in _Trading & Exchanges_. Presumably some people get entertainment value (because surely there is a kick in seeing a game's world modified beyond its normal rules of state mutation); some speculate on price trends (hence the price graph on opskins.com; what are time series graphs for if not to spark idle dreams of avarice), and others getting some sort of utility (i.e., money laundering).

Perhaps we will end up with a standardized validation simulation model that autonomous vehicles must pass for licensing. The test suite could grow as more anomalous scenarios are analyzed (ex: acceleration into gore point; white-painted trailer truck turning across divided highway; debris on highway), and it would ensure that any new vehicle on the market or OTA update would meet a minimum of safety. Manufacturers could compete on test suite coverage and acceleration curve stats, and market using rendered video of simulation results showing superhuman skill.

A decade ago a fuel-tanker truck crashed on a freeway overpass in Oakland, and the ensuing fire softened enough steel for the overpass to collapse:

https://www.nytimes.com/2007/04/30/us/30collapse.html

The overpass was reconstructed and I-580 highway reopened after 26 days:

https://www.sfgate.com/bayarea/article/A-MAZE-ING-His-reputa...

Alternately, it's amazing and inspiring what just happened; other high-end car makers with a century of experience have shipped imperfect brakes for years (1) and only redesigned them on new models in response to complaints. Tesla, by comparison, is applying the Apple model (when Apple is at its best) of continuous over-the-air product improvement, which can change one’s sense of a product's value for money.

1. MERCEDES CANCELS BY-WIRE BRAKE SYSTEM; DECISION A BLOW TO TECHNOLOGY'S FUTURE. "When the system failed, the hydraulic system took over. But that resulted in a longer stopping distance and additional brake pedal effort by the driver." http://autoweek.com/article/car-news/mercedes-cancels-wire-b...

Thanks for this response. My untrained guess is that precessional torque would cause more loss than air friction, assuming the wheel is cleanly symmetric (ex. no bolt heads sticking out). It's fascinating how many paths there are to push something up various kinds of potential energy gradient in a way that it mostly stays there by itself but is still available to us to access in a controllable way.

Wikipedia claims flywheel loss rate circa 2013 of 5% per day. https://en.wikipedia.org/wiki/Flywheel_storage_power_system

That compares to recent estimates of Tesla li-ion loss rates at under 5% per month -- 0.16% per day.

Amber Kinetics is one company building fixed flywheel storage products. http://amberkinetics.com/

They have one 32 kWh, 5-ton, 98% steel flywheel installation on Oahu; pictures here: http://amberkinetics.com/hawaiian-electric-and-amber-kinetic...

Are buried flywheels used in practice? Is the flywheel axis aligned with the Earth rotation axis? Do they pull a vacuum around it?

Unlike dividends, this channel of return seems distributed according to shareholder's trading savvy. It's often said that a company is overpaying for its shares (1) but the phenomenon minus narrative often seems that the company is intentionally playing the losing side of the trade to offer up value to shareholders who sell. This makes it an active exercise to capture that value, compared to a rule-allocated return like dividends.

Another interesting thing is that this form of buyback return channel would seem to exclude passive index funds unless they were actively shrinking (i.e. selling).

(1) Example from this morning talking how GE bought $30B of shares and later tanked. That's a pretty significant although unequally distributed return of value to those shareholders who sold vs. those still owning GE. https://finance.yahoo.com/news/general-electric-company-stoc...

> They call it “returning money to shareholders.” I call it “wasting shareholder money,” because they almost always buy at prices that are too high

You may enjoy reading Paul and Shou-Ching Jaminet's book _The Perfect Health Diet_. It includes an approachable survey of results for many supplements, as well as a considered take on why multivitamins seem less crucial than advertised.

One reason: metabolic interactions mean that proportionality matters, for example between A and D. Multivitamins are sometimes optimized for cost and pill size at the expense of this. Another reason is that people have varying baselines based on their diet, as well as how much micronutrient synthesis is happening in their gut. Because each vitamin (aka crucial micronutrient) has a dose-response curve, both too little or too much is bad; the book tries to chart out what is a good range for each and how you can achieve that through diet, lifestyle, and supplements. It is by no means a perfect book, but it is the best combination of approachable, credible, and well-researched information that I've come across.

The complexity of untangling these kinds of interactions is why nutrition research often seems to yield results that contradict earlier results. Biology is complex, but that does not mean progress in understanding it is impossible.

It's not even about a product category. It's about control of the surface area of interfaces people use. Interfaces that aggregate other services often hang poorly, but if you can build one people seek to use, you can build a brand (a name bound to positive connotations), which becomes a durable and valuable position to control. Mental real-estate in mass awareness is hard-won.

Voice control today is somewhere between gimmick and strictly dominant, depending on the situation and provider. If you consider the trends of interface design from line console to GUI to touch, each used newly-sophisticated tech to become more broadly accessible. With this broadening the stakes go up. I think that explains the proliferation of speaker chatbot products; they are just one front in a larger competition for that space people interact with directly. That particular front may or may not turn out to matter in the long run. Concurrent battle sites are car integrations and streaming devices/USB-sticks for TV's. This is why these same companies keep showing up to the same battlefronts with more or less equivalent strategies (semi-shiny new interface tech that happens to funnel you into their marketplace regime); they all want to be your concierge at the ready, in order to steer economic activity.

The Kitty Cafe concept, where you can pay to have tea and play with cats in person, is sort of an AWS equivalent for the cat ownership experience. It's fascinating that tea is an expected cofactor in this relationship.

I work in CG animation, and one thing artists do when procedurally generating geometry is tweak the size distribution of things meant to mimic natural phenomena, such as leaves and rocks. They are typically assembling these as expressions, so it's common to first reach for a simple uniform distribution in the form of scale*rand(), but more experienced folks know to go straight to a power law instead. I think it's fascinating that such an extremely high-level way to characterize natural processes could make it into an artist's toolkit like that.

> It seems like C++ developers either don't give a shit about code quality or they do, and don't make many mistakes anymore. Either way there is little demand for a static analyzer

Well, this is like claiming that authors writing in English don't care about grammar. There's too many people doing too many different things with the language -- from texting LOLs to writing academic research -- to generalize. But speaking from experience with one ~5M line industrial C++ codebase continually developed since the 90's, we care enough about quality to have a whole set of measures, automated and cultural, to support it.

We've looked at static analyzers several times but none of the evals have found issues worth the price. It turned up things like a static expression to compute a bitmask where the same flag was being listed multiple times. Which is nice, but made it feel more like a lint tool in terms of differential value added for us. If we didn't have things like valgrind memcheck on continuous integration, it might be a different story.

Can anyone recommend scraping adapters (businesses or tech) that are robust to this sort of thing? I'm talking about something higher level than, say, Beautiful Soup -- something you can configure to point at an endpoint, essentially request a sql row subscription from it, and not have to mind it too much.

Both the traversal/retrieval and data-interpretation parts seem to have interesting aspects when you consider current website design. Some websites make themselves hard even for humans to read (consider why safari reader mode exists).

This seems like a potentially valuable service, in the sense of being a schlep. I wonder how many places have home-grown scraping efforts as part of their business and how annoying it is for them to maintain.

> They seem like awfully indirect measurements of relevant skills and perspective

Well, maybe that's the thing: they aren't measurements of relevant skills or perspective. Instead, you might consider it a form of stratified sampling: trying to include sub-populations that may not be as well-represented so far but may supply valuable contributions.

Sounds like you might appreciate LFT's Parallelogram fpga demo:

http://www.linusakesson.net/scene/parallelogram/

Anyone needing an idea for a hack project: make a realtime version of this for phone or webcam. Bonus points for swapping in ridiculous sound sets, or keying augmented-reality visual fx overlays.

If anyone reading this is attending SIGGRAPH, several of us will be presenting a production talk for Piper on Sunday 2PM.

http://s2016.siggraph.org/talks/sessions/life-shorts

You are right -- that's a count of feather curves (quills & barbs), not feathers. We do not model the barbules -- though there has been interesting work done in trying to calibrate a microfacet BRDF for feathers by tracing against barbule-level details.