"If you want to change big things, you pay attention to small things."
-Rudy Giuliani on C-span talking about the Broken Windows theory
The Broken Windows theory was the catalyst for solving NYC’s crime wave in the 80’s and 90’s. NYC’s administration had been focusing on major crimes, like murder, and overlooking smaller crimes along the way. But it wasn’t working. So the city started going after petty crime that it had been overlooking: turnstyle jumpers, squeegee men, public drunks, etc. The result: All crime rates fell suddenly and continued to drop for the next ten years.
Giuliani says, “The idea of it is that you had to pay attention to small things, otherwise they would get out of control and become much worse.”
In a lot of our approach to crime, quality of life, social programs, we were allowing small things to get worse rather than dealing with them at the earliest possible stage…So we started paying attention to the things that were being ignored. Aggressive panhandling, the squeegee operators that would come up to your car and wash the window of your car whether you wanted it or not — and sometimes smashed people’s cars or tires or windows — the street-level drug-dealing; the prostitution; the graffiti, all these things that were deteriorating the city. So we said, “We’re going to pay attention to that,” and it worked. It worked because we not only got a big reduction in that, and an improvement in the quality of life, but massive reductions in homicide, and New York City turned from the crime capital of America to the safest large city in the country for five, six years in a row.
Building momentum
One key component of Broken Windows is that it shows progress. It’s not about miracles or heroic solutions or solving massive problems overnight. It’s about building momentum. It’s showing your audience that you’re headed in the right direction. It’s making visible changes, even slight ones, that show you’re doing something. Someone is on the case. People know that you haven’t abandoned them. You’re giving them a reason to trust you. You’re building faith.
Broken Windows and the web
This concept applies to all kinds of areas, not just fighting crime. In fact, others have already pointed out connections between Broken Windows and web development:
In Software Entropy, the Pragmatic Programmers talk about entropy, broken windows, and computing neighborhoods.
Don’t leave “broken windows” (bad designs, wrong decisions, or poor code) unrepaired. Fix each one as soon as it is discovered. If there is insufficient time to fix it properly, then board it up. Perhaps you can comment out the offending code, or display a “Not Implemented” message, or substitute dummy data instead. Take some action to prevent further damage and to show that you’re on top of the situation.
And Broken Windows Theory and Your Web Site says visiting a site with broken links, misleading navigation, or missing images is like visiting a neighborhood where “you get the feeling that no one cares.”
If you’re trying to create a community around your web site, I think you really have to consider this theory because it prompts you look at the value of the small and seemingly inconsequential problems on your site. By considering the ways that small problems can multiply a feeling of apathy, you may find that your time is better-spent fixing broken windows as opposed to building new houses.
Solving “street-level” problems at your site
It seems counterintuitive to focus on little problems instead of big ones. But cracking down on your version of squeegee men, graffiti, etc. could yield surprising results.
Some related ideas:
- Web apps can show someone’s “on the case” by posting a list of recent changes and what’s coming next (example). [SvN]
- Divide problems into smaller and smaller pieces until you’re able to digest them. [Getting Real]
- Show your product is alive by keeping blog posts coming post-launch. [Getting Real]
Joshua Schairbaum
on 27 Nov 06I’m not disagreeing with your “Broken Windows” methodology for web design, but I have to disagree that it works in reducing crime. I recommend you check out Freakonomics by Stephen Levitt and Stephen Dubner. They propose a completely different reason for the drastic drop in the crime rate in the 90’s, and it had not much to do with Rudy Guiliani. I know this is off topic, but I thought I would point it out.
Alex
on 27 Nov 06Jason, your comment is completely not off-topic!
Matt Davis
on 27 Nov 06Josh is quite right, there are many other theories that are equally plausible as major contributors to crime reduction throughout the 90’s such as the controversial abortion theory put forth by the Freakonomics guys, the reduction in inner city crack use, community policing, and a better economy. While the broken windows theory is probably part of the equation, crime decreased nationwide during this time period- even in communities without broken window policies.
Rob Breidecker
on 27 Nov 06Nice story. Regardless if the “Broken Windows” theory is to credit for lower NYC’s crime rate, Gulliani did an excellent job improving that city and life for the people that live in it. Hopefully, other cities will follow th model he set.
I wrote a similar type of article on my blog this weekend titled “Temporary Trees and Software Decay” (http://evolutionarygoo.com/blog/?p=61). To quickly summarize, it is about two temporary trees in my backyard that were never removed. They looked nice, but started growing into other trees around then and causing a problem.
brad
on 27 Nov 06In fact the Freakonomics theory is only one of several competing theories to Broken Windows, and the existence of all these competing theories just goes to show that nobody really knows why crime decreased in NYC but that it’s easy to come up with many intuitively plausible reasons. It also goes to show that many people still don’t understand one of the basic foundational principles of science: correlation doesn’t necessarily indicate causation. So far we’ve seen a lot of correlation but not any ironclad demonstration of causation.
But anyway, I agree that applying the “broken windows” model to Web design is a valuable approach.
Alex
on 27 Nov 06If you were going to truly compare the Broken Windows theory to fixing bad links on your website, you would claim that improving your site’s navigation will fix the “big problems” of the web like spam and malware. Fortunately, you make no such claim.
J Lane
on 27 Nov 06Whether “Broken Windows” was the main cause in the drop in crime or not in NYC, it does work. People taking ownership of where they live (not just their house, but their neighbourhood) causes a buy-in from the community and a commitment to improve things overall.
Giuliani’s biography is a great read, I’d highly recommend it. So’s Freakonomics for that matter.
ML
on 27 Nov 06Fwiw, here’s Malcolm Gladwell’s take on the Freakonomics authors dismissal of Broken Windows: “I prefer to think of Freakonomics not as contradicting my argument in Tipping Point, but as completing it.”
Jason Martinez
on 27 Nov 06Paying attention to the little things is what makes good products great products. I think this example is out of context though. In a very broad perspective ok, the overall lesson is that the little things make a difference, yes true, they create trust and exhibit quality etc… But in the case of crime in New York City this method was used to basically “cook the books.” So murders are the really terrible bad crimes and there are 4 of them a week in a given area. But comparatively, there are 100 times people can be busted for being drunk in public, 100 squidgy men that can be busted and tons of other easy, little things going on. So the crime rate, because you focus on this easy (and admittedly, unimportant) crime makes it look as though the crime rate has gotten better drastically, because the numbers show it, what they don’t show is that its less safe because before, 2 of the 4 murderers would have been busted and only 1 are now because of the lack of focus, and now we’re paying more to house harmless people in jails when the real problem, murders has not improved at all or even gotten worse. Of course i do not think this applies to the web and creating great code or web products so the article is way off. The moral of the way out of context article is that paying attention to the little things helps.
Can’t wait for another sunspots, and as tons have probably written, thanks millions for rails!
ML
on 27 Nov 06Jason, it wasn’t just the small stuff. Murder and other serious crimes dropped significantly in NYC during this period.
andrew
on 27 Nov 06A better account than Levitt’s text on the theoretical incoherence of broken windows policing (or order-based policing), and the faulty social science behind it, is Bernard Harcourt’s Illusion of Order. Beyond Wilson’s complete lack of theorizing the very concepts which he purports to be studying (order and disorder), the two key studies supporting broken windows (Skogen 1990; Sampson and Cohen 1988) don’t hold up to more robust statistical analysis, and as Harcourt shows in chapter 4 of his book, there is simply no statistically significant evidence that the drop in crime experienced by new york during the 1990s (and, really importantly, every major US city regardless of changes in policing tactics) had anything to do with zero-tolerance, order-maintenance, or broken windows policing tactics.
Eugene Loj
on 27 Nov 06The analogy that jumps into my mind is launching a site before it is ready. This recently happened with a client. Their emphasis was on getting the site out “yesterday.” A big sticking point for us was on the content side. Our fear was that a user would visit the site and see it significantly lacking in content. Ultimately we had to fold to the client’s pressure.
To add to the challenge we gave the client access to the site through content management. Because of small visual changes they insisted on making and not following simple standards, the aesthetic value of the site was compromised.
The small issues matter on your web site. Your users will notice.
Ted
on 27 Nov 06Here is why the “broken window theory” was NOT responsible for the drop in petty crime and homicide in NYC:
- The drop in crime began 3 years before Giuliani took office, when Giuliani’s opponent, Dinkins, began to beef up the police force - It came during a period when NYC was rapidly expanding the police force. The drop in crime was to a large extent proportionate to the increases in the police force. I would not expect SvN to encourage larger team sizes!
Innovative policing strategies have been widely debunked as contributing in a major way to the drop in crime, even if Giuliani likes to spin it that way.
Here is what WAS responsible for the drop in crime: - adding people to the police force - changes in crack markets - quadrupling the number of people thrown in jail - and, as the Freakenomics book says, abortion policies 20-30 years prior
For more on this I recommend reading chapter 4 of Freakenomics.
Mario
on 27 Nov 06I dont disagree that the Broken Windows theory was effective in reaching the goals set out. No one disagrees that one very effective way of maintaining order is to rule with an iron fist.
My problem here is that it ignores why people are committing crimes to begin with. Broken Windows considers the only issue to be crime, not the miserable lives people are living that causes to commit crimes. Broken Windows doesnt care if people are poor or dying. It only cares that those people are making the city look bad.
So ya, taking care of the small things is very effective. But when you only look at crime, you miss the problems of abject poverty and other social issues.
Theres this nation, I forget which, in which grafitti is punished by beating with bamboo sticks. Its very effective. Crime is way down. But there’s no fucking way I’d want to live there.
Richard Muscat
on 27 Nov 06Re: “The result: All crime rates fell suddenly and continued to drop for the next ten years.”
There is evidence by Steven Levitt in Freakonomics
“1) Unwanted children are at higher risk for crime 2) Legalized abortion reduced the number of unwanted children
and therefore
3) Legalized abortion reduced crime.
It isn’t abortion per se that matters, it is the number of unwanted children.”
Not to say that your design-related argument is invalid, however the underlying logic on which it is based could be false.
Take-away value: What may seem at first sight to be a cause-effect relationship may often turn out to be not so at all. “Conventional wisdom is often much less than wise” and people like Rudy G. will always spin things politically in ways that benefit themselves.
Walter
on 27 Nov 06I think you crossed the line here, between useful insights and “any analogy can be turned into a lesson in web design”.
In attempt to find new wisdom, it seems to me, you, guys from 37signals and The Getting Real book, are starting to jump on any far fetched theory and insist on apply it to software and site design no matter how irrelevant it is to the matter at hand.
Soon you’ll tell us “since some people wish Elvis wasn’t dead to the point of thinking he is still alive, wishful thinking is part of any successful design.”
Stop! You are trying too hard. You start to talk and behave like a guru who has run out of inspiration. Go back to basics and critical thinking.
If you want absolutely say something deep this week, say something fundamentally true : “To prevent problems grow bigger catch them while they are still small. Don’t sweat the small stuff. However, pay attention to underlying details as they arise.”
Respectfully,
Walter
Walter
on 27 Nov 06To make my point clearer :
1) One theory, called the Broken Window theory, says you can discourage criminals from your community by displaying clearly and swiftly you don’t even tolerate the smallest mischiefs.
2) An elaborate construction cannot be sound if its constituents are not sound also. This is especially true with software. Even the smallest one are always an assembly of elementary instructions from which emerges higher level proprieties and behaviours.
The Broken Window theory doesn’t prove or logically lead to reasoning #2. Crime prevention theory #1 may be false or true; it doesn’t matter since it doesn’t apply to computer science. Therefore, there is no point of citing #1 as an illustration of #2.
Scott Parks
on 27 Nov 06This is a great article and one I started to implement recently by comparring web development to building construction.
When I drive by a building under construction I see something being done almost daily – it might be small, it might be something like the roof, but I see activity. Once the shell is up that is when the electric, water, etc go in, but again, you see progress.
My customers like to see something, anything to show I care and yes, I am working on it.
-Scott
ML
on 27 Nov 06I think you crossed the line here, between useful insights and “any analogy can be turned into a lesson in web design”.
Walter, I think the ideas here apply to any field of problem solving. And fwiw, the links in the post about Broken Windows and the web (“Software Entropy” and “Broken Windows Theory and Your Web Site”) come from other authors.
Matthew
on 27 Nov 06I’m in total agreement with RG on this, and to its effective application to Web design as stated, within a marriage/relationship, to raising children, etc.
The devil is in the details.
Dan
on 27 Nov 06I think that in terms of things like search engine optimization and semantic markup that makes a site more easy to upgrade as new standards and technologies become available this theory could apply. Fixing small problems (like closing your tags) can lead to large problems being very easy to handle (like upgrading your site from html to xhtml). People puzzle over how to get a better page rank from Google, but if they provided content that users wanted to come back and read every week and linked semantically then they wouldn’t even need to pay attention to the larger issue of search engine optimization. Small considerations turn what would have been a major inconvenience into something a web designer can take in stride.
Walter
on 28 Nov 06That’s only a guess. A personal hunch. You don’t have anything to support that portion of your exposé. The Broken Windows theory may be a great story to tell and an imaginative strategy to fight crime, still it doesn’t make it a valid point. And, it doesn’t matter if you generated the hypothesis yourself or borrowed it from someone else. All you can affirm as for now is that the concept is appealing to you. You would like it to be valid for any type of problem. But, nobody has demonstrate its universality. It has yet to been proven in its own field.
With that said, I don’t refute everything you wrote. Only the part where you implicitly argue the Broken Windows theory somehow supports the precepts you teach; in this case, that it is advantageous to keep users informed of corrections being made.
Charming stories, attractive images, pleasing analogies are great marketing tools. However, they don’t translate automatically to valid points for web design.
Pramit Singfh
on 28 Nov 06Great debate on ‘broken windows’ and ‘better websites’ but you might want to do something about spam comments like above – torism etc…
Sérgio Carvalho
on 28 Nov 06Thought I’d leave the link to Levitt’s paper: The Impact of Legalized Abortion on Crime
Anonymous Coward
on 28 Nov 06I love when people read one book (Freakonomics) and make up their mind. One book.
If you really want to know about a topic, you should read many points of views. Views you agree with, views you don’t agree with, neutral views.
Reading one best seller that disagrees with other studies does not mean those other studies are wrong.
DP
on 28 Nov 06Amen. Freakonomics is far from authoritative on this. Malcolm Gladwell is among their challengers.
After reading some comments about Broken Windows, feel compelled to chime in:
Broken Windows (or the philosophy it spawned, Community Policing – CP) is not a exclusively a NY phenomenon; it was embraced in one form or another by big city police departments thruout the 80s. To argue that a nationwide drop in crime disproves the impact of Broken Windows ignores the fact that it was used in many of those cities that experienced record drops in crime (such as Boston, home of the so-called “Boston Miracle.”)
Further, in later years Giuliani’s NYPD departed widely from the CP philosophy by imposing so-called Zero Tolerance policing. Two very different approaches to crime prevention.
Also, credit should be given to where it is due: the Broken Windows concept was developed by George Kelling and James Q. Wilson, and was implemented successfully by the NY Transit Police in the early 90’s – both occurred prior to Giuliani’s election.
NY stands out because their drop in crime was so out of proportion with the rest of the nation. Broken Windows is but one of many factors cited for New York’s crime drop (others being COMPSTAT, the hiring surge that occurred under Dinkins and Clinton, etc.)
Curt Sampson
on 29 Nov 06As an Extreme Programming coach, the, “do I clean up this little mess in the code” question is something I have to deal with all the time. Unfortunately, there’s never a simple answer to that. Never doing it until you really, really have to is guaranteed to lead you into a big mess; always fixing everything you see is almost certainly going to lead to wasting time and reducing productivity.
Mostly, you have to make judgement calls based on experience. Is the chunk of code something you need to do further work on in the near future? Are you likely to forget a lot about how it works before you next look at it? Do you know a good way to improve that code right now? Is the code likely to exist in a month or three, or is it likely to vanish due to other changes in the system? How much does this code affect other things in the system?
I’ve left gross hacks in systems without ill effect, and saved myself a lot of time and effort in doing so. Sometimes it’s code that sticks around, but is never modified again. Sometimes it’s code that quietly vanishes as other parts of the system stop using its functionality.
But it’s always a judgement call; there’s no one rule you can apply the same way all the time, any more than you can always follow the advice in, “Look before you leap,” or, “He who hesitates is lost.”
Reading a bit about the Dreyfus model of learning can help you understand the relationship between these sorts of rules and guidelines and how people of varying skills make these decisions. I found Patricia Benner’s From Novice to Expert to be a good introduction to this. It deals with nursing, but that field has some interesting parallels to computer programming.
This discussion is closed.