I do not work for the NYT. I did, once upon a time, work for a news organization that was considered to be innovative with respect to using technology to further the presentation of journalism.
One of the things I worked on was a generic framework for "composite" stories -- that is, a collection of not just text and photos, but integrated presentations of text, audio/video, photos and data visualizations, with customizable templating to lay it all out and organize for the web.
Time and link rot have not been kind to the things that were done with that. For example, the data bits of this:
no longer function, and it certainly shows its age, design-wise. But that was where we were in early 2007, technology-wise, and it was built with a hacked-together-in-a-week proof of concept of the underlying framework.
What we had at the time shipped with some standard out-of-the-box templates for presenting a story, and for one-off stories we could also task someone to work on creating new ones which could be slotted in for use. When new templating work wasn't required, they could put together a composite story presentation in a matter of hours, once they had all the materials they wanted.
The next evolution of that, which I was pushing for toward the end of my tenure there but never got to complete, would have been integrated data processing and visualization tools, allowing journalists to toss various sources (my rough draft worked with spreadsheets, since those were a pretty common source) into a denormalized data store, scrub a bit, and generate useful presentations from them.
While I can't be sure, I strongly suspect that the NYT has done something similar, developing the underlying framework for tying together a bunch of content in an attractive way, and probably building some tools to simplify and speed up the generation of the data presentations. What they're doing here, and have done a couple other times, seems well within reach for someone who knows what they're doing and has access to six years' worth of advances in technology.
(anecdotally, I've heard that the NYT special projects are done in Rails; ours were done in Django since it was originally our in-house web framework, but in either case it's a strong argument for the productivity gains of those web frameworks)
Edit: FWIW, here are slides from a lightning talk I gave at PyCon in 2008, about a data-driven project and the timeline involved in it:
One of the things I worked on was a generic framework for "composite" stories -- that is, a collection of not just text and photos, but integrated presentations of text, audio/video, photos and data visualizations, with customizable templating to lay it all out and organize for the web.
Time and link rot have not been kind to the things that were done with that. For example, the data bits of this:
http://www2.ljworld.com/news/mining/
no longer function, and it certainly shows its age, design-wise. But that was where we were in early 2007, technology-wise, and it was built with a hacked-together-in-a-week proof of concept of the underlying framework.
What we had at the time shipped with some standard out-of-the-box templates for presenting a story, and for one-off stories we could also task someone to work on creating new ones which could be slotted in for use. When new templating work wasn't required, they could put together a composite story presentation in a matter of hours, once they had all the materials they wanted.
The next evolution of that, which I was pushing for toward the end of my tenure there but never got to complete, would have been integrated data processing and visualization tools, allowing journalists to toss various sources (my rough draft worked with spreadsheets, since those were a pretty common source) into a denormalized data store, scrub a bit, and generate useful presentations from them.
While I can't be sure, I strongly suspect that the NYT has done something similar, developing the underlying framework for tying together a bunch of content in an attractive way, and probably building some tools to simplify and speed up the generation of the data presentations. What they're doing here, and have done a couple other times, seems well within reach for someone who knows what they're doing and has access to six years' worth of advances in technology.
(anecdotally, I've heard that the NYT special projects are done in Rails; ours were done in Django since it was originally our in-house web framework, but in either case it's a strong argument for the productivity gains of those web frameworks)
Edit: FWIW, here are slides from a lightning talk I gave at PyCon in 2008, about a data-driven project and the timeline involved in it:
http://media.b-list.org/presentations/2008/pycon/lightning.p...
Sadly, the story itself is no longer functioning.
Here's one that's actually miraculously still online, from the fall/winter a few years back when everyone was worried about swine flu:
http://www2.ljworld.com/data/flu/
The core of that was put together very quickly, and then it was about an hour's work each week to update with the latest data coming in.