Bayesian Confirmation Theory is Not Science

Yesterday Edge published the results of their annual question: What do you consider the most interesting recent scientific news? What makes it important?

The Large Hadron Collider run was mentioned several times with different interpretations, depending on how you define “science”.

collider

What is the Large Hadron Collider Doing?

Looking for superparticles.

General relativity and quantum mechanics are currently incompatible theories. Relativity accounts for gravity in the realms of planetary orbits and galaxies and an expanding universe. Big stuff. Quantum mechanics uses electromagnetism and strong and weak nuclear forces to describe the interactions between subatomic particles. Tiny stuff.

The theories don’t scale. If E=mc2, then the amount of energy from particle interactions results in so much mass that the universe should have imploded into a black hole by now [1].

A supersymmetric model unifies the theories: Every existing elementary particle has a corresponding superparticle with opposite spin. Superpartners. In theory, at high energy, colliding particles turn into the masses of their superpartners and this emits a Higgs boson.

The Higgs boson was discovered in 2012. The next step is to find a superparticle.

What was the Scientific News?

The Large Hadron Collider has thus far found zer0 superparticles. But this year, a new run began that allows even higher-energy collisions.

I have not failed. I’ve just found 10,000 ways that won’t work. –Thomas Edison

For string theorists, this is great – We have ruled out the weaker section of the energy scale and can narrow our search! The probability of finding a superparticle on this next run has increased.

Or maybe superparticles don’t exist?

How to Reach a Scientific Conclusion

We used to rely on falsifiability. Scientific theory: All swans are white (proven false by observing a black swan).

But falsifiability doesn’t work for modern physics. For example, no one has ever observed an atom. Instead, scientists apply Bayesian confirmation theory, which assigns a confidence level to the credibility of a hypothesis.

So we have atomic theory, which is approximately 100% confirmed; and then we have string theory, which is trace-amounts confirmed.

Is modern physics is even really science [2]? Science asks the questions that are answerable; philosophy asks the questions that are not. By treating unanswerable questions as answerable, we find ourselves chasing hints of particles that might lend an extra 0.00001% confidence.

And who cares about that? Who cares if we are 0.00001% sure that every existing particle has a corresponding superparticle? If that was the frontier of string theory, string theory would be boring. Modern physics is interesting not because of its data, but because it gives us a better way of understanding reality than classical mechanics. It tells us that space-time is curved, that things behave differently at a quantum scale, that we can think of point-like particles as one-dimensional strings. And the goal isn’t to find affirmative evidence beyond some arbitrary threshold, but to continuously formulate the most reliable way of thinking at the present level of knowledge.

And that’s exactly what the ancient philosophers were trying to do.

See Also:
1. Relativity versus quantum mechanics: the battle for the universe –guardian
2. A Fight for the Soul of Science –Quanta
3. There is a 94% likelihood that the multiverse exists.
J. Polchinski. Why Trust a Theory? Reconsidering Scientific Methodology in Light of Modern Physics, Munich, Dec. 7-9, 2015.

Impro: Improvisation and the Theatre

impro

Today my boss asked if we could schedule a meeting about the server status.

Eff off, I’m on vacation, is what I wanted to say.

SURE WHAT TIME, is what I actually said. I have been conditioned for self-censorship.

Self-Censorship

Improv is action without censorship. Children are born without a filter between thought and speech, but censorship is externally enforced when we learn that certain behavior is Not Okay.

Eventually it becomes internalized, and people even censor their own thoughts. Sometimes I think some pretty messed up things and immediately worry that someone might have heard my brain. Adults struggle with creativity because they censor their imaginations for fear that it reveals their true selves. But the imagination is the true self.

Improv actors let automatic processes take over by removing self-censorship. Original thought should not require effort.

Status

People are like dogs. Every human interaction is a status transaction. A seemingly innocuous conversation:

    #1. How was your holiday? [lowers status by expressing interest in #2]
    #2. Terrible, I was working the entire time. [raises status by seeming busy]
    #1. Sorry to hear that. I had a ton of work too, but I was in the hospital all week with Dengue fever. [raises status by one-upping #2]
    #2. Oh yeah, I’ve contracted Dengue several times. I seem to catch it every time I go to St. Barth’s. [attacks #1’s status]

Even non-verbal actions communicate status. Eye contact raises status; averting the eyes lowers it. Slow movement raises status; speed lowers it. In movies, superheroes appear in slow motion, while Charlie Chaplin is sped up.

Spontaneity

To be spontaneous is to relinquish control.

A well-known rule of improv is Always say Yes. If the player opposite you says You slept with my daughter, you bastard!, it may be tempting to respond with No I didn’t! because you don’t feel comfortable playing the role of a lecher. Try saying yes.

Yes, turns out she’s really into butt stuff. Much like your wife.

Saying yes cedes control and allows the story to continue.

The intention of improv is not actually to be funny. What makes it entertaining is that it’s a reproduction of life without censorship.

Impro for Storytellers is Keith Johnstone’s second book. It’s a practical manual for theatre improv, and less relevant for those outside of Theatre Sports. Johnstone should have switched the titles of his books.

The Tail End

It’s the holidays and stuff, so here’s an important reminder from Wait But Why:

parents-small

The X‘s estimate the number of days a 34-year-old has spent with his parents, and the number of days he has left to spend with them, ever.

When you look at that reality, you realize that despite not being at the end of your life, you may very well be nearing the end of your time with some of the most important people in your life.

Spend wisely!

Read More:
The Tail End –WaitButWhy

Teaching Machines to Read, and Humans to Understand

Prismatic is dead. Sad!

Prismatic was a news reader that used machine learning to read and recommend content. Despite the fact that queries for “interest rate” often returned blog posts from ZeroHedge or sponsored content from credit agencies, it mostly did a good job.

Machine readers have come a long way. In the early days, neural networks processed data points independently. A translation machine would interpret a string like “Errare humanum est” word-by-word: To err human the.

A recurrent neural network feeds the output back into the hidden layer
A recurrent neural network feeds the outputs back into the hidden layer

Recurrent neural networks (RNNs) brought the ability to propagate past information. Words, with grammatical context, become sentences. When a machine has processed a subject and complement it knows to add a linking verb: To err is human.

A subset of RNNs, Long Short Term Memory networks, can persist early information without distortion from new inputs. By preserving long-term dependencies, a machine can connect sentences into a story.

To err is human. My wife could do no wrong. My wife is therefore…[not human].

Recent research has introduced the concept of attention [1], where a machine can selectively focus on data relevant to the question at hand. This model enables machines to answer queries on large bodies of text without trying to propagate dependencies from beginning to end.

Somewhere along the way, as machines gained the capacity for contextual comprehension, humans lost that same ability.

It used to be that humans could understand intent. We knew that words were clumsy vehicles used to communicate ideas, and so we learned to separate meaning from words. Now we can’t see beyond words.

This is not a memory-bound problem. College students in particular have excellent Long Short Term Memories, emphasis on the long. In fact, students at Harvard have identified that the Law School seal is derived from Isaac Royall, Jr’s family crest of two-and-a-half centuries ago, and that the Royall family owned slaves.

Harvard Law School seal
oppression

The seal—which adorns all of our buildings, apparel, stationery, and diplomas—honors a slaver and murderer. —Royall Must Fall

A computer might recognize that slave ownership is not Mr. Royall’s most salient feature – at least not where Harvard Law is concerned. A computer could identify that a slightly more relevant feature is Isaac Royall, Jr’s land donation, which led to the founding of the damn school.

By understanding dependencies, a computer can determine intent. It can separate someone else’s ignorance from its own. Maybe computers have more self-awareness than humans.

People say that a general artificial intelligence is still decades away. It’s clear that machines have surpassed human intelligence at least in reading comprehension and acting like a functional adult. Maybe humans should try to be more like computers.

See Also:
1. Hermann, et al. Teaching Machines to Read and Comprehend. Advances in Neural Information Processing Systems, 28 (NIPS 2015)

This Post Brought to You by Your Insurance Provider

Five-star hotel indirectly paid for by your insurance provider.
Five-star hotel indirectly paid for by your insurance provider.

I spent the last five days in Maui, thanks in large part to Gilead Sciences, Alios BioPharma, Merck, and Abbvie. This was a pharmaceutical-sponsored conference for medical doctors to learn about the results of clinical trials. And I’m not a medical doctor, but I happen to be related to one.

Some doctors were sponsored by their medical facilities as part of Continuing Medical Education. Whether bankrolled by drug company or hospital, all of it is ultimately tacked onto medical bills paid by your health insurance provider, so that’s something to feel good about.

Luau indirectly paid for by your health insurance provider.
Luau indirectly paid for by your health insurance provider.

Stuff I learned:

Clinical trials are conducted by neutral third parties. Neutral third parties whose clients sometimes treat them to lavish vacations in Maui.

Patents are weird. In the United States, a pharmaceutical company that patents a new drug is granted a 20-year term of exclusivity. Later, a second pharmaceutical company could discover a new use for the drug, and obtain a second patent. Now we have the same exact drug with two different patents. Each company can only market the drug for their own patented use case.

Thalidomide was patented in 1954 by a company called Chemie Grünenthal. It was used to treat morning sickness until a bunch of babies were born with missing appendages.

Decades later, Celgene discovered that thalidomide could be used to treat leprosy, which is apparently still a problem. Celgene obtained a new patent and now sells thalidomide under the brand name Thalomid.

Doctors have little incentive to prescribe cheap drugs. The first thalidomide patent has long expired, and someone could make a cheap generic. But the generic can only be used to treat morning sickness, because the leprosy use case is still protected.

A particularly charitable doctor could prescribe generic thalidomide for leprosy and call it “off-label use”. But did the generic manufacturer fly him to Maui?

Open bar, where your family doctor is double-fisting the night away.
Open bar, where your family doctor is double-fisting the night away.

It’s all about marketing. BigPharma invests heavily in marketing to make sure their drugs get prescribed. By “marketing”, I mean visiting doctors and showering them with gifts.

Oops, I don’t mean gifts. Medical institutions have policies against physicians receiving gifts. However, if a sales rep wants to take a doctor out to dinner at French Laundry, or invite his family to an all-expense-paid trip to Maui, that’s totally cool.

Moral of the story: Marry a doctor. Ideally one that specializes in infectious diseases requiring expensive drug therapy.

Dessert table and my brother's head, both indirectly paid for by your insurance provider.
Dessert table and my brother’s head, both indirectly paid for by your insurance provider.

*The FDA permits physicians to prescribe medications for other than their intended use. So a generic substitute for a patented use case is okay. However, this opens the physician’s medical facility to lawsuits from both the patent-holder and the patient.