October 19th, 2014

A few remarks about someone else's remarks about a disease that has been in the news lately

A friend posted a link to a blog post written by a nurse, about ebola. I didn't like it. (Are we surprised? No, we are not surprised.) I waited a while to respond, and chose _not_ to respond where the link was originally posted, because FB comments are just not great for the way I like to write about things.

Please feel free to go, ugh, I never want to hear about that stupid disease again and skip right on over this post.

I am not a health care professional, altho I am related to some; perhaps one or more of my friends and relatives will chime in with an improvement to this.

Here's what started me writing about ebola:


I have some really serious problems with this writeup.

The most significant problem I have is the idea that you could infect someone before you had any symptoms.

"Just like with the flu or hand foot and mouth disease, you can be spreading it to others before you show a symptom*(apparently not many see the *, so please read the elaboration at the bottom).”

You generally do have symptoms. In the case of the flu and ebola, the symptom may be a general run-down feeling and a low grade fever which isn’t definitively any one particular illness. But it is a symptom. I am trying to understand why we think an unenforceable travel ban would somehow be useful — but we aren’t talking about things like, say, asking everyone who has a low grade fever and a general run down feeling to maybe stay 3 feet away from everyone until they are clearly, identifiably sick with whatever they are sick with. You know, paid sick leave from the government, type of thing. That wouldn’t just help with the terrifying but low-probability ebola, but the less terrifying, more probable flu. Which we are now in the season of. It would also deal with the doesn’t-have-ebola-but-might-spread-it person, too (whether sticky-fingered or asymptomatic carrier).

In the lower part, the author talks about how someone can have matter from a sick person on them, spread it around, but never get sick from it themselves. This is true, altho the relevance to the discussion is pretty limited, and tends to spread fear more than anything else. The kind of contact tracing and isolation we are doing is actually addressing this kind of risk. Otherwise, they wouldn’t be bleaching the crap (literally) out of a middle school attended by kids who were on the same flight as someone who really did have ebola. Kids who in all likelihood, don’t have ebola and never will, just like the lab tech on the cruise ship, but who we are asking to limit the amount of people they have contact with until we know for sure.

The second problem I have is the idea that this is a disease we are unfamiliar with because of where it is from. Laurie Garrett (“The Coming Plague”) and Richard Preston (“The Hot Zone”) both did books about this 20 years ago that are still in print (and available in ebook form) and still sell well. A bit over wrought, and with their own problems, but anyone who read them Back In the Day is well aware of how ebola kills.

"It is foreign to the US, both literally and figuratively.”

"Imagine if cancer was infectious, and you lived in a country with zero cancer, and someone thought it would be a good idea to fly a few people in. I think there would be a different attitude.”

Here’s the deal. (For the moment, we’re going to ignore the fact that some cancer actually is infectious, because the author of the piece appears to know that and to have simultaneously forgotten it. Plus, cigarettes.) If you ban something that people really want to do, you’d better fucking well have an amazing plan for how you are going to enforce it. Because if you _let_ people do it while regulating it, you have a chance. If you _ban_ it, and there are ways to sneak around it, they will. And in the meantime, language that focuses on how it comes from Over There feeds into a lot of terrible politics and fear and bad policy and etc. that we don’t need more of. (viz Republican Duncan Hunter saying IS are crossing over the Mexican border and he knows because we’ve already apprehended some. Denied immediately by the Homeland, obvs)

The third problem I have, which could be the worst of all, but probably isn’t actually that big of a deal because everyone in this country is so terrible at understanding numbers anyway, is the idea that we know how deadly ebola is. We don’t. Here are some things we _do_ know. It is actually kind of difficult to catch. It is survivable _without_ modern nursing care, especially if your ancestry is from the region in which ebola has animal reservoirs. We’re just starting to really understand this right now, but sometime in the next few years (or decades), we’ll probably have some testing to determine how resistant a given person will be to ebola. Now that ebola has moved _outside_ that area, and into an area with more mobile, urban people, it is more lethal than it typically is where it originated (altho it is not completely clear why that is). Here’s another thing that anyone who read Laurie Garrett knows: ebola is THE classic example of nosocomial amplification. And that nurse didn’t fucking mention it _at all_. When you have a seemingly unstoppable ebola epidemic, the way you stop it is simple: close the hospitals and tell everyone to stay home. About a month later, it’s over. Why does this work? The people who die, die. The people who live, live. And the people who might have caught it by helping or being around someone who was sick, or by being tested by health workers who are themselves sick … don’t.

What we’re going to find out with this epidemic is whether the hospitals we (the British army, Doctors without Borders, Samaritan’s Purse) build in Africa are better than the ones which amplified ebola in past epidemics. They probably are. And then we’ll find out what the side effects of that are, because if you have hospitals that don’t amplify themselves but you don’t have enough space in them, you may get waiting spaces that amplify. Sort of like the circles of hell.

In the meantime, let’s not lose track of the numbers so far, in our fears about the numbers in the future.


Those numbers are showing about 5000 deaths out of about 9000 total cases. And WE KNOW that the total cases number could well be way bigger proportionately than the total deaths is. That is, we’re probably closer to the right number of deaths from ebola than we are to the total number of cases of ebola.

How do we know this? Well, people have been going around checking for antibodies to ebola and finding them in people who weren’t identified as cases. Quite a few people, actually. And we also know from every other studied epidemic _EVER_ that when the panic subsides, we find out that the identified cases were a tiny fraction of the total cases, but a lot of diseases that seem disastrously lethal turn out to fly under the radar of, Oh I Just Thought I Had a Really Bad Cold in many people. Or, as in Typhoid Mary, no cold at all, just the last person you want preparing food for other people.


“In some outbreaks, up to half of people at high risk of infection who were watched never got sick, but they had antibodies against ebola in their blood.”

And this is true of stuff we “know” is 100% lethal. “One study showed that people in Peru appeared to have survived rabies … One in 10 people tested had antibodies to rabies but many couldn’t recall having gotten sick.”

So should we worry about Ebola Mary?

"Dr. Leroy's team studied 25 individuals who never developed symptoms although they lived with family members and cared for them without using gloves and other precautions in two outbreaks in Gabon in 1996.

Using standard virologic techniques, the scientists from Gabon, Germany and France said they could not detect the virus in the blood of the healthy contacts. But Dr. Leroy's succeeded by using a technique known as polymerase chain reaction to grow the tiny amount of virus present."

That’s a _really_ low viral load. Probably don’t need to worry about that. (ETA: Altho we might need to worry about the lab techs PCRing ebola virus -- and then catching it themselves, giving it to their family, etc. This is why we don't even let people study smallpox anymore. People's lab technique is frequently Not Good Enough.)


"Although there is evidence of asymptomatic carriers, the very low levels of virus detected in these individuals suggest they do not pose a significant source of transmission.”

(That article actually has a really nice discussion of R0 if you can sit through it. The R0 analysis does not take into consideration cultural context — but the control intervention analysis at the end does, qualitatively.)

There are things we _really_ should be freaking out about. I nominate global warming and the need to take well understood steps to stimulate our economy (repairing infrastructure and other large projects with a high degree of public support that provide jobs and get money moving around again), along with protecting access to reproductive health care for all women. I’m sure you’ve got a few of your own. As long as it isn’t ebola or IS, or Miley Cyrus twerking, I’m happy to listen to the arguments as to why there is a problem and what we should do about it.

More about those auto analysts

This would be the half-promised post about Adam Jonas, Ravi Shanker, James E Faucette, Paresh Jain, Neel Mehta and Laura Lembke's September 4, 2014 "Industry View" "North America Insight" Morgan Stanley Research paper, "Rent-a-Car Meets Tech: Head-on Collision". As you may recall, I waste time online. A lot of it. Some of that time (an embarrassingly large amount of time) is spent reading Business Insider (I think of it as the soap opera and bon bons part of my day, neither of which I have ever actually consumed).

The note starts out asserting tech "will consolidate the taxi/livery market". As near as I can tell, the not-my-own-car market is divided up into taxis (hail), chauffeur/car service and car rental, with ride sharing services (Uber) and car-sharing services (Zipcar) as relatively new and disruptive entrants. The research team has some good observations to make. Fleet management competency is core to the car rental business. By getting good at fleet management in insurance rentals, Enterprise was able to step into the car rental business at a very high level and compete strongly. Part of this team's argument is that the taxi/livery market presents an opportunity to start up a new, big fleet manager that could then, sort of as an after thought, squat on the car rental market. Their thesis is that the mature oligopoly of car rental is a lot more precarious than it appears to be. I'm not prepared to argue with that. It's what they get into on the way to that that I have problems with.

(1) Will Uber really get into fleet management, or even "closer to the asset"? They have dipped their toe in -- and then pulled that toe right back out when it didn't go brilliantly. They might go a bunch of rounds on this before committing, if they ever commit. A bunch of established players think of Uber as a driver network, and that sounds more correct to me.

(2) Tech is probably going to put the airport counter down. National is an arm of Enterprise and something like National's sign up, state your preferences and walk out to pick a car (or to a reserved car, more likely, at higher priced levels through other names) seems likely. But getting rid of one step out of the airport and one line to wait in may have a lot less of an impact than it seems, unless airports back down on enforcement of contracts, which seems sort of unlikely. I do know that at Schiphol, you can rent a car without going anywhere near a counter OR a lot: a guy meets you, you sign on a tablet, he walks you out to the car which is parked in the pick up/drop off lanes. Drop off works similarly. Super weird, feels sketchy at first, and then you wonder why it isn't always like this. But airports can kill this kind of operation or fee it to death to preserve the existing players, and it is important to not lose track of that.

The next component of the argument involves tech organizing the massive market that is taxis, chauffeur, livery market. It is enormous, and the very biggest participants globally are well under the 1% size. This is where a significant piece of hand waving occurs. If someone says, getting 1% of a market isn't too hard, right? It is easy to think, 1% that's not much, sure, that should be do-able. Maybe not easy, but do-able. Well, if you get 1% of a ginormous thing, then you have a really big thing, big enough to squat on and crush some or all of the car-rental oligopoly. But if the reader/listener/victim actually thinks through the 1%, they won't agree to it and the argument goes *poof*.

Basically, IF Uber bought vehicles and IF Uber kept growing really fast for a while and IF they were not themselves disrupted by stuff like Lyft and IF IF IF, then Uber could put Hertz or whatever out of business. Possible? Sure. Likely? Well, the _last_ time someone produced a related argument, it involved Zipcar, and what happened to Zipcar? They were bought by Avis Budget. This is how oligopolies stick around and this is how businesses solve the innovator's dilemma: they watch the up and comers and buy strategically. Makes for a boring world for an analyst.

But the analysts are not yet done! The next chunk of the paper involves how little time we spend in our cars. Even people spending a couple hours a day in a car are massively underutilizing that capital expenditure if you compare it to a factory that runs a single shift a day. It's true, cars are massively underutilized and they are that way because we all want our cars for approximately the same few hours a day and only some of us can time shift and moving the cars around (removing the sippy cups, kid seats, stuff to drop off on our way to or from an errand) is non-trivial. The analysts make the same observation that every public transport, ride share and car pool advocate has ever made (you can double capacity and half the cost by sharing with just 1 other person!). And anyone who has ever taken public transport or participated in ride shares and car pools knows all the problems that go with that.

I've made the argument before that the easiest way to reduce the number of cars on the road is to reduce the number of cars in a household. Taking the last car out of a household is the hardest one to remove. Making the driving offspring share with a stay-at-home parent is sometimes possible. If public transport is available, the go-to-workers in the household may not need a car during the day. But the household has a hard time giving up the last car, and household size has shrunk. You'd never know it based on this report. (And this factor is often concealed by people who mock those who have range anxiety preventing them from buying a BEV. Inevitably, the mock-er has access to someone else's ICE or hybrid to borrow whenever the trip is long or complex.)

Some of the car-sharing issues would be reduced (but not all! We still have to deal with the sippy cups, special seating, garage door openers, stuff to drop off at the middle class guilt reduction station on the way home, sports equipment for the kids to use at their practice in the evening, etc.) if the shared vehicle could drive itself from where it dropped my husband off at work (he'll need a better place to store his bicycle, which currently lives in the back of the van) to whoever was going to use it to, say, go to Costco and then yoga (that would be Not Me, since I don't do either). And that brings us to the most incredible thesis of all: vehicles that don't need drivers, not at all.

There are a bunch of self-driving solutions out there, all partial so far (available only at certain speeds, in certain locations, etc.). It is easy to imagine a world in which self-driving solutions were less partial. It is less easy to imagine a world in which self-driving solutions are the norm. Much less easy. Predicating an investment decision (the car rental oligopoly is under threat) on self-driving vehicles behaving in this manner seems a little ... ahead of the world. But these analysts are here to put a price on the unimaginable:

"Our team estimates that a fully autonomous car requires about $3K of extra hardware content, plus initially $10K of R&D and profit margin which we expect to fall by one-half by [I really could not possibly make this up] 2020. Assuming an autonomous car could be sold for $8k of total additional cost, this is equal to approximately 6 weeks wages for an Uber driver. The robot car pays for itself in 6 weeks."

The _entirely imaginary_ robot car pays for itself in 6 weeks. It's no wonder the guys over at BI love covering the Morgan Stanley auto team. That paragraph is gonna stick with me probably forever.

Sensing that this might be a teensy bit aggressive (hey, I used to write code for a living. I know from aggressive estimates produced by inadequately supervised junior team members), they hedge. "even if the autonomous car cost an extra $60-70k, it can pay for itself in driver cost elimination in merely 1 year."

So, here is what they missed. There are no legal costs mentioned. And I am pretty sure there will be a fuck ton spent on lawyers trying to craft law to let self-driving cars drive around EVEN WITH A DRIVER RIGHT THERE. There will be a metric fuck ton spent on lawyers trying to craft law to figure out who is responsible for accidents involving self-driving cars driving around with a sleeping driver. And while, like Han Solo, I can imagine quite a lot of money, I cannot actually imagine enough money to pay off everyone to make it legal to have cars driving around without anyone at all in them ... before 2020.

They mostly calm down after this. Mostly. And they spend some time discussing how Enterprise might just decide to really compete, as opposed to play nice as part of the oligopoly, thus destabilizing the industry and presenting us with yet another opportunity (a la Wal-mart and Amazon) to discuss whether we should invoke utility regulation schemes and/or antitrust schemes to return to the more expensive and worse world of the then past.

ETA: FOR THE RECORD, some of my all time favorite sf stories are the Telzey Amberdon stories by James H. Schmitz. Telzey's world involves flying, self driving cars. I have no problem with flying, self driving cars in sf. None whatsoever. Zero. I am good with that. But I don't expect sf when I am reading research notes from Morgan Stanley. Next thing you know, someone is going to be saying we'll all be traveling with anti-grav jetpacks powered by fusion before 2030. And it'll be so safe you won't need helmets. Of course, they'll totally leave out all the other risks that helmets protect us from, like bugs impaling themselves on our faces, which anyone who has ever been on a motorcycle knows all about.

ETAYA: Looks like there's a bit of IHS fiction to blame for the Morgan Stanley group's numbers.


I want whatever they have


"Egil Juliussen, principal analyst for infotainment and autonomous driver assisted systems at IHS Automotive. Juliussen co-authored the study with IHS Automotive senior ADAS analyst Jeremy Carlson", which is/was available for free to members of the news media (I bet I don't count -- if you have a copy of the report, let me know. Title is: Emerging Technologies: Autonomous Cars—Not If, But When).

I love that you can take a couple of guys who muck about with an already highly optional package in cars which is always at risk of disintermediation by smartphones and/or aftermarket choices, and they can say shit like this:

"Accident rates will plunge to near zero for SDCs, although other cars will crash into SDCs, but as the market share of SDCs on the highway grows, overall accident rates will decline steadily,” Juliussen says."

I'm betting he's too young to have encountered all those jokes about what the world would be like if Microsoft were responsible for flying the planes and so forth. Because that quote right there is like a sort of sick little joke.

"The study also notes some potential barriers to SDC deployment and two major technology risks: software reliability and cyber security. The barriers include implementation of a legal framework for self-driving cars and establishment of government rules and regulations."

Wonder what they said? Given how long the Google Books lawsuit took to finally come to the correct conclusion, and especially given that a lot of lawyers were pretty convinced for a year or two there that the wrong conclusion was going to stand, it is no sure thing that self-driving cars with no human present will _ever_ be allowed out on the road. Seriously, someone needs to explain to me how Homeland Security isn't going to put a ten year road block on this thing? All I have to do to commit mayhem is order up a bunch of cars, load 'em up with something appalling, and then tell them to go park somewhere at the same time and wait for the appalling thing to do its appalling business. There's a really good reason why we don't let baggage on an airplane that isn't associated with a person on that airplane.

So, you know, if the guys who wrote this have something that enables them to forget the horrifying world we live in for long enough to produce a report, I kinda want to know what it is. Because that sounds nice.

ETA: You can apparently all laugh at me in less than 10 years when I will apparently be riding around in one of these myself.


There are real definitional problems. Some people mean no person in the car when they say autonomous. Some people mean person in the car.


And the google taxi idea seems to top out at 25 mph, putting it in neighborhood vehicle regulatory terrain, iirc.


Adaptive Cruise Control

Here is JD Power explaining Adaptive Cruise Control:


Basically, cruise control plus some kind of sensor system to notice what is in front of you (camera, radar, lidar, wtf) and some hardware/software to use the telemetry sensibly to maintain following distance. I believe I will be getting this as part of one of the option packages on my i3 (which is supposedly through production).

In the JD Powder 2012 write up, it was about $1500.

When Matthew Debord wrote about his experience with ACC this fall (and this is a guy who writes about the MS auto team's wacky analysis), it was a $1000 option on a Ford Fusion. Definitely has reached mass market.


I'm looking around for other instances where I can get some insight into pricing when these kinds of incremental changes enter the market vs. some years later when they become typical of the market or at least widely available. I don't have _any_ sense of how fast pricing drops on The Fancy Stuff (such as power windows and power steering back in the day, and ACC and back up cameras today). If you know of a source -- or even if you have some rule you use in your own head -- I'm interested.

ETA: A little blast from the past. Pricing for ACC on Ford models in 2009.


$1200 was a steal at the time, versus $2000 or more, but the coverage was focused on getting below $1000.

ETAYA: In the 2014 model year, the Prius Plug-In got some standard features including ACC ripped _out_ to help bring the price down. They were still available as a tech package, with the total coming in less than the 2013 price.


ETA OMG! I'm sending you to the wayback machine. Google found me, I got 404'ed, pulled it out of the cache, but wanted to give you a link that might still work when you read this.


This is awesome! I occasionally (ha!) mention that I used to have a Treo phone and a folding keyboard and moved all my accounts online and yada yada yada back in like 2003. And then everything went to shit and I was entirely back on paper or near enough a few years later. Very annoying. This time around it is sticking, tyvm.

In 2004, the Sierra had a top trim level that included Dynamic Cruise Control. It was an expensive trim line, but the economy was going good and people were excited about the feature. "Personally, I thought it was the start of a new era, where all cruise control would be radar/laser-guided in the years to come." Me and the Treo! In almost the exact same time frame (boy, I could tell you things about what happened to RCA back in the oh never mind).

"it became a chore to manage for the remainder of the trip. ... it is programmed to stay hilariously far away from obstructions. ... it could only use engine braking to slow the van. [loud!] ... it would annoyingly floor the throttle to get back up to speed [also loud!]"

The punchline is a bit lengthy, but at least I didn't have to live it!

"The ultimate insult to injury is that the adaptive cruise control in the Sienna works by using input from a laser sensor in the front bumper’s air intake – you know, the area that is usually damaged when rocks are flung up from the car in front, or you hit a cone or a tall curb. It didn’t take much to kill the laser sensor, and it would stop working even if the lens was dirty. Imagine that, the front of a family van getting dirty! When the system detected dirt or a dead sensor, you would not have cruise control unless you told the van you wanted it to stay in non-guided mode from the moment you started it up, and then never dropped below 40kmh after that. So, the alternative to doing this hokey pokey was to replace the $1000-$2500 sensor, which needed hours of labor to align. Let’s reiterate. Rock hits your front bumper (normal occurence), you can either spend $2500 to fix it, alter your normal driving routine, or drive your $40,000 van, now without cruise control."

This is probably why we all prefer options to cost under $1000. That way it doesn't hurt so much when they break down or annoy us.

Google's Self Driving Car is the Worst Driver Ever


Has never driven in snow OR heavy rain
Cannot avoid pot holes
Would ignore a police officer waving for it to stop
Would fail to recognize a new stop light
Cannot drive in a large, open parking lot or a multi-level parking garage
Cannot read a traffic light if the sun is directly behind it
Objects on the road are avoided regardless of what they are (so a plastic bag would be treated like a rock)

Google's self driving car is not really driving on the road. It is driving on a map.

"intricate preparations have been made beforehand, with the car’s exact route, including driveways, extensively mapped. Data from multiple passes by a special sensor vehicle must later be pored over, meter by meter, by both computers and humans. It’s vastly more effort than what’s needed for Google Maps."

Yeah, that whole google taxi thing is probably more than a few years off.