# Measure for Measure

My daughter is moving to Colorado, and she and her older brother recently took a four-day road trip from his house in Pennsylvania to her new apartment. They both posted their photos on Shutterfly; many of the shots are essentially the same.

The big difference between their photo albums—and it’s makes a huge difference—is that he took a few extra minutes to add captions. So from his pictures I know that the bronze statue of Abe Lincoln they both stood next to is in Vandalia, which was once the capital of Illinois; that the baseball game they went to was in Kansas City; that the giant cross by the highway is in Effingham, Illinois—which is enough information for a Google search that says that the cross is 198 feet tall and was built by the Cross Foundation.

Of course, location information, such as Vandalia, Ill., and Kansas City, Mo., can already be included in a photograph’s metadata, if the camera has GPS. In my ideal universe, all cameras would, and they’d even have little keyboards so you could add a caption right when you take a picture. Photo metadata is phenomenally useful, and, in a world of photo clouds like Shutterfly and Flickr, it’s getting ever more so.

You know what else needs metadata? Engineering numbers. That’s the premise behind Allen Razdow’s new start-up, True Engineering Technology.

Razdow was a co-founder of MathSoft, the company behind Mathcad. In a way reminiscent of Stephen Wolfram’s idea that the Mathematica universe would benefit from databases that could be queried (thus, Wolfram Alpha), Razdow decided that the software on an engineer’s desktop, like Mathcad—but also Microsoft Office—needed metadata for the numbers that move from one program to another. If I had to choose one of these ideas as a winner, it would be Razdow’s.

The idea apparently came slowly to Razdow, who wrote MathCAD back in the 1980s (for MS-DOS!), and with good reason—it requires you to think of numbers in a paradoxical way. While the common conception of computers is that they turn everything into numbers, Razdow’s insight is that the reality is just the reverse.

We take an engineering number, maybe it’s the hydrogen permeability of palladium, or the specific gravity of the railroad ties you just shipped to a customer, and put it into a report that strips it of almost all of its meaning—what reference book the number came from, or when and where it was measured and by whom, the tolerances, and so forth. We take numbers that are ripe with engineering meaning and mathematical context and turn them into flat text. Often—and paradoxically this happens particularly with those bastions of number-crunching, spreadsheets—you don’t even directly know the unit of measurement, because that’s contained in a column heading or a footnote or some other surrounding text.

Consider all the numbers that get used and reused for years, within your company and outside of it. Imagine you’ve worked out a more precise measurement of the hydrogen permeability of the particular palladium alloy to be used in an upcoming product. Or auditors from the Nuclear Regulatory Commission or the Food and Drug Administration have arrived to examine a new report that contains 70 different key engineering numbers that need to be checked, 50 of which were taken from a report that was vetted last year.

Razdow has in mind a plug-in that would encourage you to create metadata for important numbers and would let a number retain that metadata when it’s cut and pasted from one application to another, whether it’s MathCAD to a Word document, or to a PDF, or vice versa. You can hover over someone’s number and see some of that metadata, or, if they make it public, you can click on it and get all of it from a Web page devoted to that number on a public site that Razdow’s company will maintain. True Engineering Technology will make its money by selling a server appliance that will host and manage engineering numbers within an enterprise.

It’s a clever and much-needed idea. Like a lot of other Web 2.0 notions these days—think RSS, for example—it will need widespread adoption by users—in this case engineers—and the software applications that they use. Here’s hoping that happens.