Many of the corrections that appear in the press are notable thanks to the significance or amusing nature of the mistake, or because they speak to a larger truth about journalism. But often the most important corrections are the ones that are never published.
This is because the best research we have about correction rates suggests that the vast majority of errors made in American newspapers—roughly 98 percent—are never corrected. That reality is now affecting journalist and author Scott Rosenberg on almost a daily basis.
I interviewed Rosenberg back in September about MediaBugs, a project he recently launched thanks to a Knight News Challenge grant. (I’m an unpaid advisor to MediaBugs, though I have nothing to do with its day-to-day operation.) The point of MediaBugs is to create a place where people in the San Francisco Bay Area can report errors found in local media. Then Rosenberg and associate director Mark Follman look at the bug (error) report and, if it seems valid, follow up with the offending outlet. They’re trying to increase public participation in the correction process, and, along the way, help discover and correct mistakes that otherwise might have fallen through the cracks.
This is important work. We need more corrections, rather than fewer. (“Fewer Errors, More Corrections” would be my slogan if this were an election. I’m convinced it would resonate with the Tea Party movement, thus ushering me into office.) One of the biggest barriers to correction is that people often have no idea how to go about getting one. Media outlets can unwittingly make it frustrating, confusing and ultimately pointless to report an error. That leaves members of the public feeling frustrated, and I think it also reduces their level of trust in the press.
Rosenberg experienced this dynamic first hand when he recently tried to report an error to The Wall Street Journal. Here’s an excerpt from a blog post that recounted his long and largely fruitless correction hunt:
I looked for some link on the Journal site for “corrections” or “report an error.” No such link exists on the Journal home page, nor did searching voluminous “Help” and “Customer Service” pages turn up anything. The “Contact us” page offers three general email addresses for feedback, labeled as follows:
Send a comment/inquiry about an article or feature in The Wall Street Journal to:wsjcontact@dowjones.com.
React to something you’ve read on WSJ.com at: newseditors@wsj.com.
Offer a comment/suggestion about features and content on WSJ.com at: feedback@wsj.com.
I challenge anyone who is not a part of the WSJ organization to interpret which of these three lines of inquiry would be an appropriate choice to report an error …
Aside from sending e-mails, he also posted a comment on the offending article and left a phone message. Nothing happened. It was only after his blog post made Romenesko and was retweeted on Twitter that the Journal leapt into action and fixed the error.
“A news organization should not expect the public to understand its internal structure [in order to request a correction],” Rosenberg told me.
He says we need a very clear standard for reporting an error, something that every news organization uses and that readers can learn to recognize.
“Part of the solution is for every site to have a button that says ‘Report an error,’ ” he says. “I’m debating whether one of the central focuses of MediaBugs should be a campaign to make it a Web standard so that every site has this button. ‘Share this’ became standard two or three years ago …”
Another example would be RSS. One way that it achieved mass adoption was by becoming standardized on major blogging platforms and then adopted by major Web sites. At the same time, the adoption of the RSS icon created a universal visual cue that helped drive understanding among Internet users.
This is rich: Perhaps two weeks ago I posted a correction here, at the CJR, on an article "How Big Is Big? When numbers are meaningless" [By Merrill Perlman] about how numbers in news articles don't mean much unless interpreted in terms that readers know. The error was that a square mile is "about" 640 acres, when, in fact, it is exactly 640 acres.
Then, I went on in my comment to make an error of my own. (Serves me right, I suppose). I was quite wrong about the size of a US football field. This was helpfully corrected by another reader.
However, the original error stands in the article, underscoring by example the need for an effective error-correcting mechanism.
#1 Posted by Nicholas Spies, CJR on Sat 12 Jun 2010 at 02:15 PM
Didn't see your initial comment, Nicholas. Sorry. Will correct Merrill's article now. If this sort of thing happens again, try e-mailing editors@cjr.org. We're usually pretty responsive.
#2 Posted by Justin Peters, CJR on Sat 12 Jun 2010 at 06:08 PM
Lesson learned for me: Don't trust Apple's Calculator for area conversion. It says 1 square mile is 639.999998790047318 acres. When converting the other way, it says 640 acres is 1.00000000118506 square miles. I don't know if that's true of all versions of Calculator. Mine is 4.5.2 (97), the version included with Snow Leopard.
I also did a quick sampling of online conversion utilities. Of the first 10 Google results for "area conversion," three of the online utilities got the answer wrong (including the top-ranked conversion utility) when converting from square miles to acres or from acres to square miles or both.
Thanks for bringing this up, Nicholas.
#3 Posted by David Nagel, CJR on Mon 14 Jun 2010 at 09:41 PM
I recommend a funky but extremely useful application called Frink (found at http://futureboy.us ) whose author, Alan Eliasen, recognized the lack of physical units in programming languages and wrote Frink to redress this. Frink has many, many systems of units, not just physical sizes, weights, etc but also relative values of currencies, varying values for some over time (with automatic lookup of tables of such info on the web) and much more. It's a free Java app, thus cross-platform.
Alas, even Frink gives the wrong answer for: 1 mile^2 -> acres
249999000001/390625000 (exactly 639.99744000256) , and I have no idea what the fraction it bases its wrong conclusion on represents. [I haven't brought this to the attention of the author, yet.]
This shows that conversions must be tempered with a bit of common sense. When a result is very nearly a round number, it is worth taking another route to see if your first result is not simply a round-off error.
#4 Posted by Nicholas Spies, CJR on Wed 16 Jun 2010 at 06:43 PM
Actually, Frink is exactly right, and the original statement that a square mile is "about" 640 acres is correct. Nicholas wrote to me and told me that Frink was wrong. Here's my response to him, which hopefully sheds some light on the issue of these historical units of measure.
Actually, Frink is exactly right, and it's important to know why it's right.
Other converters get this wrong due to sloppy research.
The short answer is that the furlong (and league,
acre, rod, statute mile, township, chain, etc.) are all defined in terms
of the *survey* foot. This is the definition of the foot used in the
U.S. up until July 1, 1959, when its value was slightly changed (by 2
parts in a million, as you can see) to the current value of the
"international" foot as exactly 0.3048 meters. Previously, it was
exactly 1200/3937 meters, which can't be represented as a finite decimal
number.
Surveyors still use the old definitions, and the correct definition (by International standards body and by law) of the acre and other units need to be based on the old "survey foot" definition.
Any time Frink gives you a result that is a multiple of the fraction
499999/500000 when working with lengths, you'll know the old survey foot was involved.
If you use, in Frink:
1 statutemile^2 -> acres
then you'll get the result you want. The "statute mile" is the *old*
mile using the outdated pre-1959 definition. The modern mile is just
called the "mile" or sometimes the "international mile" if someone's
making sure which mile is to be understood. Frink also uses the term
"surveyfoot" if you want to use the old, outdated definition of the
foot, say, when converting acres to square feet.
Surveyors still use the old definition, and if you want to be
correct, (and I do,) you need to use the right base definitions for all
the derived definitions. This is a common question, and people are
often suprised to hear the history (unless they work in a geographic
field, and then they run into it 50 times a day.) Ask a surveyor about
the fun this creates. Every one of them will have some stories.
Also see:
http://www.sizes.com/units/foot.htm
or NIST Special Publication 811.
http://physics.nist.gov/Pubs/SP811/contents.html
(This uses underlining to show the units that were defined in terms
of the survey foot. Be warned that later versions have typos and
mistakes due to a change in the software they used to produce it. I
reported these to them and they've acknowledged them as errors to be
fixed. I haven't checked those publications in the last couple of years
to see if the changes made it in.)
I really do scrupulously research these things (I've probably
researched every major measure in Frink's data file at least 5 times,
usually against the most authoritative references I can
find--international standards bodies and the like) so if you see
something unexpected, odds are there's a reason for it. I'm always
surprised when I dig in and research some of these units. If you see a
conversion that seems surprising, I'll bet all the money in my wallet
that it's right, and there's a reason for it.
As a corollary, I've probably found and reported as many typos in
NIST publications as anybody out there. :)
Thanks for the question, though! I get this a lot. I've seen
various web pages where someone runs a similar calculation through and
blames roundoff or something when they should blame surveyors. :)
> > I was curious what the fractions, in this case 249999000001/390625000)
> >
#5 Posted by Alan Eliasen, CJR on Wed 16 Jun 2010 at 08:00 PM
I just verified that the current version of NIST Special Publication 811 (linked above) *has* been properly typeset to indicate the correct definition of acre, survey foot, etc., which matches Frink's definition, of course. See section B.6. It also discusses the history of the "statute mile" and "survey foot", so you can prove to yourself that Frink's exact rational number is, indeed, the correct answer.
It's often, no, *always* necessary to do your research on units of measure with official, authoritative documents from your national or international standards body, and not believe the vast number of confused and poorly-researched sources out there. (But I'll repeat that for a couple years, even NIST had it wrong when they changed typesetting programs!)
As I said before, I'll bet the authority of Frink's research against any other unit conversion utility out there any day. If you want to be right, use Frink. I've noticed that other unit converters like the Google Calculator and Wolfram Alpha are very poorly researched, and don't understand the difference between many units like these. You can prove I'm right by following the definitions set forth by international standards bureaus.
If you want to see how a unit is defined, I highly recommend peeking into Frink's standard data file:
http://futureboy.us/frinkdata/units.txt
Nicholas had a good point when he stated that if a conversion came out close to a whole number, you should think about more deeply. However, as you can see, common sense has little or nothing to do with institutional inertia of surveyors, so you have to do a bit deeper research to get the *right* answer. :)
#6 Posted by Alan Eliasen, CJR on Wed 16 Jun 2010 at 10:30 PM
FYI,
Sending to editors@cjr.org results in:
Sorry, but this mailing list does not accept submissions by email.
Your message was not accepted for distribution.
#7 Posted by Nicholas Spies, CJR on Thu 17 Jun 2010 at 05:28 AM
The editors@cjr.org address should be working now.
#8 Posted by Justin Peters, CJR on Thu 17 Jun 2010 at 10:29 AM