In response to my earlier posts on VW cheating I have heard several people say “I don’t know engines well so I don’t follow most of what you’re saying”. This is a familiar hurdle, true for most specialized technical fields.
I don’t mind hearing this because I am a believer in bridging. I see no point in shaming people who lack hands-on engine experience or have not thought deeply about the economics of transportation. A technical argument should be able to stand on its own, such that it can be explained to anyone.
So here I will attempt to build a bridge from being a long-time engine tuner to the growing number of very smart IT and infosec people without any real engine experience who suddenly now are looking into smog topics.
More specifically I will answer from experience whether removing engine DMCA immediately would help in the case of VW cheating.
Three Levels of Analytics
On the beginning end of an analytic spectrum, the thought that immediate DMCA removal “probably would help” is a binary form of assessment: see something say something. DMCA is a prior known harm. It has done harm elsewhere. When DMCA is noticed therefore its removal is a simple reaction.
Next on the spectrum is knowing that DMCA can be a harm yet wondering based on ranked data if removal will achieve an objective. Seeing DMCA used by a German car company could mean every German car company is suspect. A ranking system begs the question of how to know when and if safe transition away from DMCA is possible? Is it after German cars no longer are available for sale?
The training examples I suggest to answer this question are from other scandals related to privacy. Lance Armstrong, like VW, was a winner caught cheating. However Lance wasn’t the problem, he was a symptom of demand. He represented a far wider problem.
Using first level analytics (see Lance with privacy and say something) would not be the right approach. Likewise second level analytics are insufficient because Lance was not the only cheater.
Getting beyond level two analytics is very hard. Anyone with audit experience knows it can be a losing battle on the ground unless you have real infrastructure in place to support a search for knowledge. You have to be able to store data, evaluate and adapt. The better your tests the more your cheating adversaries will circumvent them so you need some way to win that race.
A sophisticated level of knowledge is a third level of analytics, which I will call heatmap. As signs of cheating emerge, none very special on their own, the probability is warmer overall. Privacy is not completely lost, but reliable indicators of cheating are developed broadly. This involves sensors so fast, unique and rich in detail that the cheater can not afford to keep ahead of them.
There are two more levels of analytics above heatmap unnecessary to discuss here. Suffice it to say a third level gets us to where we need; it should answer whether and when removing DMCA would be improving air quality.
I use my own experience to work through finding a third level analytics answer. It comes from tuning many engines and even making my own fuel over the last decade. Here are two reasons why I think removing DMCA is a distraction from the main issue: free market risks and the economics of performance tuning.
Free Market Risks
Removing DMCA would be great for innovation and cost improvements from shared knowledge. It would create a more free and unregulated market. That however is not going to magically make pollution stop.
More of something and cheaper doesn’t imply clean. In fact it could be the opposite as the market innovates toward more power for less money. Removing DMCA arguably means the market continues in the worst possible direction and pollution simply increases.
Can we avoid innovation going awry? Yes, with regulation specific to the objective. DMCA is a weak control for issues of competitiveness and innovation, only slightly related to the issue of keeping air clean. Removing it should come when we are able to regulate for clean air.
Removing Lance Armstrong’s privacy could actually make his cheats more pervasive and harder to detect by auditors. So could we improve detection without removing privacy completely? Absolutely yes.
Some suggest the VW cheat was caught using sophisticated testing. I think that’s an exaggeration but we still should look at the tests as an example to model. The auditor success really was in perseverance and perspective more than doing anything clever or novel. Someone kept thinking mpg and power advertised were too good to be clean, so they applied a clean-specific test where VW did not.
Take a moment to think of the VW cheat this way:
- When you are stationary (garage, warm up in snow) you get cleaner air
- When you drive, you get more power but it is dirty
This is exactly, and I mean exactly, what typical American customers demand of manufacturers. It is considered acceptable to pollute in the areas least likely to be measured. This is why you can buy “off-road use only” performance parts (meant to be used privately) and then drive them around on roads (publicly) without any real risk of prosecution or fines.
So with pervasive cheating and cheating ingrained in the American engine market why did regulators focus on one company? In brief because it is harder to ban pollution by cheating American consumers than it is to go after a wealthy German company with a minority of vehicles on the road.
To put this in perspective VW already had their cars banned from the California market in 2004. They came back in 2008 with some incredible new numbers and sales took off. All of this has been blogged here extensively before.
A good auditor sees improvement and immediately starts thinking skeptically; how did a small car sales winner get so good so fast (the answer is Bosch, who actually developed “off-road use only” codes). And then the auditor hunts. Sending a car across the country with sensors is not a super special or novel idea, which perhaps you have read in my prior blog posts (e.g. Jaguar boasted 62mpg in cross country test).
Auditors today are closing in on manufacturers because the market functions in a somewhat predictable manner. Changing this abruptly by opening up innovation could lead to many more polluters, groundswell of people acting more like VW (because you’ve removed VW from the equation) and even take us towards weakening of other controls focused on clean air. A focus on a winner with a clear-cut case is a very efficient form of regulation but insufficient, since the problem is widespread.
All of this says to me removing DMCA and opening up a free market without other forms of regulation in place would likely be a clean air setback. It would be like demanding the recipe for cyclist performance enhancing drugs be public in order to reduce their use. Unless cyclists and race organizers are prepared to regulate against use, releasing the recipe can lead to far more cheating and less chance of stopping it.
Performance Tuner Economics
It is well known in the engine market that DMCA does not stop people from completely reverse engineering their cars. Performance tuning firms, not to mention customers themselves, often reverse firmware and/or write their own. In fact you could say there is a symbiotic relationship where the weak enforcement of DMCA allows manufacturers to learn from the after-market crowd what power enhancements to sell next.
Note here there is literally no market for clean enhancements. You simply can not find after-market products designed to get the cleanest possible emissions from your engine.
What VW did was realize that customers wanted more power, more mpg, as they always do. This translates to more convenient “workarounds” and double-speak to avoid regulations of being clean. Thus instead of customers paying $100 and taking 10 minutes to after-market tune their engine, VW essentially modeled customer behavior and provided a solution in software.
VW probably figured why leave the fixes to after-market performance companies. They also likely saw it as a temporary workaround to get back into the market sooner (2008) instead of when they had figured out how to actually comply: both power and clean (2013). Classic product manager risk behavior.
The pervasive cheating that drives VW to do the same is both good and bad. On the one hand it is bad because the market obviously and flagrantly pollutes and no one has budget or tools to stop it at the widespread consumer level. On the other hand it is good because VW took the unrepentant customer bait for better cheats, brought it in-house, and gave regulators a one-stop shop to issue a fine and make an example for everyone to see.
Using our Lance Armstrong example, he cheated more and better than all the other cheaters, which made him the best person to take-down in front of everyone as an example. Some people say VW had 11 million cars affected and this is a lot. Unfortunately this is not a lot in the big picture of cheating.
I mentioned before that California took action in 2005 and knocked VW out of the market. This was because VW was big enough to be a centralized high-profile target but small enough and consumer-centric enough to be made into an easy example. Much more difficult would be for regulators to go after Ford, GM, Kenworth, Caterpillar, John Deere, etc..
Instead of only affecting a few million consumers a regulation at the much larger cheater level could seriously impact business processes and even shut them down. It is common to hear truck drivers complain that if they have to drive a clean engine in order to operate in California they will go out of business; lower mpg or less power to stop polluting is a very hard business decision for hundreds of millions of drivers.
DMCA therefore doesn’t really stop people from innovating (albeit in non-clean direction). So it would have to be enforced far more strictly to help keep air clean. That would be a very bad thing. Harming innovation to reduce pollution sounds backwards because it is. The same resources instead of trying to enforce DMCA could be used directly for enforcing actual clean air controls. The goal being when you finally remove DMCA the resulting innovation would be pointed in a positive direction.
This is why I say stop wasting time talking about DMCA in pollution circles (a mostly non-barrier to reversing and tuning) when you directly could be addressing the actual problems of cheating for actual air quality controls.
Building a Better Solution
In conclusion, I hope I’ve built the argument well enough to stand on its own, no special engine experience necessary. We need to be building a far better surveillance network to monitor for clean air and a far more effective response system for enforcement. This probably sounds shocking so the ethics and norms of behavior have to be ironed out. We should put it in terms of other pollution success stories.
When you see someone smoking a cigarette you say something to them. If that person doesn’t comply you invoke authority. Obviously you can’t tell on the spot you are getting cancer but you have it on good authority that seeing a smoker is reason to act. DMCA of the cigarette industry, such as recipes for mixing and rolling, seem mostly irrelevant because they are.
Thus we really should ask ourselves for engines how do we build a comfortable living environment still capable of finding and stopping engine-smokers?
Imagine every loud pipe you hear is reason enough to say something. Generally loud pipes are after market power improvements that intentionally increase pollution. The ear is no perfect sensor but it’s a start (albeit California regulators have been arguing they can decouple noise from pollution). Imagine neighborhoods using air quality sensors deployed to help build a heatmap; for example monitoring outside popular restaurants collecting data on SUV emissions left behind. You then deliver to the restaurant their pollution results and fine them based on their customer behavior.
There are many possibilities of great impact to consider and plan. Arguments about removing DMCA are mostly irrelevant to clean air economics and technical problems.