Chris X Edwards

Sometimes I feel like the main utility of me physically coming in to work is to change the water cooler bottle. Too heavy today. Recovering.
2017-07-21 08:22
Misheard "robot planning" as "rabbit planning". Ok, that is quite different.
2017-07-20 15:20
Saw some PR noise about some firm's "evolutionary leap forward". Does this mean it is likely not to survive this change à la evolution?
2017-07-12 13:36
Did you see Facebook's comment? Commenting on someone else's comment? We're all now being watched by the worst gossiper in history.
2017-07-12 12:19
Misheard "machine learning" as "mushy learning". No loss of meaning.
2017-07-08 08:37
Etc.
--------------------------

The First Rule Of Idiot Driver Deathmatch

2017-07-15 13:16

Oh no, it’s not that. You can can talk about Idiot Driver Deathmatch all you like. In fact, I think it’s helpful to and so I will. The first and only rule of Idiot Driver Deathmatch is this:

DO NOT DIE!

I am currently recovering from one of my most serious brushes with non-compliance of this important rule. As I mentally enumerate the long list of bicycle crashes I’ve experienced, this one was the most disturbing.

I had woken up early to beat both the heat and a decent percentage of the idiots. The light was pleasantly not bright but it was about 10 minutes past dawn and visibility was good. So I thought. I live at the top of a non-trivial hill (~100m elevation from sea level). I am always very careful descending it. So much so, that I usually take side streets to avoid the main road’s busy traffic and some known vicious road defects (note that I know the condition of this road intimately). Eventually however, to reach the safety of the park down by the bay, I would need to cross the freeway.

Freeways are like great rivers, impassable except at bridges. This meant that to cross the freeway I would need to return to the busy main road. As I approached the last street that would take me to the main road, I saw a car approaching from the other direction. This car is special because it left one of the last clear memories before things became interesting.

Before seeing that car I remember I was mentally evaluating the literary merit of Metallica lyrics. No conclusion on that important issue was reached because a high priority thought suddenly filled my mind, "What is that idiot doing?" The car I had seen was driving stupidly. I can not, unfortunately, recall exactly what the problem was, but it was red flagged. I’m pretty sure there was both a speed and a trajectory problem. I want to say it was driving strangely slowly and kind of over too far as if it was going to pull over at any moment and send a text or something. But it just kept driving like that. He (or she) turned right on to the last road to the main road and I turned left, incidentally following.

This car was so disturbing to me that I decided to cut through the car wash to avoid it at the light where it was going to be waiting to turn right. I popped out of the car wash and on to the main road. I’m sure I passed the problem car. I know from hindsight that at this point I was 5 seconds or so from a crash. Because I did crash one might suppose I was mentally not properly prepared. I feel I can reject this because I know this piece of road to be one of the most treacherous and lethal. That I’ve ever encountered. In the entire world. And, I know this freeway crossing very well. I never take it lightly. I can not stress how horrible and dangerous this 300 meters of cycling is. At this exact spot, there is even a "Share The Road" sign which should be interpreted by all wise cyclists as "Cyclists Forsaken, You Are Going To Die". (My wife stopped cycling because of this freeway crossing.) (One of the most terrifying moments of my life was a completely different occasion in this exact same spot where I was high-watt sprinting up to the freeway bridge, riding apace with and between the cars turning right and the ones going straight; my chain snapped. By some stupendous miracle, I did not go down. I take this piece of road extremely seriously.)

The last very clear memory was getting around a parked car west of the car wash and then looking back to evaluate what my liability was, i.e. what kind of traffic was back there. I know it was either three or four cars. Presumably the flagged idiot was there but no positive ID on that.

The next thing that happened was I found myself looking at my garage’s keypad. I typed in the code. The next thing that happened, I was ringing my own doorbell. The next thing that happened was my wife answered the door and I was standing there confused and, it was now apparent to me, pretty badly injured.

Demonic Possession

This is a great opportunity to say some things about demonic possession. As a rule, I do not believe in supernatural things but I do believe in demonic possession. It turns out those are not contradictory. I have seen demonic possession up close and in person and, wow, it is something you won’t forget. If you see it you’ll probably want to write about it in any foundational religious texts you’re writing. I’m calling it by the traditional name, "demonic possession", but let’s consider what the putative actual non-supernatural effects are. Here’s Wikipedia’s conception of it.

Descriptions of demonic possessions often include erased memories or personalities, convulsions and fainting as if one were dying. Other descriptions include access to hidden knowledge and foreign languages, drastic changes in vocal intonation and facial structure, the sudden appearance of injuries or lesions, and superhuman strength. Unlike in channeling, the subject has no control over the possessing entity…

I have seen first hand the vocal intonation and facial structure change in someone "possessed". I completely understand how that phenomenon would be ascribed to "the devil" or some malevolent supernatural entity. Witnessing this first hand has been one of the most interesting and disturbing things I have ever seen. I am convinced that much of the extraordinary behavior described in religious texts was inspired by this phenomenon. If you have faith in supernatural causality, then it’s easy to see a connection. If you have faith in an utter lack of supernatural causes, I believe it is even easier to find an explanation.

After witnessing this phenomenon on several occasions I was left with a powerful feeling of, "Well, that was very weird and a complete mystery." Later in life I had the good fortune to read The Origin of Consciousness in the Breakdown of the Bicameral Mind by Julian Jaynes. I would like to quote Richard Dawkins' reaction to it because it is exactly mine too.

It is one of those books that is either complete rubbish or a work of consummate genius, nothing in between! Probably the former, but I’m hedging my bets.

Whether the book is true, not true, or somewhat true is not so important. What can not be denied is that the book inspires fresh thinking about topics like demonic possession. (Note that the topic of yesterday’s post, Neal Stephenson, was specifically inspired by Jaynes' work when he wrote his most famous book, Snow Crash.)

Jaynes was quite specific in how he divided up the mind’s architecture (into two chambers, naturally). My hypothesis is simply that the mind is made up of many different cognitive actors. No more specific details are required to further hypothesize interesting ramifications. If even vaguely correct, this is of extreme importance to AI researchers seeking to synthesize human-like behavior.

I will summarize the necessary important components of my belief in demonic possession for the purposes of this story. 1. There are no "demons" per se. 2. The "demon" personality is really a cognitive actor that in normal circumstances does not control the person’s output (facial expressions, voice, body language). This is why the rare cases of interacting with it directly are so disturbing (like the Uncanny Valley). 3. Some cognitive actors can take control of some tasks in normal operation. For example when driving, you can have a conversation with someone while a different (but not at all inconsequential) cognitive actor does the actual driving.

If that seems freaky to you and the idea of "you" actually being a team of cognitive contributors seems wrong or weird, consider that the physical architecture of the brain is definitively configured that way. Jaynes' book is perhaps most useful to really get an internal appreciation of that fact. Jaynes felt like much could be explained by the distinct left and right hemispheres of the brain. And that may be so. I personally suspect that various concentric layers are just as important in these matters (medulla, amygdala, cerebellum, cerebrum, neocortex, etc.).

One of the reasons The Origin of Consciousness has been such an important book to me is that after reading it I started to view each person I encountered as if they were a "team". I found that explained their behavior much better than the ordinary view of personality, the one that inspired singular pronouns (I, me, he, her). This is especially true of "crazy" people and less apparent with well adjusted people. But the model never seems to fail. I began to appreciate this mental architecture in my own mind. I believe that the foundation of good mental health is optimal teamwork between cognitive actors.

I do not believe that demonic possession involves supernatural agents and, furthermore, I am on pretty good terms with all of my cognitive actors who generally all work very well together. This is why the pieces fell into place quickly.

Where were we in the story?

I was at home, face swollen and bleeding, serious road rash on my entire right side, head to foot. My left knuckles were bruised. Most serious, my shoulder was quite smashed. I watch enough pro cycling to know how they do a rough field check for a broken collarbone and I was failing that check.

(Know why my shoulder looks blue in this photo? Because it is blue.) pretty_beat_up.jpg

W… T… F…?

I knew I had been on the bike. The road rash was an unmistakable ancillary byproduct. What I did not remember was the crash. Nor did I remember how I got home. One possibility was that someone gave me (and my awkward heavy bike) a ride home while I was not conscious. But that was not a satisfying explanation. How would they know where I lived or to put me in front of my non obvious garage? The second possibility is that "I" walked the bike back up the hill. I did not remember doing that. The final possibility, which I now believe is confirmed, is that "I" rode the bike back up the hill without the benefit of the cognitive actor which is writing this account (i.e. the topic of Metallica lyrics was on hold).

For the last two days since the incident, I have walked down the hill retracing my route and trying to remember what exactly transpired. This morning I went down at the same time of day. I am now certain that I must have crashed within 20m of the "Share The Road" sign. I focused on possibilities of how I could have crossed the main road to get home and I had an insight. I remembered pushing both of the crosswalk buttons at the car wash light.

What’s so fascinating is that when you consider where Grim Chris (as I shall refer to him) dropped me off, a theme emerges. Right in front of my garage keypad. Right in front of my doorbell. Grim Chris apparently does not like buttons. And this was likely why the memory of pushing both of the crosswalk buttons was the anchor I used to reconstruct the event. (I am re-evaluating "my" interesting strong aversion to push button shifters.)

This was a tricky reconstruction because I’ve ridden up this hill hundreds of times. It would be easy to introduce past memories. But as I re-walked the route, Grim Chris glared in the direction of a smoker who had been seen that morning in one of the units of my ("non-smoking") apartment complex. Grim Chris, apparently, does not like smokers. That memory was clearly there and published to my entire mind.

I am now pretty sure how I got back home. I suppose by now Grim Chris is just pretty good at riding a bicycle up hills while ignoring pain. Good job, Grim!

The big question that remains is what exactly happened? In my hundreds of diverse and spectacular bike crashes, I have never had one that I can not remember. But reviewing the scene this morning is pretty damning. The last solid memory of a deliberate conscious action was checking behind me to see what evil I would be fighting. I know that I registered 3 or 4 cars. Such quick assessments are something I am very good at. And then… blank.

I know that I was thrown down extremely fast. This was the kind of crash that can easily kill you. The side of my head took a pretty hard hit. I am pretty sure that if I had not been wearing a helmet I would be either in the hospital right now or moved on to the crematorium. My right arm was pinned under me; it never had a chance to react. My left hand apparently swung around and punched the road to counteract the rotational velocity of my 20mph+ body suddenly making horizontal contact with the road. I am thankful for my obsession with wearing gloves (for just such occasions) as my hands kept most of their skin.

The next question is what drops a cyclist that fast? The answer is one of three things.

  1. Spectacular catastrophic mechanical failure of the bicycle. The breaking chain of my earlier incident for example. I’ve also had broken pedals, cranks, shoe cleats, handlebars, and stems which have had stunningly swift and vicious effects. Forks, skewers, etc. could also have done it.

  2. A serious road defect. One of my most memorable crashes was quite a teachable moment for me. I was riding up a hill. Slowly, as one does. And the next instant, I was on the ground. Whaaa? It turned out that there was a crack in the road just wide enough to grab a wheel and give it a surprising twist. I don’t need to even mention ice and road paint in the rain (which is even more insidious than ice).

    I do not believe either of those factors were in play. As I have said, I know this road extremely well and going down to the scene confirms there is nothing that would snag a wheel. The bike is in good shape. (Lovingly protected from damage by the right side of my body!) That brings us to number three.

  3. A motor vehicle.

Our familiarity with cars makes them seem like a part of us. But they are not. They are very literally superhuman. Their strength is unimaginable. It must be harnessed with engineering, not imagining. If even a small part of a car’s active energy is transferred to you, it is very likely that you will die.

Standing at the scene of the incident this morning I watched the traffic for quite a while. The northbound ones usually make a pretty linear transition from their lane to the right turn lane just after the parallel parking zone. But the parking zone is a bit nebulous. On a bicycle I must exactly mirror (not match!) and cross this path of cars. Under normal conditions, I simply must count on the cars to not just mow me down in cold blood. (I too often wear road colored clothing but on the morning of the crash, I was wearing a bright yellow shirt). We know there were three or four cars behind me and that one of them had already been flagged as driving badly. We know there was nothing in the straightforward road that should have given me spontaneous problems. The only conclusion I have left is that I was tapped, likely on the pannier which is on the left. And this put me on the ground in what to human timescale would be called "immediately".

Although I barely complied, it seems I did just follow rule one of Idiot Driver Deathmatch (barring some unperceived latent intracranial edema). My superhero imaginary friends saved my ass and took good care of me. I am thankful for that. One thing my entire team is completely intent on is the fact that Idiot Driver Deathmatch is not a one way street. I am coming for you, Idiot Driver. Grim Chris is coming for you. He can be pretty damn intense. I hope, Idiot Driver, that someone takes you out of the equation soon. If need be, however, I will knock you off the road personally. Now then… I need to get back to work studying autonomous car technology. And now you know why.

Review: The Rise And Fall Of D.O.D.O.

2017-07-14 19:08

This is a book by Neal Stephenson and Nicole Galland. I’m a huge fan of Neal. It’s safe to say he’s one of my favorite fiction authors. I liked this book, but I didn’t love it as much as some other Stephenson books.

The first irritant was some little detail problems. For example, on p670 the number "31142" is quoted as the number of square miles needed to search for a fugitive who can run 100 miles. That may seem overly nerdy to pick on, but this is a Neal Stephenson book.

On p647 we are told about some "large" vikings. Not large with respect to other vikings, but large with respect to modern people. That smells wrong to me since I have walked through many art museums and have seen thousands of suits of armor and historical garments. The fact is that finding Vikings who would seem large and imposing to modern Americans would be very hard in my opinion. Neal, with his LARP hobby (which makes an obligatory appearance), should know that.

And of course there is this rule of fiction writing which I just made up: if you do not want your plot to get into a muddle, do not write about time travel! Of course, Neal is a special writer and if someone is going to pull that off, he’s probably going to do it as well as it can be done. I think he did a great job with it, but I’m still convinced that it is more trouble than it’s worth. Neal and/or Nicole took the approach of using a combination of two popular fiction genres, Harry Potter and quantum mechanics esoterica. In the story there are magic-using witches and their magic has something to do with parallel universes as described by some of the weirder branches of quantum mechanics. (I always like to say that in some other universe, quantum mechanics has a better explanation.)

Ok, so there’s time travel. It’s explained away about as well as the plot of "Interstellar". I’ll suspend disbelief to be a good sport. So what do our main characters do with this power? They concoct the most byzantine (literally going to Byzantium to explain that word) Rube Goldberg missions. It just seems like such a clever author could have thought of less complicated ways to use time travel to, say, make money. For instance, instead of trying to go back in time and plant a mundane contemporary item that is very valuable today, why not just go back to some known treasure loss and figure out exactly where the lost galleon sunk?

Oh well, best not to worry about the details. In fact, with respect to some of the inevitable time travel plot muddle, p725 says, "..don’t think about it too much." Indeed.

Brooks vs. AI: No Point In Qualming

2017-07-12 08:51

I was looking over this IEEE Spectrum survey of famous smart people’s opinions on AI and I just had to laugh at Rodney Brooks' response. Actually, it’s the way he put it that was funny. The sentiment is the exact same that I share on the topic. What’s so funny is that he takes special digs at Ray Kurzweil. I’m in total agreement, having declared Kurzweil a "crazy futurist" back in 2015 (and if I declare someone a crazy futurist, their futurism has to be pretty damn crazy).

I love how he totally deflates this question.

When will we have computers/robots recognizably as intelligent and as conscious as humans?

Not in our lifetimes, not even in Ray Kurzweil’s lifetime, and despite his fervent wishes, just like the rest of us, he will die within just a few decades. It will be well over 100 years before we see this level in our machines. Maybe many hundred years.

Then there’s this sensible answer which basically says that long term prognostication is more of an artform than a science. Still needling Ray!

How will brainlike computers change the world?

Since we won’t have intelligent computers like humans for well over 100 years, we cannot make any sensible projections about how they will change the world, as we don’t understand what the world will be like at all in 100 years. (For example, imagine reading Turing’s paper on computable numbers in 1936 and trying to pro­ject out how computers would change the world in just 70 or 80 years.) So an equivalent well-grounded question would have to be something simpler, like “How will computers/robots continue to change the world?” Answer: Within 20 years most baby boomers are going to have robotic devices in their homes, helping them maintain their independence as they age in place. This will include Ray Kurzweil, who will still not be immortal.

And then he takes a jab at Nick Bostrom who is a smart guy and a refreshingly creative contemporary philosopher. However, I also agree with Brooks that Bostrom’s AI squeamishness is premature.

Do you have any qualms about a future in which computers have human-level (or greater) intelligence?

No qualms at all, as the world will have evolved so much in the next 100+ years that we cannot possibly imagine what it will be like, so there is no point in qualming. Qualming in the face of zero facts or understanding is a fun parlor game but generally not useful. And yes, this includes Nick Bostrom.

On the topic of AI taking over a world I’ll still be living in, I too am qualmless.

Review: Messy

2017-06-29 13:11

I just finished Messy—The Power Of Disorder To Transform Our Lives by Tim Harford. It was an ok book. The basic premise is that if things are too neat or too well organized, it can be easy to stop paying proper attention which can cause accidents or prevent creative problem solving. Etc. The book was well done and made that case pretty well across a lot of diverse fields.

However, I quickly started to feel some philosophical discomfort. To start with, the book is pretty well organized. I usually very much appreciate a book where the chapters are neatly organized into predictable numbers of pages. This book’s chapters had page counts whose population standard deviation was exactly 2.0 pages which is remarkably consistent. The copy editing was fastidious. The grammar correct and orthodox. The material was meticulously supported by 17 pages of extremely fine print end notes, all immaculately organized. The author says directly that he has worked on this book for five years and it is apparent that it was not dashed off thoughtlessly.

There were even times in the material where the truth started to sneak out. For example on page 198, he talks about mistaken identity causing people to get listed on the creepy unconstitutional no-fly list and he says, "…we should [listen to victims of unusual errors] and set up mechanisms to sort these errors out quickly." At this point I realized the problem—for someone, the no-fly list itself was a mechanism to sort out "errors". One man’s disorder is another’s order.

Even the book’s flagship example of a mildly famous musician’s technique to inspire creativity was questionable as an example of the premise. The idea is that the musician had these cards with some semi-random directives on them and when creativity was felt to be lacking, a card could be taken and the (messy) random directive pursued or contemplated. The problem with this example, is that this technique is a technique. This very structured system is extremely clearly defined and conceptually tidy. Contrast, for example, musicians who might have a whole lifestyle where they constantly go to random parties, meet random people, and do random drugs. Fans of Kieth Richards and Lemmy Kilmister know that works too. Fans of Bon Scott, and John Bonham however, know that this strategy can easily produce limited success (ending in choking to death on your own vomit in your early thirties).

I personally divide neatness into these three categories.

  • Engineering competence

  • Aesthetic

  • Cargo cult

Engineering neatness is neatness that is required to do something properly. Engineering neatness has an explanation, a good one, why it is beneficial. Like not using your fingers to clean a camera lens but using your fingers to clean a gauge block.

Aesthetic neatness is just that weird impulse some of us sometimes have to have for things to be organized, put away, etc. Think of this like those people who stack rocks or make sand mandalas. It’s just something some people like. Go for it! No justification needed!

Cargo cult neatness is mostly found in large corporate environments and recommendations from the AMA. If you don’t know what a cargo cult is, it is extremely interesting. In this type of neatness, though there is little understanding of the mechanism, if some other people had some sort of success with some kind of neatness in a similar situation, it is believed that it should be applied liberally and universally. To be fair, this book did try to focus on this kind of thing (for example, making people have clean/tidy cubicles for no real pragmatic reason turns out to be counter productive).

So what’s the right answer then? I’m well convinced that people with an overzealous compulsion for tidiness are not doing themselves any huge favors. But I still feel that if you have to get stranded partway up the climb to the global maximum, the neatness side of the slope beats the messy side. My advice is keep your eye on the ball. If having a messy desk is getting the job done faster, have a messy desk. If the job itself is cleaning the desk, then tidy the desk. This book was like a guide to help anorexics eat more. That is simply not the normal problem normal people have. But if you have that problem, finding that wholesome sensible balance is just as valid.

Emissions Cheating Endgame

2017-06-20 09:46

I love reading The Morning Paper where @adriancolyer translates computer science papers into English. With my keen interest in automotive software I took a special note of this excellent paper.

Only after deciding to write about this fascinating work did I notice that this was from my friend, the brilliant and witty Kirill Levchenko at sysnet.ucsd.edu.

I don’t have anything techincal to add to this fine work, but reading TMP’s take on it, I have to agree that the endgame here is one of two things. First, there could indeed be an AI race to constantly game the system. Currently there are some fairly crude, but entirely effective, heuristics to determine when the system is being watched (tested). But one could imagine a more sophisticated testing regime that required the wheels be turned randomly and the speed varied. An AI classifier could be conceived of which would still endeavor to cheat, and the race would be on.

The other option, and this seems like the just approach, is to not test the car occasionally (using extrapolation to find out how much it likely pollutes). No, the correct approach is to mandate full time pollution accounting. Your car should have a "measured polution" counter that just tallies it up and you get sent a bill accordingly.

The next topic, of course, would be cheating the pollution sensors. Don’t worry, there could be montoring sensors to ensure the correct functioning of the pollution sensors. And other turtles.

--------------------------

For older posts and RSS feed see the blog archives.
Chris X Edwards © 1999-2017