Find file
Fetching contributors…
Cannot retrieve contributors at this time
1053 lines (836 sloc) 61.5 KB
<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="">
<title><![CDATA[Accursed Ware]]></title>
<link href="" rel="self"/>
<link href=""/>
<name><![CDATA[John Krauss]]></name>
<generator uri="">Octopress</generator>
<title type="html"><![CDATA[Whither Rent Regulation]]></title>
<link href=""/>
<content type="html"><![CDATA[<p>The rent stabilization law that keep apartments affordable for millions of New
Yorkers was just renewed, <a href="">warts and all</a>.</p>
<p>The fight in Albany between tenant advocates and politicians centered around
closing loopholes in the law. The worst allows stabilized apartments to exit
the program when their rent gets high enough, resulting in the loss of
thousands of affordable apartments every year &mdash; far exceeding new construction
of affordable apartments.</p>
<p>But if you ask <a href="">the agency in charge of rent stabilization</a> which buildings
are losing stabilized apartments, they won&rsquo;t tell you. In fact, they reject
any Freedom of Information Law requests for data regarding the number of
apartments in the stabilization program by building, citing unclear privacy
<p>The only information they&rsquo;ll release is a <a href="">list of buildings</a> that have
apartments in the program. The list doesn&rsquo;t indicate whether every apartment in
a building is stabilized, or if only some are. <strong>Buildings that are 100%
stabilized look the same on this list as buildings with just one apartment left in
the program.</strong></p>
<p>The secrecy blanketing the stabilization program makes it very difficult to
understand how loopholes in the program affect affordability in different
neighborhoods over time. Not only does this make life hard for tenant
advocates, but it provides cover for landlords who fail to tell the state
(register) their stabilized apartments. Registration is voluntary &mdash; another
loophole in the law &mdash; and failure to do so could be an indication that they
are overcharging their tenants.</p>
<p>If a landlord doesn&rsquo;t like charging the legal rent, they can simply &ldquo;forget&rdquo; to
register. It&rsquo;s up to the tenant to take them to court to comply.</p>
<h3>Hidden in plain sight</h3>
<p>Remarkably, the number of stabilized apartments in each building over the last
seven years is hidden in plain sight, in property tax bills. With help from
a few civic hackers, I built <a href=""></a>, a collection
of every tax bill going back to 2008 for every building that might be
stabilized in New York City.</p>
<p>Putting together this site required downloading hundreds of thousands of tax
bills PDFs over several months because New York City&rsquo;s <a href="">Department of
Finance</a> (DoF) wanted $50,000 to mail me files that could already be found
free online. More on that in the next post.</p>
<blockquote><p>&hellip;approximately 1,000 hours at a cost to you of $50.36 per hour is needed to
fulfill your request&hellip; <cite><a href="">DoF Records Access Office</a></cite></p></blockquote>
<p>Using this data, I&rsquo;ve put together <a href="">a
that shows precisely where stabilized apartments have disappeared, remained,
or been constructed between 2007 and 2014.</p>
<p><noscript><img src="" alt="Changes in stabilized apartment counts 2007-2014" /><a href="">5</a></noscript></p>
<iframe width='100%' height='520' frameborder='0'
allowfullscreen webkitallowfullscreen mozallowfullscreen oallowfullscreen
<p>The map is meant to enable quick comparisons between four classes of buildings.</p>
<li><p><span style="color:red;"><strong>Major Losses</strong></span>, in red, buildings where
<em>50% or more</em> of the apartments in the building have left stabilization between
2007 and 2014. These buildings are densest in the Upper East Side of
Manhattan, where the high-rent vacancy loophole has allowed very large numbers
of apartments to exit the program. However, Bushwick is also a hotspot for this
class, which would suggest that landlords there are failing to register the
stabilized apartments in their buildings.</p></li>
<li><p><span style="color:#cc0;"><strong>Decreases</strong></span>, in yellow through
orange, buildings where <em>up to 50%</em> of the
apartments in the building have left stabilization. The Upper East Side again
leads the way in losses here, but there are significant patches of loss along
the high-density stabilized corridor from Prospect Park to the ocean in
Brooklyn along Ocean Ave, and along Queens Boulevard in Forest Hills.</p></li>
<li><p><span style="color:blue;"><strong>Steady</strong></span>, in blue, buildings where the
number of stabilized apartments has <em>remained the same</em> or very slightly
increased. In Manhattan, Chinatown and
Washington Heights have held their apartments relatively well; much of the Bronx
along the Grand Concourse has, too. Those landlords who are registering in
Bushwick, Brooklyn often are registering the same number of apartments 2014 as 2007.
Many buildings on Ocean ave in Brooklyn have held steady, too.</p></li>
<li><p><span style="color:green;"><strong>Increases</strong></span>, in cyan through green,
buildings where the number of stabilized apartments has <em>increased 10% or more</em>.
These are often new developments with affordable housing.</p></li>
<p>Buildings with more apartments have larger diameter circles, and precise unit
counts, subsidies, and addresses are available by hovering or clicking on
their circle.</p>
<p>Color is based off of the percentage change in stabilized apartments over the
total number of apartments in the building. So a building with 20 apartments
total that had 10 stabilized in 2007, and 5 stabilized in 2014, would be a 25%
loss. Its circle would be orange.</p>
<p>By using the map and its underlying data, advocates should be able to identify
<li><p>just starting to lose significant numbers of stabilized apartments</p></li>
<li><p>whose landlords have stopped registering with the state, and thus may contain
tenants at risk of overcharge</p></li>
<p>It is also possible to break down losses and gains by any political boundary,
so electeds can understand how loopholes are affecting rental affordability
for their constituents.</p>
<h3>Get the data!</h3>
<p>The underlying data is available in several CSV tables on
<a href=""></a>. The underlying scrapers, which download
the bills, and parsers, which turn the HTML pages and PDFs downloaded into
tabular data, can be found <a href="">on
<p>The guide to using the data is on
<a href="">GitHub</a>.</p>
<title type="html"><![CDATA[⅕ of NYC collision data wrong]]></title>
<link href=""/>
<content type="html"><![CDATA[<p>New York has really stepped up its open data game in the last year. Data sets
like <a href="">ACRIS</a> (the city property records) and <a href="">MapPLUTO</a> (GIS) that used to
cost thousands of dollars are now free. While it&rsquo;s in beta with limited
access, the city now makes available its official geocoder, too. This openness
is drawing into harsh relief excctly how poor the quality of the city&rsquo;s
collision statistics are.</p>
<p>The NYPD has been releasing intersection-level collision reports every month
since August 2011. I&rsquo;ve been collecting and processing these reports from an
obfuscated Excel (originally only PDF) format into <a href="">machine-readable
<p>These reports are the basis for <a href="">NYC Crashmapper</a>, an interactive map of
these collisions across the city. Unfortunately, the actual releases do not
include geographical information; they include only the borough, police
precinct, and street intersection. With some help from <a href=";group=1ab97dd0518e47e59e47686f479e241d">Tom Swanson</a>, most
of these intersections were geocoded using ArcGIS, while new intersections
appearing on the monthly report were batched through Google&rsquo;s public geocoder.</p>
<p>This process yielded longitude and latitude for about 97% of the almost 44,000
intersections with collisions on record for the last several years. However,
there were clearly problems with the data. Sometimes Google would get it
wrong, placing an intersection well outside the boundaries of New York, or
simply place an intersection in the wrong police precinct. Occasionally
I would do cleanings to filter these out, but at heart Google&rsquo;s geocoder is
something of a black box. I knew some percentage of the intersections had the
wrong longitude/latitude, but it was impossible to know precisely how many, or
how wrong they were.</p>
<p>While NYC&rsquo;s <a href="">own geocoder</a> has been floating around for a while, it was
originally a Windows-only executable. I messed around with it, but booting up
a Virtualbox and clicking around in a GUI couldn&rsquo;t be easily integrated into
an automated workflow.</p>
<p>Now the same data is available, although it requires an app registration and
delay (several days, for me), as the <a href="">Geoclient API</a>. Six REST endpoints
provide longitude and latitude, alongside a host of other info (city council
districts, police precincts, etc.) as JSON based off of address, intersection,
blockface, place name, BBL, or BIN input. I wrapped the API in <a href="">Python
bindings</a> to make it a bit easier to work with.</p>
<p>Upon receiving email confirmation from DoITT that my app had been approved to
use the API, I ran through the 44,000 intersections. There doesn&rsquo;t appear to
be any kind of rate limiting, and the process went relatively quickly. There&rsquo;s
no batch feature at the moment, so each request must be made individually.</p>
<tr><th></th> <th># of intersections</th> <th>%</th></tr>
<tr><th>Successfully geocoded</th> <td>36058</td> <td>82.37</td></tr>
<tr><th><i>&#8221;&#8221;&#8221; (precinct match)</i></th> <td><i>33700</i></td> <td><i>76.99</i></td></tr>
<tr><th><i>&#8221;&#8221;&#8221; (precinct mismatch)</i></th> <td><i>2358</i></td> <td><i>5.39</i></td></tr>
<tr><th>Streets do not intersect</th> <td>4139</td> <td>9.46</td></tr>
<tr><th>Streets intersect twice</th> <td>2651</td> <td>6.06</td></tr>
<tr><th>Streets intersect more than twice</th><td>626</td> <td>1.43</td></tr>
<tr><th>Invalid street name</th> <td>149</td> <td>0.34</td></tr>
<tr><th>Place name instead of street name, non-addressable</th><td>77</td><td>0.17</td></tr>
<tr><th>Misspelled street name</th> <td>32</td> <td>0.07</td></tr>
<tr><th>One street name was &#8216;UNKNOWN&#8217;</th> <td>21</td> <td>0.05</td></tr>
<tr><th>Place name instead of street name, addressable</th><td>19</td> <td>0.04</td></tr>
<tr><th>Other (&#8220;not part of&#8221;)</th> <td>2</td> <td>0.00</td></tr>
<p>The success rate was much less than the combined ArcGIS/Google approach: almost
18% of the intersections could not be associated with lon/lats. However, the
Geoclient&rsquo;s error messages are great. Instead of pulling the wrong lon/lat or
nothing at all, it identifies exactly what&rsquo;s going wrong: in almost 1 in 10
intersections in the reports, the identified streets <em>don&rsquo;t even intersect</em>.</p>
<p>In many cases, this can be chalked up to a misspelling, generally a missed
prefix or switched up Avenue, Street, or Place &mdash; for example, the intersection
of 32 Street and Flatbush Ave in Brooklyn. Clearly East 32 Street must have
been intended. In many other cases, the information must simply be wrong. One
intersection, of Ocean Parkway and Seton Place in Brooklyn, is merely
<p>The &ldquo;streets intersect twice&rdquo; phenomenon is interesting, and even approximately
solvable. This happens in cases like the <a href=",+Brooklyn,+NY+11235/@40.5807057,-73.9650921,17z/data=!4m2!3m1!1s0x89c24444753fe54d:0xc02ed3f455d9a605">intersection of Brighton 3 Street
and Neptune Avenue</a> in Brooklyn, where the cross street jogs
a short distance. It&rsquo;s possible to ask the geocoder for the more northerly or
southerly intersection, however, the NYPD data is still ambiguous as to which
was closer.</p>
<p class="center"><iframe src="!1m14!1m8!1m3!1d3030.2287594726517!2d-73.9650921!3d40.5807057!3m2!1i1024!2i768!4f13.1!3m3!1m2!1s0x89c24444753fe54d%3A0xc02ed3f455d9a605!2sNeptune+Ave+%26+Brighton+3rd+St%2C+Brooklyn%2C+NY+11235!5e0!3m2!1sen!2sus!4v1391291435007" width="400" height="300" frameborder="0" style="border:0"></iframe></p>
<p>The last significant chunk of missed geocoding is because of intersections
&ldquo;more than twice&rdquo;. This can happen in intersections like <a href=",+Brooklyn,+NY/@40.6257719,-73.9331992,18z/data=!4m2!3m1!1s0x89c24355b603e4cf:0xc31b80b14b1a7c1b">Avenue K and King&rsquo;s
Highway</a> where the layout of one of the streets results in
multiple crossings.</p>
<p class="center"><iframe src="!1m14!1m8!1m3!1d1514.0931806910887!2d-73.9331992!3d40.6257719!3m2!1i1024!2i768!4f13.1!3m3!1m2!1s0x89c24355b603e4cf%3A0xc31b80b14b1a7c1b!2sKings+Hwy+%26+Avenue+K%2C+Brooklyn%2C+NY!5e0!3m2!1sen!2sus!4v1391292147906" width="400" height="300" frameborder="0" style="border:0"></iframe></p>
<p>There were an additional slightly more than 5% of intersections that were
geocoded, but whose police precincts did not match. Since this is all
intersection-level data, this is to be expected in some cases: intersections
can skirt precinct boundaries, and in such cases it would be unclear which
precinct the &ldquo;real&rdquo; report would come from. However, there were odd cases
where the NYPD filed an intersection under the wrong precinct, for example,
filing the <a href=",+Brooklyn,+NY+11223/@40.5957238,-73.9650048,17z/data=!3m1!4b1!4m2!3m1!1s0x89c244f5dc18ba85:0xcae280f4057393a3">intersection of Avenue V and Ocean Parkway</a> in
Brooklyn in the 60, instead of the 61 precinct where it actually is.</p>
<p class="center"><iframe src="!1m14!1m8!1m3!1d3029.5483480885487!2d-73.9650048!3d40.5957238!3m2!1i1024!2i768!4f13.1!3m3!1m2!1s0x89c244f5dc18ba85%3A0xcae280f4057393a3!2sOcean+Pkwy+%26+Avenue+V%2C+Brooklyn%2C+NY+11223!5e0!3m2!1sen!2sus!4v1391272537346" width="400" height="300" frameborder="0" style="border:0"></iframe></p>
<p>Even if the data quality on the existing reports were perfect, the information
would be hobbled by the identification by intersection. The data is assembled
from the state&rsquo;s <a href="">MV-104</a> forms, which do not have to specify a location more
precisely than an intersection. Officers fill out these forms on-site, and any
mistakes or inaccuracies on the form will be entered into the database, without
any sort of further checking.</p>
<p>Google was far more liberal dispensing coordinates. Its geocoder was able to
provide coordinates for the vast majority of the intersections Geoclient
couldn&rsquo;t, however these coordinates are very suspect.</p>
<tr><th></th> <th># of intersections</th><th>%</th>
<tr><th>Geocoded in both</th> <td>36055</td> <td>82.37%</td>
<tr><th><i>&#8221;&#8221;&#8221; (&gt; 0.01 lat/lon difference)</i></th><td><i>180</i></td><td><i>0.41%</i></td>
<tr><th><i>&#8221;&#8221;&#8221; (&lt; 0.01, &gt;0.0003 lat/lon difference)</i></th><td><i>378</i></td><td><i>0.86%</i></td>
<tr><th>Google only</th> <td>6490</td> <td>14.83%</td>
<tr><th>Geoclient only</th> <td>3</td> <td>0.00%</td>
<tr><th>Neither</th> <td>1226</td> <td>2.80%</td>
<p>In cases where both geocoders captured a lon/lat but there was a significant
difference, the NYC geocoder was superior. For example, it was able to capture
the <a href=",-73.9764686,18z/data=!4m7!1m4!3m3!1s0x89c2598378c717a7:0x266d24f6c67a6b6d!2s27+1st+Ave,+New+York,+NY+10003!3b1!3m1!1s0x0:0x649aab3e087f8735">intersection of East 27 St and 1 Avenue</a> in Manhattan. Google was
unable to understand that East 27 St continues even though it is closed off to
cars, and erroneously identified the intersection of East 2 St and 1 Ave.</p>
<p class="center"><iframe src="!1m18!1m12!1m3!1d1511.5139148409382!2d-73.97646857235236!3d40.73941314916777!2m3!1f0!2f0!3f0!3m2!1i1024!2i768!4f13.1!3m3!1m2!1s0x0%3A0x649aab3e087f8735!2sNYU+Med+Center%3A+Hilz+Max+J+MD!5e0!3m2!1sen!2sus!4v1391294585211" width="400" height="300" frameborder="0" style="border:0"></iframe></p>
<p>Even fairly minute differences in longitude or latitude signal problems with
the Google geocoding &mdash; for example, a difference as small as 0.0003 degree
indicates that Google mixed up Beach Terrace and Oak Terrace on their
<a href="'33.62%22,+-73%C2%B054'48.60%22/@40.80934,-73.9135,17z/data=!3m1!4b1!4m2!3m1!1s0x0:0x0">intersection across Beekman Ave</a> in the Bronx.</p>
<p class="center"><iframe src="!1m18!1m12!1m3!1d1509.9293108982622!2d-73.9134544024479!3d40.80910044385041!2m3!1f0!2f0!3f0!3m2!1i1024!2i768!4f13.1!3m3!1m2!1s0x0%3A0x0!2zKzQwwrA0OCczMy42MiIsIC03M8KwNTQnNDguNjAi!5e0!3m2!1sen!2sus!4v1391296500495" width="400" height="300" frameborder="0" style="border:0"></iframe></p>
<p>While the Geoclient geocoder is not able to make clear statements about many of
the intersections in the NYPD collision dataset, it does a far better job for
those it identifies. Those it cannot, which Google does, are ambiguous in the
cases where the Geoclient&rsquo;s error was that there are multiple intersections
(about 7.5%), and generally indicate an NYPD data entry error in the cases
where Geoclient recognizes no intersection (almost 10%).</p>
<p>What can be done now that it&rsquo;s so clear how problematic these reports are?</p>
<li><p>The NYPD should be pressed to cooperate with DoITT to geocode intersections
on the reports they file, and double check those that don&rsquo;t pass through the
geocoder. Even if the reports, being state mandated, can&rsquo;t be modified, the
geocoder is the perfect tool to catch bad data at the time of entry.</p></li>
<li><p>If the Geoclient API identifies a problem, the officer who entered
problematic information should be asked to clarify. The correction or
comment would be entered into the monthly release.</p></li>
<li><p>The NYPD could switch to a daily feed of MV-104 reports, and such
a check could be worked directly into the filing process.</p></li>
<title type="html"><![CDATA[A YouTube CLI for Mac]]></title>
<link href=""/>
<content type="html"><![CDATA[<p>Out of laziness, I mostly listen to music off <a href="">YouTube</a> these days. As a
library it&rsquo;s pretty unbeatable, and I don&rsquo;t have to worry about plugging in an
external HDD with a big library.</p>
<p>Using the browser for selecting songs is a drag, but I&rsquo;ve always used
visual interfaces (originally iTunes, then a combination of <a href="">VLC</a> and
<a href="">Vox</a>). Also, Flash player munches on your battery pretty bad. And
I generally try to avoid executing anything with Flash.</p>
<p>Turns out there&rsquo;s a CLI for YouTube, <a href="">youtube-viewer</a> with vim-like
keybindings no less! Since it requires a newer version of Perl than what&rsquo;s
bundled with OSX, you need to do a little work to get it running.</p>
<h3>Perl 5.16.0</h3>
<p>The Perl version bundled with OSX in <code>/usr/bin/perl</code> with Mac OSX 10.7.5, is
5.12.3. Considering Apple&rsquo;s tendency to let the CLI tools bundled with OSX
rot, I doubt newer versions have a more up-to-date version. It&rsquo;s entirely
possible that some system-level functionality depends on an old version of
Perl, so updating at the root level would be a Very Bad Idea.</p>
<p>Fortunately, there&rsquo;s a <a href="">good setup tool</a> available on <a href=""></a>:</p>
<pre><code class='bash'>$ curl -L | bash</code></pre>
<p>This will install 5.16.0 in <code>~/perl5/perlbrew</code> and add a line to your
<code>.bashrc</code>, updating your env Perl version.</p>
<p>This one&rsquo;s easy.</p>
<pre><code class='bash'>$ brew install mplayer</code></pre>
<p>Snag the source:</p>
<pre><code class='bash'>$ git clone</code></pre>
<p>Follow the instructions, but drop the last <code>sudo</code> &mdash; provided you make
sure to execute <code>Build.PL</code> with your perlbrew (rather than system) Perl, then
it will be installed in a user-specific directory.</p>
<pre><code class='bash'>$ perl Build.PL
$ ./Build
$ ./Build test
$ ./Build install</code></pre>
<pre><code class='bash'>$ youtube-viewer
=&gt;&gt; Search for YouTube videos (:h for help)</code></pre>
<p>Or, if you don&rsquo;t want an MPlayer window popping up outside your Terminal while
a video is playing:</p>
<pre><code class='bash'>$ youtube-viewer -n
=&gt;&gt; Search for YouTube videos (:h for help)</code></pre>
<title type="html"><![CDATA[Introducing NYC Crashmapper]]></title>
<link href=""/>
<content type="html"><![CDATA[<p>For over a year, I&rsquo;ve been archiving and processing the New York City Police
Department&rsquo;s collision data on the <a href="">Crash Data Band-Aid</a>.</p>
<p>With some help from <a href="">generous people</a>, all valid intersections have been
geocoded in the past several months. So now all the collision data, in addition
to being free for download above, can be visualized and analyzed in your
browser using <a href="">Crashmapper</a>!</p>
<p><a href=""><img src="" alt="NYC Crashmapper" /></a></p>
<p>The project is free and open source, you can browse the code <a href="">on Github</a>
(it&rsquo;s in the <code>map</code> directory).</p>
<p>Building the map&rsquo;s been fun. There&rsquo;s no server-side app &mdash; it&rsquo;s just
a static site on <a href="">S3</a> &mdash; and all data points for some 30K intersections
over more than 24 months are loaded into your browser for fast analysis. The
base layers are from <a href="">OSM</a>, and the mapping library is <a href="">Leaflet</a>. I&rsquo;m
using a slightly optimized <a href="">fork</a> of the awesome <a href="">Leaflet.markercluster</a>
plugin. The &ldquo;heatmap&rdquo; is really a bunch of CSS drop shadows around marker
clusters. The real deep-link custom URLs are possible by setting the error
page on S3 to <code>index.html</code>, so that all links are served up through the app&rsquo;s
<a href="">Backbone</a> router.</p>
<p>Crashmapper could be adapted to your town. Contact me on the left if you&rsquo;re
<title type="html"><![CDATA[New Google Maps Mutilates Bike Layer]]></title>
<link href=""/>
<content type="html"><![CDATA[<p><em><strong>Update 8/1</strong>: Google fixed it yesterday! Thanks everyone who filed tickets
with them, maybe it made a difference. :)</em></p>
<p>Anyone who bikes in New York City and uses <a href="">Google Maps</a> to look up their
route is in for an unpleasant surprise.</p>
<p>The new-look Google Maps totally mutilates NYC&rsquo;s previously excellent bicycle
lane layer. I wouldn&rsquo;t be surprised if it messed it up elsewhere. Since the
old Google Maps looks to be gone for good, the best that can be presented for
comparison is <a href="">OSM</a>&rsquo;s layer:</p>
<p><img src="" alt="Brooklyn Bike paths in OSM" /></p>
<p>Here&rsquo;s the new, messed up Google Maps:</p>
<p><img src="" alt="Brooklyn bike paths in Google Maps" /></p>
<p>Bergen and Dean St., major routes for bikers in northern Brooklyn going east
or west, are gone, along with Flushing Ave.</p>
<p>Manhattan didn&rsquo;t do any better:</p>
<p><img src="" alt="Manhattan bike paths in Google Maps" /></p>
<p>First and Second Aves, which have miles of dedicated bike lanes separated from
auto traffic, are unmarked. It&rsquo;s the same for 8th and 9th Aves. Google&rsquo;s
negligence renders the island&rsquo;s respectable bike lane network invisible.
Here&rsquo;s the reality:</p>
<p><img src="" alt="Manhattan bike paths in OSM" /></p>
<p>A bike share system just rolled out here, and this is an awful time to make it
hard for new, inexperienced bikers to figure out where the safest routes are.
The CitiBike app does have paths on it, but it&rsquo;s still more intuitive for
many people to use tried-and-true Google Maps, especially if they&rsquo;re looking up
a route on their computer beforehand.</p>
<p>Google had this data before, and it&rsquo;s unclear why they shouldn&rsquo;t have it now.
Here&rsquo;s to hoping they notice, actually care, and fix it quick.</p>
<title type="html"><![CDATA[New NYC buildings less efficient]]></title>
<link href=""/>
<content type="html"><![CDATA[<p>Large buildings constructed in New York City since 1935 are by today&rsquo;s
standards less energy efficient.</p>
<script type="text/javascript"
New","title":"Median \"Energy Star\" over year built,
Built (5-year
1"} </script>
<p>Reading <a href="">this article</a> from Metropolis Mag on what&rsquo;s wrong with green
architecture, and <a href="">LEED</a> certification in particular, I noticed references to
New York City measuring energy efficiency for some famous buildings: the
Chrysler Building, the new 7 World Trade Center, and the Seagram Building,
amongst others. Drawing from a <a href="">Times article</a>, they observed that the newer
buildings, even LEED certified ones, performed worse:</p>
<blockquote><p>Among other reasons for this failing, the Times pointed to the widespread
use of expansive curtain-wall glass assemblies and large, “deep-plan” designs
that put most usable space far from exterior walls, forcing greater reliance on
artificial light and ventilation systems.</p>
<p><i><small>from <a href=""></a></small></i></p></blockquote>
<p>The Times and Metropolis mag stuck to anecdotal evidence, not looking beyond
a few buildings. Fortunately, the city put all the scores <a href="">online</a>.</p>
<p>So it&rsquo;s easy to test: are buildings constructed since the advent of mid-century
modern building techniques less energy efficient?</p>
<p>Of 4000 reports, about 1600 recorded an actual &ldquo;Energy Star Score&rdquo;,</p>
<blockquote><p>ENERGY STAR Score: A 1-to-100 percentile ranking for specified building
types, as calculated by Portfolio Manager, with 100 being the best score and 50
the median. It compares the energy performance of a building against the
Commercial Buildings Energy Consumption Survey (CBECS), a national database,
and independent industry surveys for that building type. This rating is
normalized for weather and building attributes in order to obtain a measure of
<p><i><small>from <a href=""></a></small></i></p></blockquote>
<p>1600 ain&rsquo;t a bad sample size. And what buildings are on the sheet?</p>
<blockquote><p>&hellip;all privately-owned properties with individual buildings over 50,000 square
feet or with multiple buildings with a combined square footage over 100,000
square feet&hellip;</p>
<p><i><small>from <a href=""></a></small></i></p></blockquote>
<p>Big office buildings.</p>
<p>The essential context, &ldquo;When was this building built?&rdquo; was missing. But there
were <a href="">BBL</a>s, the NYC property identification number. Since the <a href="">Department
of Finance</a> makes available BBL-indexed <a href="">tax roll data</a> with when buildings
were built, all&rsquo;s that had to be done was to download the tax roll, <a href="">convert
it</a> from MS Access format (fun!) and join it to the environmental ratings.</p>
<p>Keeping the sample size respectable by looking in five-year-increments, the
pattern is clear: large buildings constructed prior to 1935 do much
better by today&rsquo;s energy efficiency standards than those constructed after.</p>
<p>I&rsquo;m making the <a href="">environmental data joined to year built</a> available as a CSV
for anyone to play with. Below is a chart including sample sizes, for the
statistically curious.</p>
<script type="text/javascript"
New","booleanRole":"certainty","title":"Median \"Energy Star\" over year built,
Built (5-year
1"} </script>
<title type="html"><![CDATA[Preparing a Mac laptop for your Grandma]]></title>
<link href=""/>
<content type="html"><![CDATA[<p>My grandma got a new laptop for Mother&rsquo;s Day. She&rsquo;s in her nineties, and first
started using a computer a few years ago. Email, watching music on
YouTube, and playing bridge online are now essential parts of her life.
Her first computer was an old iBook, which still works fine, but the services
she uses have rendered it unusable:</p>
<h3>The pace of obsolecence on the Internet is unacceptable.</h3>
<p>A huge number of websites use JavaScript for questionable bells and whistles,
under the assumption everyone has a really fast JS interpreter. A PowerPC
running OSX 10.4 simply can&rsquo;t run it fast enough, and these sites don&rsquo;t degrade
gracefully &mdash; they freeze up.</p>
<p>Flash doesn&rsquo;t make this better. Adobe simply gave up on this at a certain
point, and <a href="">Ten Four Fox</a>, which
was her only option after a certain point, didn&rsquo;t really properly support Flash
either. Losing YouTube was harsh for her, and I can&rsquo;t imagine why it was
really necessary &mdash; when she first got the computer, it worked fine.</p>
<p>The computer was fine, but the internet dropped support for her. And in the
most frustrating way: everything ran slower, broke in random and unexpected
ways, and it was very hard to explain to her what was going on, why, or what to
do about it. Sadly, the only option was to get her a new computer.</p>
<h3>Setting up a new Mac for an old person is an epic pain in the ass</h3>
<p>It&rsquo;s probably not better on Windows or Linux distributions, and since she&rsquo;d
learned on a Mac, that was the only practical option. Fortunately the
interface didn&rsquo;t change too drastically between 10.4 and 10.8, but the factory
setup of the OS is piss poor for an old person. Because either the guest
account was active by default, or she (or someone else) accidentally activated
it, I had the pleasure of setting her computer up to make it accessible to her
several times. So the steps got stuck in my head:</p>
<li><p><strong>Disable the guest account.</strong> Go to System Preferences, unlock
your prefs, click on &ldquo;Guest User&rdquo;, and uncheck &ldquo;Allow Guests to log on to
this computer&rdquo;. Otherwise, it&rsquo;s possible to log on, set things up, save
files, and set up bookmarks on Safari &mdash; and lose everything as soon as you
log out.</p></li>
<li><p><strong>Auto log in the correct account.</strong> In &ldquo;Security and Privacy&rdquo;, uncheck
&ldquo;Require password for sleep and screen saver&rdquo;. Uncheck &ldquo;Disable automatic
login&rdquo;. Select the proper user in &ldquo;Automatically log in as&rdquo;, enter their
password, and press &ldquo;OK&rdquo;. For bonus points, you could set up a separate
non-admin account for them, although I didn&rsquo;t do this.</p></li>
<li><p><strong>Allow apps only from the app store.</strong> I didn&rsquo;t do this, but I probably
should have &mdash; she doesn&rsquo;t use any applications that aren&rsquo;t in-browser, so
anything else is probably a security risk.</p></li>
<li><p><strong>Make the cursor bigger.</strong> She could barely see the cursor on the new
computer, partly because the screen resolution was much higher. Go to
&ldquo;Universal Access&rdquo;, then &ldquo;Mouse and Trackpad&rdquo;, and make the cursor bigger for
<li><p><strong>Make everything bigger.</strong> Her new laptop has the same size screen, but much
higher resolution. LCDs don&rsquo;t look great below their natural resolution, but
for someone without great eyes, the slight blurriness is not noticeable at
all &mdash; and it&rsquo;s the best way to make sure everything is bigger. I cut her
resolution from 1440 by 900 to 1024 by 640. Go to &ldquo;Displays&rdquo;, then
&ldquo;Display&rdquo;, and reduce the resolution.</p></li>
<li><p><strong>Always make the screen as bright as possible.</strong> Still in &ldquo;Displays&rdquo;,
increase the brightness to maximum. Uncheck &ldquo;Automatically adjust
brightness&rdquo;. Then, go to &ldquo;Energy Saver&rdquo;, then &ldquo;Battery&rdquo; and disable
&ldquo;Slightly dim the display when using this power source&rdquo;.</p></li>
<li><p><strong>Increase general contrast.</strong> Go to &ldquo;Universal Access&rdquo;, then &ldquo;Seeing&rdquo;, and
raise the &ldquo;Enhance contrast&rdquo; setting a bit underneath &ldquo;Display&rdquo; &mdash; a little
bit goes a long way.</p></li>
<li><p><strong>Make the cursor slower.</strong> By default, the cursor is pretty jumpy if you
don&rsquo;t have very sure fingers &mdash; go to &ldquo;Trackpad&rdquo; and lower the tracking
<li><p><strong>Turn off two-finger scrolling.</strong> In &ldquo;Trackpad&rdquo;, uncheck everything under
&ldquo;Scroll &amp; Zoom&rdquo;.</p></li>
<li><p><strong>Turn off trackpad actions, especially fancy multi-touch actions.</strong> In
&ldquo;Trackpad&rdquo;, uncheck everything under &ldquo;Point &amp; Click&rdquo; and &ldquo;More Gestures&rdquo;.</p></li>
<li><p><strong>Maximize and always show the Dock.</strong> Go to &ldquo;Dock&rdquo;. Drag the &ldquo;Size&rdquo; slider
to make it big as possible. Uncheck &ldquo;Automatically hide and show the Dock&rdquo;.</p></li>
<li><p><strong>Turn off Hot Corners, Exposé, and all the other shit that unexpectedly
changes screen layout.</strong> This is a big one, and can be extremely hard to
explain over the phone when something goes wrong. Go to &ldquo;Mission Control&rdquo;,
and set all the &ldquo;Keyboard and Mouse Shortcuts&rdquo; to &ldquo;&ndash;&rdquo;. Then click &ldquo;Hot
Corners&hellip;&rdquo; and set all of them to &ldquo;&ndash;&rdquo;, too.</p></li>
<li><p><strong>Remove unnecessary crap from the Dock.</strong> The only app my grandma uses is
Safari, really. Sometimes on the phone it&rsquo;s helpful to have her go to System
Preferences. I removed everything from the dock excepting those two things,
which also made the Dock bigger, and easier for to click on Safari.</p></li>
<li><p><strong>Remove unnecessary crap from the menu bar.</strong> By default, Bluetooth and
Time Machine are in the menu bar, and are useless and possibly confusing.
Since she doesn&rsquo;t store any files on the computer, Time Machine is entirely
unnecessary. Go to their respective preference panes and remove them.</p></li>
<li><p><strong>Always show scroll bars.</strong> Personally, I hate Apple&rsquo;s move towards hiding
scrollbars unless you maneuver the mouse to that tiny space on the edge of
a window you need to scroll. And it sucks really, really hard for someone
has trouble with slight trackpad movements. Go to &ldquo;General&rdquo; in &ldquo;System
Preferences&rdquo;, and select &ldquo;Always&rdquo; under &ldquo;Show scroll bars&rdquo;. While you&rsquo;re
there, you may as well change sidebar icon size to &ldquo;Large&rdquo;.</p></li>
<li><p><strong>Make fonts bigger in Safari.</strong> I set her up with Safari just because it&rsquo;s
bundled with the OS and is the most straightforward, although surely Firefox
and Chrome have similar options. Go to &ldquo;Preferences&rdquo;, then &ldquo;Advanced&rdquo;, then
check off &ldquo;Never use font sizes smaller than&rdquo;. I set hers to 16, although
this would be a little risky for someone who browses more random sites (her
repertoire is pretty set.) Some sites are nigh-on unusable when you futz with
their formatting that much. Probably 12 is safe.</p></li>
<li><p><strong>Make the Safari homepage something useful.</strong> Go to &ldquo;Preferences&rdquo; in
Safari, then set both &ldquo;New windows open with&rdquo; and &ldquo;New tabs open with&rdquo; to
&ldquo;Homepage&rdquo;. Then set the homepage to Google, or something else that&rsquo;s not</p></li>
<li><p><strong>Log into all the services. Make sure they stay logged in.</strong> I logged her
into Gmail, and checked off the option to keep her logged in.</p></li>
<li><p><strong>Remove stupid bookmarks.</strong> By default, Safari is loaded with a bunch of
bookmarks she didn&rsquo;t need. She doesn&rsquo;t use Facebook or Twitter. She doesn&rsquo;t
use iCloud. I have no idea who uses the &ldquo;News&rdquo; or &ldquo;Popular&rdquo; folders. I got
rid of them all.</p></li>
<h3>Some stuff just sucks and you can&rsquo;t fix it</h3>
<p>The Adobe Flash updater is really confusing, requires multiple clicks just to
upgrade, it doesn&rsquo;t explain what it does or why it needs to run. It&rsquo;s
extremely difficult to explain to someone without a pretty solid comfort with
computers what the hell is going on when it keeps popping dialog boxes in front
of them. But keeping Flash up to date is important, and necessary for yotube
Hopefully Flash will die soon and this can finally be done away with.</p>
<p>The Notification Center menu item, and App Store stuff in general, is new
and unnecessary for her in 10.8. Apple&rsquo;s doesn&rsquo;t let you remove it. That
sucks, but it&rsquo;s something that hopefully won&rsquo;t cause any problems.</p>
<p>iCloud is a pain in the ass in a similar way. I wasn&rsquo;t sure exactly how it was
happening, but my Grandma kept asking me questions about iCloud that make me
think it randomly pops up stuff asking her to set it up. Since it didn&rsquo;t
happen when I was there though, I couldn&rsquo;t be sure how to make it shut up.</p>
<h3>In conclusion</h3>
<p>It&rsquo;s nuts that to make a new laptop relatively unthreatening and accessible for
an enthusiastic 90-something requires going across a dozen System Preferences
panes and sub-panes, and simply having to accept certain things as inevitably
confusing. I don&rsquo;t know what the solution is, and it may not be possible to
make the set up something that my grandma itself could do, but the steps above
took nearly an hour to figure out (and a good 20 minutes to reproduce.)</p>
<p>Furthermore, most of the internet is a usability nightmare for the elderly.
Tasks like clicking on a form field to fill it out, seeing that a white menu
drop down on a white menu is, in fact, a drop-down, or navigating to a &ldquo;Log in&rdquo;
button can become idiotically arduous tasks. The fad for hiding controls when
you&rsquo;re not using them (looking at you, scroll bars) is particularly
frustrating. Controls themselves are often tiny, with no clear way to make
them bigger &mdash; YouTube is particularly bad with this.</p>
<p>Setting up this laptop several times, and seeing her struggle through the
usability nightmare that is most of the internet, made me think about how
little account I take of usability in my own designs. It&rsquo;s hard &mdash; without
seeing someone actually work through these problems, the choke points aren&rsquo;t
obvious. In all honesty, I don&rsquo;t think it&rsquo;s practical for web designers to do
so: the expense is too great and the impact is small. The makers of OSs and
browsers are in a far better position to make using the internet more
manageable for the elderly.</p>
<title type="html"><![CDATA[JS-Free Live Chat]]></title>
<link href=""/>
<content type="html"><![CDATA[<p>I built a <a href="">real-time web-based chat</a> without a line of client-side code. No
JavaScript, no Flash, no (ew) Java applet. 100% HTML. It even works in
<a href="">Internet Explorer 4</a>, emulated with <a href="">SheepShaver</a>.</p>
<p><a href="" target="_blank">
<img src="" width="640" height="311" />
<p>Because I remember the canonical <a href="">node.js</a> example being a chat server, I
had to call it <a href="">No.JS</a> (no-dot-JS, got it? Haha. Sorry.)</p>
<p><img src="" alt="No (dot) JS" /></p>
<p>To be fair, the backend could have been built in JavaScript using Node, instead
of Python using <a href="">Brubeck</a>. It would make for the dandy perversity of having
exclusively server-side JavaScript.</p>
<p>All the code for it is <a href="">on GitHub</a>, so all you Node-hackers should go ahead
and build a JavaScript edition&hellip;</p>
<hr />
<p>Why&rsquo;d I do this, and how does it work?</p>
<p>I&rsquo;m building a real-time browser-based multiplayer game, using the same
templates (<a href="">Mustache</a>) on the client and server-side. Instead of messing
around with WebSockets, I used AJAX long polling to push updates from the
server to the client. Since the server could render all the views, I set up a
little <code>Accept</code> header goodness so that browsers with JavaScript turned off
could still play&hellip; but the problem of pushing the state remained.</p>
<p>There is the peculiarly irritating <code>&lt;meta http-equiv="refresh" ...&gt;</code> tag,
embedded in my consciousness from some late-90s &ldquo;dynamic&rdquo; interfaces. Every
few seconds, the page would flash to refresh with possibly new content.
Maddeningly, you would be thrown back to the top of an identical (or very
slightly modified) page.</p>
<p>Throwing the meta refresh inside a <code>&lt;noscript /&gt;</code> would make the game kind-of
work, while not irritating the majority of JavaScript-enabled users.
The refresh still wouldn&rsquo;t be live, though. And the more frequent it was,
the more irritating it would be.</p>
<p>So why not just long-poll the whole page?</p>
<p>Browsers wait for a response before wiping the contents of a window. The prior
page works the same while a new page is loading. The server can dictate
exactly when the page should reload.</p>
<p>The stakes are higher than when long polling an AJAX request. If the server
doesn&rsquo;t get back to the client with a response eventually, the browser doesn&rsquo;t
merely log to the error console &mdash; it replaces the page with a message that
the server is down. Responding with identical content at least twice a minute
prevents this.</p>
<p>To put it all together, the Mustache template would look like this in the
&lt;!-- Only fuss with this for those without JavaScript. --&gt;
&lt;!-- The content of the refresh header is generated by the handler. --&gt;
&lt;meta http-equiv="refresh" content=&#123;&#123;refresh&#125;&#125; /&gt;
<p>And this in the body:</p>
&lt;!-- Exciting rapidly changing content! --&gt;
<p>The function to generate content, using <a href="">Redis</a>, would look like:</p>
<pre><code class='python'>def get_page_content(old_id):
# If an old ID was specified, and it&#39;s equal to the current ID,
# subscribe to updates on the &#39;updates&#39; channel of our Redis database.
# Then listen for an update. Whatever triggers the update should also
# modify the ID.
if old_id and old_id == db.get(&#39;id&#39;):
pubsub = db.pubsub()
return {
# Populate the refresh header in the template with the new
# ID. The `0;` means that the browser will immediately
# try to reload the page -- but will hang, because the
# reload is for the new ID.
&#39;refresh&#39;: &quot;0; url=?id=%s&quot; % db.get(&#39;id&#39;),
# This would be a separate function, but you get the point.
&#39;content&#39;: json.loads(db.get(&#39;content&#39;))
<p>The handler, using <a href="">Brubeck</a> with <a href="">gevent</a>, would look like:</p>
<pre><code class='python'>class PageHandler(MustacheRendering):
# The route is specified elsewhere, but this method will be called
# when there is a vanilla HTTP GET for the page.
def get(self):
# Pull out the last loaded page ID. Will be `None` if none was
# specified.
old_id = self.get_argument(&#39;id&#39;)
# See `get_page_content` above.
context = gevent.timeout.with_timeout(30, get_page_content, old_id)
return self.render_template(&#39;page&#39;, **context)
except gevent.timeout.Timeout:
# The timeout above was for thirty seconds. If nothing changes
# after that time, the client is simply redirected back to the
# original page with no ID argument at all. This will cause
# an immediate reload.
# We do this instead of redirecting to `&quot;?id=%s&quot; % old_id`
# because that would cause another hang, and possibly another
# 3xx response if we hit this timeout again. If nothing
# changes for a few minutes, which is likely, then the repeated
# redirects would wipe the page with a `too many redirects`
# error. This avoids that eventuality. :)
return self.redirect(&#39;?&#39;)</code></pre>
<p>No.JS is pulled from the game&rsquo;s chat component. Chat is tricky, because a
refresh wipes forms. You&rsquo;re typing, someone says something&hellip; and everything
you were about to say goes to the bitbucket. Keeping the chat messages inside
an <code>&lt;iframe /&gt;</code> solves this problem. I switched it out for a regular frame
in No.JS to support older browsers.</p>
<title type="html"><![CDATA[NYPD crash data band-aid]]></title>
<link href=""/>
<content type="html"><![CDATA[<p>Yesterday, <a href="">Streetsblog</a> wrote about the previous day&rsquo;s New York
City Council hearing on traffic safety. There have been several
particularly gruesome incidents in the past year where New York
drivers maimed and killed pedestrians and cyclists. The New York City
Police Department has not only barely investigated the crime scenes,
but it has been incredibly callous towards the victims, their
families, and the press.</p>
<p><a href="">This part</a> stuck out to me:</p>
<blockquote><p>Council Member Jessica Lappin got into an animated discussion with
Petito over traffic crash data. When Lappin asked why NYPD is
releasing data in PDF form — and only after the council adopted
legislation forcing the department to do so — Petito replied that the
department is &ldquo;concerned with the integrity of the data itself.&rdquo;
Petito said NYPD believes data released on a spreadsheet could be
manipulated by people who want &ldquo;to make a point of some sort.&rdquo; An
incredulous Lappin assured Petito that the public only wants to
analyze the data to improve safety, not use it for &ldquo;evil.&rdquo;</p></blockquote>
<p>I already knew that they were finally releasing &mdash; after the Council
forced them to &mdash; crash data as <a href="">idiotically obfuscated PDFs</a>, but
reading that they justified this out of concern for &ldquo;the integrity of
the data,&rdquo; was so galling that it goaded me into action. I would make the
data accessible as friendly, <a href="">parseable CSVs</a>.</p>
<p>After all, isn&rsquo;t the purpose of data to &ldquo;make a point of some sort&rdquo;?</p>
<p>Since I am hosting copies of their data in a format (spreadsheets, oh
my!) amenable to the forces of evil on my server, I recommend that you
download the code to roll your own NYPD traffic crash CSVs:</p>
<pre><code>git clone
<p>Once you&rsquo;ve got it, run the shell script:</p>
<pre><code>cd nypd-crash-data-bandaid/
<p>You&rsquo;ll need python, <a href="">xpdf</a>, and wget. Full instructions are in the
readme on <a href="">Github</a>.</p>
<p>The NYPD replaces the previous month&rsquo;s files every month. Thus I&rsquo;ve
included a sample crontab, which checks the site for a new PDFs every
day. Of course, they don&rsquo;t rename the files, which means that it&rsquo;s
necessary to download one and do the PDF conversion merely to
ascertain what month it applies to. Incredible.</p>
<p>There&rsquo;s an extra script,, which makes it easy to generate RSS feeds for when the NYPD has updated their data. You can see it in action <a href="">here</a>.</p>
<p>I&rsquo;d like to thank David Turner for writing the <a href="">original</a> NYPD crash
data PDF->text converter, which was only linked to in an article from
a <a href="">few months back</a>. I had to modify the output to reflect the
multidimensionality of the vehicle types and accident categories in a
flat file, but the parser itself was golden. I&rsquo;d also like to thank
Matthew Kime, who came up with the band-aid idea for the name.</p>
<p>Now we just need someone to compare this to the NYS Department of
Transportation&rsquo;s crash data as visualized through Transportation
Alternatives&#8217; <a href="">CrashStat</a>, which only runs up 2009.</p>
<title type="html"><![CDATA[The Long Reach of Industry Research]]></title>
<link href=""/>
<content type="html"><![CDATA[<p>PIPA is, for the moment, dead in the water. Reid is not going to vote
on cloture. His <a href="">statement</a> reads, in part:</p>
<blockquote><p>&hellip;Counterfeiting and piracy cost the American economy billions of
dollars and thousands of jobs each year, with the movie industry
alone supporting over 2.2 million jobs&hellip;</p></blockquote>
<p>The part about &ldquo;supporting 2.2 million jobs&rdquo; looked familiar. Last
night, I checked out <a href=""></a> when I heard that anonymous had
knocked it down. It was already back up, and displayed prominently on
the main page was the same &ldquo;supporting 2.2 million jobs&rdquo; claim. In
<a href="">more detail</a>:</p>
<blockquote><p>We support a national community of 2.2 million workers — costume
designers to make-up artists, stuntmen to set builders, writers to
actors, accountants to dry cleaners — employed in all 50 states of
our union.</p></blockquote>
<p>Their <a href="">PDF fact sheet</a> breaks things down a little more:</p>
<blockquote><p>The industry is a major private sector employer, supporting 2.2 million jobs, and nearly $137 billion in
total wages in 2009:</p>
<li><p>Direct industry jobs generated $40.5 billion in wages, and an average salary 26% higher than the
national average:</p>
<li>There were nearly 272,000 jobs in the core business of producing, marketing,
manufacturing, and distributing motion pictures and television shows. These are high
quality jobs, with an average salary of nearly $79,000, 73% higher than the average salary
<li>Additionally, there were over 430,000 jobs in related businesses that distribute motion
pictures and television shows to consumers.</li>
<li>Industry also supports indirect jobs and wages in thousands of companies with which it does
business, such as caterers, dry cleaners, florists, hardware and lumber suppliers, and jobs in other
companies doing business with consumers, such as video retailers, theme parks and tourist
<p>OK, so &ldquo;support&rdquo; means &ldquo;do business with.&rdquo; Fair enough, and by that
metric I alone supported hundreds of grocers, bartenders, video
retailers, laundromats, chefs, musicians, etc. in 2009. It is a far
cry from the enumerated make-up artists, stuntmen, set builders,
writers, and actors. Reid&rsquo;s phrase, &ldquo;the movie industry alone,&rdquo; is
similarly deceptive.</p>
<p>Most of those 2.2 million workers are the dry cleaners. And I think
that very few of them would lose their jobs if the movie industry
simply ceased to exist, which is not what is at stake.</p>
<p>Ironically, I see no reason why those &ldquo;430,000 jobs in related
businesses&rdquo; couldn&rsquo;t be jobs with companies involved in the piracy of
movies, or, more reasonably, companies who would be fucked by the
consequences of legislation like PIPA.</p>
<p>What is most irksome is that Reid is happy to parrot a misleading
factoid <em>directly</em> from the MPAA&rsquo;s website, when the government spends
several billion dollars a year on the <a href="">Census Bureau</a>. Which does
its own <a href="">Economic Census</a> every five years, and keeps other tabs annually.</p>
<p>So what information cost several billions of dollars for Reid to blow
off in favor of his donors? I looked up employment by
<a href="">NAICS category</a>, which is 512110 for &ldquo;Motion Picture and Video
<p>There were 146,539 &ldquo;Paid employees for pay period including March 12
(number)&rdquo; in 2009. <a href=";prodType=table">Check it out</a>.</p>