Mark Draughn, Author at Windypundit

Model Justice — Stalled On the Learning Curve

December 13, 2018 by Mark Draughn 2 Comments

The “Model Justice” series of posts (starting here) is chronicling my attempts to build a software model of certain aspects of the criminal justice system. As I explained in the most recent post, my first attempt at building a model ran into a dead end, so now I’m figuring out what to do for my next attempt.

I’ve identified several possible ways the model could be improved, and based on some feedback from lawyers Mark Bennett and Matt Haiduk, I think I want the new version of the model to eventually implement the following features:

Allow defendants to have some degree of risk love instead of risk aversion when making plea bargaining decisions.
Allow risk preferences to vary depending on case quality.
Allow limits to the extremes of case quality. E.g. Always some chance of acquittal/conviction no matter how strong/weak the case.
Allow prosecution and defense to have different views of the quality of the case.

In addition, I think I’d like to include a few other ideas:

Allow mixes of different types of cases. E.g. Different severity of crime.
Incorporate the way absolute sentencing severity — including collateral consequences — affects risk tolerance.
Incorporate the effects of defendants being stuck in jail pending trial.

I also want to switch to a different model of plea bargaining — one informed by economic thinking — that I’ll describe in a future post.

Before I can do any of that, I need to make a big change to my approach. The system modeling tool I’ve been using, Insight Maker, is designed to model entities that transition through a series of states. That seemed appropriate tool for modeling a defendant going through various stages of the criminal justice system: Arrest, pre-trial, trial, jail, release, and so on.

However, nearly all of the things I want to change will have an effect at only one point in the process: Plea bargaining. Even things that do not happen until after the trial — prison, collateral consequences — are only part of the model because of how they affect plea bargaining. So the cumulative effect of all these changes will be to greatly complicate the model’s plea bargaining logic.

Unfortunately, while Insight Maker is a fantastic (and free) tool, it’s not really engineered for that kind of complex programming. For what I’m planning, Insight Maker falls short in several ways:

You can add code to an Insight Maker model to control state transitions, or to execute independently as actions, but there’s no real place to add libraries of code that can be called from other code, which makes it awkward to do anything non-trivial.
Insight Maker code is using its own private language, invented by the author of Insight Maker. It seems quite competent for a small language, but it doesn’t feel like a full-strength programming language.
The editing environment for Insight Maker is primitive. It has an editor window implemented in JavaScript, but you can only open one piece of code at a time, and it doesn’t come close to the convenience features available in a modern language-aware integrated development environment.
When developing software, it’s nice to be able manage multiple versions of the code base so you can try different approaches and easily back out of bad ideas, but there’s no easy way to manage multiple versions of Insight Maker code.
You can download a model into a file, which you can edit in other tools, but you can’t easily modify or use the model in that format.
Insight Maker doesn’t have libraries of open source code that you can leverage in your program to add features or speed development.
Insight Maker graphic output is limited to only certain types of data and certain types of charts.
Insight Maker doesn’t appear to be thriving. The official Insight Maker forum has only had 15 topics posted in the last three months, and the Insight Maker news stream hasn’t been updated in almost a year and a half.

I considered a number of possible alternatives to Insight Maker. I would love to play around with a professional-strength modeling tool like AnyLogic, Vensim, Powersim, or isee systems’ Stella/iThink packages. AnyLogic in particular can be extended with Java, which would allow me to use a whole world of open-source Java code. On the other hand, these tools are heavily focused on modeling complex business processes, and they are priced as if a corporation was buying them. (Some of them have free or low-cost versions, but those often have serious limitations, not all of which are documented.)

Something like MATLAB/Simulink, in addition to possibly being pricey, is also really intended for modeling physical systems, not system dynamics or the sort of agent-based modeling I was doing. Wolfram SystemModeler is similar, which is a shame, because I already have a home user license.

Speaking of Wolfram, I could probably use Mathematica, because you can do anything in Mathematica, at least according to the folks who make Mathematica. I have a home user license for that too, and I’ve already used it on this project to generate some of the graphics for these posts. The problem is, I get frustrated with Mathematica when I try to do anything complicated, because it’s not like a normal programming language.

But that got me thinking…maybe what I really need is not a simulation tool, but an ordinary programming environment in which I can build a simulation. It wouldn’t have the fancy drag-and-drop block diagrams that all those simulation products provide for building models, but I’m not sure I need that for what I’m doing. I think the connections and flows are simple enough that I can model them with ordinary programming structures.

In that case, the obvious choice for me would be C# and the .NET Framework, since that’s what I work with on my day job. I have access to a ton of development tools, and I’m very familiar with useful code patterns, the development toolchain, and the general approach to developing a C# project. On the other hand, C# is geared towards building production-quality systems, but since I’m still designing the model, I would probably benefit more from a development environment geared toward ad hoc queries and prototyping.

I think I need something that works like R Studio, where I can type in simple commands to change things, run code, and see the results, so I can easily try lots variations and ideas. Or maybe I’d be better with a notebook-style interface, like Mathematica. Both of those also provide data visualization, so you can change something, run it, and see a graphical representation of the result. I don’t actually want to use either of those, however, because I find both R and Mathematica’s “Wolfram Language” to be very annoying for writing anything substantial.

And that brings me to Python. Even though Python has been around for over 25 years, I have never really needed it for anything I was doing, so I’ve pretty much ignored it until now. I tend to prefer heavily-typed languages like C++, Java, and C# that compile down into machine code to run efficiently. Python is an interpreted scripting language that is usually used for writing short programs that stitch together powerful components. (Also, it uses indentation to delimit blocks, which is different from practically everything else in the world these days.)

However, unlike scripting languages such as Perl (or special-purpose tools such as R), Python was actually created from the beginning with full-strength programming features, and it has grown to become one of the most popular computer languages in the world, especially in the data science community. (So if nothing else, learning Python could turn out useful later.)

For my purposes, Python ticks several important boxes:

Python is a full-strength software system with support for multiple-file software projects and organizational abstractions such as classes, modules, and packages.
Python is a mature, well-known, and widely used language.
Python is supported by numerous integrated development environments, including Eclipse (with the PyDev extension), Sublime, Visual Studio, Spyder, and PyCharm (for which I already have a license).
Python projects are just files in directories, so you can easily edit them in a variety of ways and version control them with git.
There are tons of usable third-party open source Python libraries at PyPi.
Python is thriving and growing as a programming language due to its popularity among the data science community, so there are plenty of resources.
There are development environments for Python that support both an R Studio-style environment (Spyder) and a notebook-style environment (Jupyter).

Of course, to use Python, I’m first going to have to learn Python. Writing tiny programs is easy and quick, but I want to build up a library of modeling components, so I need to learn about things like Python classes, Python modules, and Python packages, as well as figure out a development environment, all before I can start building the model.

I think it’s going to be a while.

How To Save Yourself From the New WordPress Editor

December 8, 2018 by Mark Draughn 2 Comments

Oh my God, I ran automated updates on the blog Friday afternoon without realizing that it was also going to update WordPress to version 5.0. So when I started to write a post that evening, I was met with this monstrosity:

WordPress has been threatening us with their new Gutenberg editor for months now, and I guess it’s finally arrived, despite the fact that the trial Gutenberg plugin has been getting abysmal reviews.

What the hell were they thinking? Gutenberg is trying to be some kind of low-rent page builder, allowing users to drag and drop feature blocks into a page. People who use WordPress to build commerce sites love that sort of thing, because it gives them complete control over the design of every page on an individual basis. I use WordPress for blogging, however, so I hated Gutenberg on sight.

Look, I just want to write stuff. I don’t want or need to make every single page look unique and different. Blogging for me is about content, not visual design.

When WordPress started talking about Gutenberg, I ignored it, because it figured it would just be another editor. I didn’t expect such a drastic change. I didn’t expect that I would have to stop and learn a new way of posting stuff. In essence, WordPress Gutenberg has broken the fundamental workflow of blogging: Open the page and start writing.

In the long run, It’s not going to be a huge deal for me. I deal with technical blogging stuff all the time, and I’ve worked with page builders before, so I’m sure I’ll get used to it. But I suspect some of the more curmudgeony bloggers out there (I’m looking at you, Greenfield!) are going to hate this change.

Fortunately, there’s a solution: Go back to the old editor with the Classic Editor plugin:

Note that it has already been installed by over 700,000 users. I’m one of them. I threw this post together in a few minutes in the old editor.

Update: If you only want to disable Gutenberg for some of the users on your blog, there’s the Disable Gutenberg plugin, which gives you detailed control over when and for whom Gutenberg is active.

#ThotAudit Is Really About Ethics in Tax Collection

November 27, 2018 by Mark Draughn 1 Comment

Oh dear God. I saw a tweet from Mistress Matisse, wondered what the heck she was talking about, and made the mistake of clicking on the tag, and now I’ve discovered a whole new world of idiocy in something called #ThotAudit.

First of all, I had to learn a new word: “thot.” Apparently, it’s rapper means something like “slut” or “whore,” but it’s a relatively new word that showed up around 2011. Because, you know, American vernacular English didn’t yet have enough derogatory terms for sexually active women.

(When I was in high school, I literally didn’t understand why guys used the word “slut” to demean girls. “Wait a minute…you’re insulting girls…for wanting to have sex with us…? Why would you do that?”)

Anyway, #ThotAudit is some kind of weird campaign to report sex workers to the IRS for tax evasion, and it’s organized by exactly the kind of people you’d think would organize such a thing:

Online sex workers have slammed a viral campaign by right-wing “incels” and men’s rights activists to mass report them to tax authorities for allegedly failing to declare income.

The hashtag #ThotAudit, which appears to have been started by a Facebook user calling themselves “David Wu”, began to trend on Sunday after being promoted by controversial pick-up artist and “legal rape” advocate Daryush ‘Roosh’ Valizadeh.

“Online thots are finding out that all income generated from their breasts and vagina is taxable,” Mr Valizadeh tweeted.

“Men are aggressively organising to report all thots. I don’t blame them: these girls are getting a free ride via beta bux and a broken sexual marketplace that is rigged in the favour of females.”

Man, that whole “income generated from their breasts and vagina” line is creepy as fuck. Can you imagine having to spend every day of your life being that guy? Who worries about stuff like that? I guess the same kind of guy who whines about a “broken sexual marketplace.” #ThotAudit is so aggressively stupid that I don’t know where to start.

I guess I might as well start by anticipating that someone will accuse me of “white knighting.” That’s a term that pickup artists, MRAs, and other assholes use when some other guy steps in to defend a woman they’re bothering. It’s meant to taunt the other guy for using the appearance of chivalry in a weak attempt to get laid, because they can’t imagine any other reason for talking to — or about — women.

The thing is, I’m not white knighting, because I’m not defending these women. They don’t need my help defending against something this dumb.

First of all, it’s not like any of these #ThotAudit assholes have actual knowledge of tax evasion. They don’t have any information that they could give in a statement in a legal setting that would help establish any of the elements of tax evasion.

All the #ThotAudit assholes have is a suspicion that these women must be evading taxes because they are sex workers. And that’s on the basis of Twitter feeds and escort web sites, which IRS agents can easily find by themselves without help from the #ThotAudit twits. Heck, the #ThotAudit twits don’t even have the real names of the women they claim they are reporting to the IRS.

But let’s suppose, just for the sake of argument, that some #ThotAudit twit is actually able to get IRS investigators to take interest in a sex worker that he brought to their attention. If the investigators are able to bring the legal powers of the federal government to bear on the problem (I have no idea what level of proof is needed for this) they could probably trace a Twitter account or website to an actual person. And if they did that, what would they find?

Probably, a taxpayer.

In the case of purely online sex workers, like cam girls or phone sex performers, they have to be able to accept electronic payments, which will create records in the banking system that will make the income visible to the IRS, pretty much guarantying they will have pay taxes on the income. Things get a little more complicated with sex workers who accept cash payments in person, such as strippers or escorts. If the #ThotAudit jerks get IRS investigators to look into the finances of someone like that, what will they find?

Probably something familiar: A cash-oriented business.

The IRS is of course deeply suspicious of cash-oriented businesses because it’s hard for the IRS to know how much money they really receive when there are no electronic bank records of the payments. Everyone suspects that a lot of cash-based businesses cheat on their taxes, so the IRS would be right to be skeptical about the tax returns of sex workers. But, well…the point of unreported cash income that no one knows it exists. The IRS isn’t going to be able to do much without proof of income.

And it’s not as if most sex workers don’t pay income taxes. Sure, if a customer pays them $300 cash, and they go to Target and use the money to buy groceries or clothing or a television, they won’t leave a paper trail, so they can cheat by not paying those taxes. But if they lead anything like a normal middle-class life, they probably have a ton of expenses that leave a trail. If they have a $2000/month mortgage, a $600/month car payment, and $300/month in utilities, plus $500/month for health insurance and another $1000 per month in miscellaneous credit card payments, they need to be able to show almost $6000/month in income to cover those expenses.

That means reporting somewhere around $18,000 in income every quarter and making estimated tax payments. It will be more if they want to put some money in the bank or pay into an IRA. So if the IRS takes a look at them because of #ThotAudit, they’ll find a woman reporting over $70,000 in annual income. They may suspect that she’s bringing in even more as unreported cash income, but it will be hard to turn those suspicions into evidence.

Given that the #ThotAudit douchebags aren’t reporting people in other cash-oriented business — bartenders, food truck operators, bodega owners, dry cleaners, farmstand owners, landscapers, housekeepers, dance teachers, handymen…or for that matter, male sex workers — it’s pretty clear this is just another excuse to harass women.

But it’s not even effective as harassment. I had to think for a while to convince myself that the threat to tattle to the IRS was toothless. But to actual sex workers, who know far more about the tax implications of their business than I do, this has got to be incredibly obvious. Sex workers on Twitter are certainly not frightened. They’ve been mocking #ThotAudit without mercy.

Look, sex workers deal with more than their fair share of assholes online. They are used to fending off time wasters and religious nuts and every would-be edgelord who thinks it’s epic to ask them if they suck dick for a living. They are used to harassment, and they have strategies for dealing with it. They are not impressed by someone threatening to mail a piece of paper to the government.

Model Justice – Plea Bargaining Blues

November 13, 2018 by Mark Draughn Leave a Comment

As I explained in the first post of this series, I’m trying to build a computer model of certain aspects of the criminal justice system. This week, it’s not going well.

My goal is to build a model of how our justice system handles plea bargains, with an eye toward understanding (1) why a system built around the highly formalized decision-making mechanism of the trial doesn’t actually have many trials, and (2) what we can do about it. As I am neither a professional systems modeler nor a lawyer, this is mostly intended as a learning experience.

I plan to eventually give the model a bunch of parameters such as time spent waiting for trial, case quality, sentencing regime, defendant risk tolerance. If I do a good job building the model, then I should be able to find a set of parameters that make the model produce results that look like our criminal justice system: Lots of plea bargains, very few trials, etc. Then I can experiment with ideas for reform by tweaking the model parameters to see how the model responds.

I’m taking an iterative approach to building the model, starting with a skeleton that has some of the major moving parts of the criminal legal process, and then elaborating on parts of the model to flesh it out into something more realistic.

The most recent model is very simple: There’s only one crime, people are arrested at a fixed rate, they wait in jail for a fixed time, and then they have a trial, after which they may go to jail for a fixed period. I suspect my lawyer readers will be annoyed to learn that the entire trial — the core of their profession — is modeled as a single draw from a random distribution to get the verdict. Plea bargaining, my whole reason for doing this, isn’t even in the model yet.

It’s that last problem that I tried to fix. I’m trying to add a plea bargaining phase somewhere between arrest and trial, and I’d like to tell you about it. This story will not end well.

I started by thinking about the basic decision the defendant has to make during plea bargaining: Take the prosecution’s offer or reject it and go to trial. I think there are two main question to address: What does the defendant know? And what does the defendant prefer?

I’m assuming the defendant has knowledge of a few basic things:

The amount of time in the plea offer from the prosecution.
The possible outcomes of the trial, consisting of:
- The probability of being found guilty.
- The sentence they will serve if found guilty.

Here you see more of my assumptions: Everybody has the same information, and it’s correct information as far as it goes. Note that I’m not saying anyone knows how the trial will turn out ahead of time — I’m modeling it as a draw from a random variable — but I am assuming that both the prosecutor and defense know what that probability is, and that both can use it to make decisions.

When thinking about a random event, it’s helpful to consider the expected outcome, which is calculated as the sum of the value of each outcome times the probability of that outcome. If you buy a raffle ticket which has a 1-in-10 chance of winning $50, we can calculate the expected value of your ticket as follows: Your 1-in-10 chance is a way of saying p = 0.1 for winning, so we multiply that by the payoff, $50, and get $5. Your 9-in-10 chance of losing has p = 0.9, but you get nothing for that, so the value of losing goes to $0. Add up the expected value for all possible outcomes (5 + 0), and you get the expected value of your ticket: $5.

Similarly, if the defendant knows he will be found guilty half the time (p = 0.5), and he knows the sentence is 10 years, he can calculate the expected sentence as (0.5 x 10 =) 5 years. Once he receives the prosecutor’s plea offer, he knows everything he needs to know (under this model) to make a decision.

The key issue controlling the defendant’s decision is then his attitude toward the risk of going to trial. If the prosecutor’s plea offer is 5 years in jail — exactly equal to the expected sentence calculated above, then you could argue that the defendant should be indifferent to the choice between taking the offer or going to trial, since the expected sentence is 5 years either way.

The problem with that approach is that it ignores risk, and people don’t usually ignore risk. All other things being equal, people usually prefer to reduce their risk. There is a massive amount of evidence for this, including the existence of the entire insurance industry, where customers pay fixed premiums to avoid the risk of having to suffer losses. If this risk aversion extends to plea bargaining decisions, a defendant will choose a certain sentence of 5 years in jail over a 50% chance of spending 10 years in jail, even thought the latter also includes a 50% chance of going free.

But if the defendant prefers a 5-year plea over a trial with a 50% risk of a 10-year sentence, what happens if the prosecutor makes an offer of 5 years and 1 month instead? Well, the insurance industry is pretty profitable, which means that people are paying more in premiums than their insurance companies are paying out to cover losses, which means that (on average) customers lose money when they buy insurance. So I think it’s very likely that a defendant will agree to lose an extra month of life in prison to protect against the risk of losing an extra five years of life in prison. If criminal defendants are risk averse, they would be willing to accept a plea bargain for more time than the mathematically expected sentence.

I am modeling this risk aversion by calculating an adjustment to the probability of guilt that represented how that probability would feel to a risk-averse defendant. E.g. a 50% (p = 0.5) probability of guilt with a 10-year sentence has a mathematically expected sentence of 5 years, but to a risk averse defendant, even a 6-year sentence might be preferable to a 50% risk of a 10 year sentence.

I decided to calculated the adjusted risk-averse factor by calculating an expected probability of an acquittal as 1 – p, then raising it to a power specified as a model parameter, and then converting back to a probability of guilt by subtracting it from 1 again. E.g.

1 – (1 – p)^a

where

p is probability of guilt

a is the risk aversion model parameter

I picked this method of modeling risk aversion because (1) I vaguely remember reading an article about risk aversion that modeled it like this, (2) it has the nice property that p = 0 and p = 1 both map to themselves, so it stays in range, and (3) I like the look of the curve. Here, for example, are the curves with a set to 1, 1.5, and 2 for comparison. Remember a = 1 means no risk aversion.

Now that I had a simple model for defendant choice, it was time to model the prosecution.

One of the fundamental differences between defendants and prosecutors, at least for purposes of my model, is that defendants are individuals who seek their own advantage in a single criminal case, whereas prosecutors work together across hundreds or thousands of trials to achieve a common organizational goal. So while I feel reasonably comfortable in assuming for purposes of the model that most criminal defendants will try to minimize their risk-adjusted prison sentence, I’m not nearly as confident in figuring out what exactly prosecutors are trying to achieve.

To keep the model simple, I’m going to assume that prosecutors want the opposite of what defendants want. That is, if defendants want to minimize the risk-adjusted sentence they expect, prosecutors want to maximize the aggregate expected prison sentence across all defendants.

You’ll notice I didn’t say the prosecution goal was risk-adjusted. That’s because if defendants are like insurance buyers, prosecutors are like insurance companies. For an individual home owner, the loss of their house to fire would be financially devastating without an insurance policy. But to the company that issued the policy, it’s just business. They issued policies for tens or hundreds of thousands of homes, and they are well aware that a bunch of them are going to burn down every year. Meanwhile, they will be collecting premiums on all of them, and very nearly all of them will survive through the year with no loss whatsoever. In investment terms, insurance companies are “diversifying away” the risk.

Prosecutors can do something similar. Using the example from above, a defendant going to trial against a 50% chance of a 10-year sentence has a mathematically expected sentence of 5 years, but he’s not going to serve 5 years. He’s going to serve either zero or 10 years. On the other hand, a prosecution team that deals with 100 of these trials will have a mathematically expected sentence of 5 years per case, which adds up to a aggregate expected sentence across all cases of 500 prisoner-years. And they’re actually going to get about 500 prisoner-years, which is 50% of the theoretical maximum of 1000 prisoner-years. In other words, in the long run, the prosecutor can earn the expected sentence on average.

Furthermore, defendants can only accept or reject the deal that the prosecutor offers them, but the prosecutor’s office can offer whatever deal they want, including dismissal or no deal, to all of the defendants. This means it’s ultimately the prosecutor who gets to decide which defendants get plea bargains.

However, they have to do that within a resource constraint: There’s a limit on the number of trials the justice system can support in any given time period (at least for the short term). So if 100 cases enter the system, there might only be enough capacity in the system for 30 full trials. For the system to keep functioning, the remaining 70 cases will have to be disposed of with a plea bargain.

But how do they choose which defendants get plea deals and which ones go to trial? Well, consider this batch of defendants:

ID	Probability of Guilt	Sentence if Convicted	Mathematically Expected Sentence	Risk Aversion Adjusted Probability of Guilt	Acceptable Plea	Difference Between Acceptable Plea and Expected Sentence
A	0%	10	0.000	0.000	0.000	0.000
B	10%	10	1.000	0.178	1.778	0.778
C	20%	10	2.000	0.299	2.991	0.991
D	30%	10	3.000	0.405	4.054	1.054
E	40%	10	4.000	0.503	5.030	1.030
F	50%	10	5.000	0.595	5.946	0.946
G	60%	10	6.000	0.682	6.817	0.817
H	70%	10	7.000	0.765	7.653	0.653
I	80%	10	8.000	0.846	8.459	0.459
J	90%	10	9.000	0.924	9.240	0.240
K	100%	10	10.000	1.000	10.000	0.000

The first three colums show the probability of being found guilty, the fixed 10-year sentence, and the resulting mathematically expected sentence. The fourth column shows the probability of guilt adjusted for risk aversion (using an exponent of 3/4), and then the plea that the defendant would be willing to accept based on that risk adjustment. Finally, the last column shows the difference between the plea and the mathematically expected sentence. I.e. the premium in prison time that the defendant is willing to pay to reduce the risk of a 10-year sentence.

At this point, the prosecution plea bargaining strategy seemed obvious to me: If the goal is to maximize aggregate prison time, then the prosecution can maximize total prison time by offering plea deals to those defendants willing to accept the most additional prison time beyond beyond the mathematically expected sentence. Thus, if they only have resources for 3 trials for the defendants above, they should dismiss the unwinnable case A, plea out B through H, and go to trial on defendants I, J, and K.

That’s when I realized I had screwed up: If the prosecution is trying to maximize aggregate prison time, and if defendants will agree to more than the mathematically expected sentence because they are risk averse, then why would the prosecution ever go to trial on any case? By the assumptions I had made, they could always get more prison time with a plea offer. And yet…trials happen.

There’s a mistake in here somewhere. Possibly more than one. Either I’ve oversimplified my model, or else it’s flat-out wrong. Or both. A few possibilities spring immediately to mind:

Maybe I was wrong to assume both sides see the same probability of guilt. The sides in each case could have randomly different views of the quality of any given case. For all I know, all trials occur because the prosecution thinks it will be easier to prove guilt than the defense does.
Defendants may be risk-preferring instead of risk averse. Conventional wisdom is that criminals are more willing to tolerate risk than law abiding people. Or risk-preferring people, criminal or not, may be more likely to attract police attention.
Prosecutors may be optimizing for something other than aggregate prison sentence, such as conviction rate. Or prison times may combine in a non-linear manner: E.g. five 20-year sentences may be better for a prosecutor’s career than twenty 5-year sentences.
Maybe collateral (non-prison) effects of a conviction affect defendants’ decisions in ways that matter. E.g. If any conviction, including by plea bargain, could result in deportation, defendants might grasp at even small chances at trial.
I may need a more complex model of the trial outcome than just a probability of conviction. Maybe there are important phases or types of cases. Or maybe one side has more information and can consistently make better choices.
Maybe I need to account for interactions between cases in the prosecutors’ decision making. E.g. taking 5 aggravated battery cases to trial may use up resources that could be used on a murder case, which could affect how they make plea bargains.

Time to go back to the drawing board. At the very least, I’ve got some thinking to do. And maybe it’s time for me to actually read something written by people who study this stuff.

A Carnival of Street Photography

November 12, 2018 by Mark Draughn Leave a Comment

Singer-songwriter Natalie Merchant released her first solo album, Tigerlilly, in 1995. I remember its top-ten single “Carnival” was all over the radio for months. And of course there was a music video to go with it.

I was disappointed with the video. I liked Merchant’s songs, but the video felt like she wasn’t really trying: It’s shot in black-and-white, and it just shows Merchant lip-syncing the song while walking the streets New York City. I thought it was the sort of thing an artist would throw together when the record company wants a video and the artist doesn’t really care.

Twenty-four years later I just caught the “Carnival” video by accident when YouTube automatically played it after something else. And I discovered it was wonderful. The video hadn’t changed a bit, of course. But I have.

About twelve years ago I got seriously interested in photography, and in the process of learning how to take pictures, I read about famous photographers. I was especially interested in the street photographers, such as Henri Cartier-Bresson, Garry Winogrand, Walker Evans, and Lee Friedlander. Familiarity with their work gave me a better artistic context in which to view the “Carnival” video.

What first caught my eye was that Merchant was taking pictures of people and things on the streets:

And the video would sometimes show the still photos that could have resulted:

That…looked a lot like she was doing street photography. And then just seconds after that, the video showed this image of a man standing the the deep shadows, with just a bit of light hitting his face:

That’s a classic street photography technique.

I had to restart the video and watch it a lot more closely. Not only is Natalie Merchant doing street photography in the video, but many of the scenes in the video itself are shot in the distinctive style of street photography. It’s four and a half minutes of artistically harsh shadows, strong angular lines, interesting people, strange reflections, unusual poses, and careful juxtapositions.

Heck, she’s even using a Leica camera, preferred by many street photographers:

It was clear that someone on this video knew their way around street photography. It turns out the video was directed by Melodie McDaniel, who is a professional photographer when she’s not making music videos and commercials. The photos on her website (especially in the archival section) show some definite street influence.

If you want to see more of that lovely street composition, here’s the full “Carnival” video:

Main Menu