A few days ago, Gideon at a public defender posted a very thoughful response to my post discussing whether criminal defense lawyers help their clients get away with crimes:
We fight. We fight tooth and nail. Not because we absolutely believe that our client is innocent—in fact, most of the time we don’t care whether the client is guilty or not.
We use phrases like “make the State prove its case” and “the burden of proof is on them and if we don’t make them meet it, the slope will start slipping”.
That’s only a piece of the beginning of Gideon’s response, but I’m going to take it and run in a different direction. To start things off, I want to talk about how quality is measured.
One of the aims of statistical quality control is to observe and then eliminate variance from an industrial process in order to improve the quality of the output. If some chemical mixture is supposed to cook at 190 degrees, it often doesn’t matter much in terms of quality control whether the temperature is 185 or 195, as long as it’s the same temperature every time.To tell if that’s the case, you have to take temperature measurements every time you reach that step in the process.
But taking the temperature is itself a process that is subject to variance: Do you take the measurement in the middle of the chemical vat, or near the edge? Do you measure the temperature at the surface of the mixture, at the bottom, or somewhere in the middle? Does it matter if the probe is inserted at an angle? Such variations in the measurement process could cause you to draw incorrect conclusions.
To reduce the effects of measurement variance, quality control experts like to use operational definitions of the quantities to be measured. So instead of saying “measure the temperature of the chemical mixture” the quality control manuals will spell out the procedure in detail:
Use Binford X2102 digital temperature meter fitted with a Type B 25-inch probe. Turn the meter on and allow the electronics to stabilize for 30 minutes. Then open the measurement port at the center of the top of the chemical vat and insert the probe 18 inches into the vat. Wait one minute. Record the temperature indicated on the probe display.
Having such operational definitions for every measurement will improve the quality control statistician’s ability to judge the quality of the process.
Quality judgments are not always as objective as a simple physical quantity. Sometimes they involve an inherently subjective evaluation.
When a web search company is considering a change to their search algorithm, they need to make sure it’s going to be an improvement. The new algorithm should produce search results that are more relevant than the old one.
Unfortunately, there’s no objective way to measure relevance. It’s an inherently subjective concept: Search results are only relevant to the people who use them. To test a search algorithm, you have to have real people evaluate the results.
A further complication is that people are different. Some people who search for “Catholic girls school” want to find a place to send their daughters, and some people want to find pornographic fantasies about Catholic school girls. As with any poll, the solution is to use a randomly selected group of people and average away the differences.
In essence, this testing process—evaluation by a pool of random strangers—is the search engine company’s operational definition of how search quality is measured.
I’m probably stretching the analogy too far, but I tend to think of the criminal trial process as our society’s operational definition of guilty.
The use of a jury, especially, makes the whole process sound a lot like a scientific measure of a subjective quality:
In order to prove someone guilty, a prosecutor has to gather all his evidence and testimony and present it in a formalized manner to a disinterested and isolated group of twelve strangers in such a way that they become unanimously convinced beyond a reasonable doubt that his claim of guilt is true despite the efforts of a dedicated opponent assigned the job of challenging his claim.
(That’s a summary, of course, a virtual platonic ideal. The real operational definition of guilt consists of the entirety of criminal procedure, and it’s messy and ragged and forever changing.)
One of the things that’s kind of surprising about this process of finding guilt is how distrustful it is of the people in the justice system. Not only is the entire process of deciding guilt handled by a separate branch of government, but even the people working in that branch are not trusted to decide guilt, which is left to the jury.
Even then, the prosecutor is not allowed unobstructed access to the jury. His every move is opposed by a person assigned the task of doing exactly that: The defense attorney.
That’s why when I say that criminal defense lawyers are there to help their clients “get away with” their crimes, I really don’t mean it in an offensive way. They play an important part in a high-stakes decision making process. It would be far too easy to convict people of crimes if there wasn’t someone assigned the specific duty of zealously representing the defendant’s interests.
I’ve always thought that Ken Lammers put it well:
Basically, whether you are trying to get a good deal or a verdict of not guilty (as few and far between as those are), as a defense attorney you almost invariably represent a client’s liberty interest against society’s long term interest in making him conform and the government’s use of power to either force conformity or seek vengeance.
I suspect that most criminal defense lawyers do not draw inspiration from the thought of playing a role in a system of checks and balances, but they don’t have to. As long as they are inspired, they will serve their purpose.
Leave a Reply