Boston.com’s Homegrown Revolution
by Mark Toner
Wander the narrow streets of Boston’s North End, and The Old North Church comes into view around a blind corner.
 |
| An online-grocery van stops to make a delivery near Boston’s historic Old North Church. |
Beneath the steeple where Paul Revere once hung two lanterns to mark the beginning of the Revolutionary War idles a foot soldier from a more contemporary battleground: a homeruns.com online-grocery delivery van.
Make no mistake, the historic gateway to the Northeast has emerged as a New Economy powerhouse. Saying The Boston Globe’s 1995 launch of Boston.com was heard around the world may strain journalistic license, but someone was listening. Namely Bill Gates, who touted the site’s unique melding of news, arts-and-entertainment listings and other information as the model for Microsoft’s Sidewalk online city-guide network. Sidewalk has since rolled out and rolled itself up, but the city-site approach now is commonplace.
And the revolution continues. A stone’s throw from the site of the historic Boston Tea Party, Boston.com’s technical staffers are moving away from what could be considered the standard approach to managing a complex Internet operation. Instead, they’re embracing the open-source movement that offers free–and freely modifiable–software in the place of proprietary solutions.
 |
| An inflatable penguin in Boston.com’s technical department hints at its Linux leanings. |
"We really use open-source software," says software-development director Stephen Reppucci. And following the ethos of the open-source movement, Boston.com also freely shares some of the tools it develops. The homegrown code developed to serve up the site’s MP3 digital-music service, for instance, now is in place at four other papers.
"We like to walk the walk," he adds.
TOO HIP TO BE SQUARE
Tie-clad visitors feel out of place stepping into the lobby of 320 Congress St., a renovated warehouse in the Fort Point district where Boston.com shares space with several other technology companies. Jeans, T-shirts and ponytails are everywhere, and staffers volley good-natured barbs back and forth in the elevator.
Marketing Director Stephanie Shore greets us in the second-floor reception area, where a neon Boston.com sign glows above a Boston Globe newsrack–"our link to the Globe," she explains.
A quick stroll reveals many of the details that we’ve all come to expect at a dot-com workplace: exposed brick walls and timbered ceilings, a break room with the obligatory couches and equally obligatory Sony PlayStation, and restrooms a newly hired office administrator keeps stocked with hand lotion, perfume and hairspray.
All but a handful of Boston.com’s 80-odd employees work at the 3-year-old Congress Street office. Staffers previously worked downtown in more traditional office space perched directly above the Boston Marathon’s finish line–"helpful when we were covering it," says Content Manager Michael Manning.
Elevators are now the only barriers between departments, which flow from area to area across Boston.com’s three floors. Much like a print newsroom, the site’s first-story content area could be mistaken for the front office; staffers obligingly point visitors upstairs to reception. A staff of 10 produces Boston.com’s arts-and-entertainment coverage, MP3 site–home to downloadable music from local bands–and digitalMASS, featuring online-only content about the region’s bustling technology sector.
 |
| A Boston Globe newsrack reminds Boston.com visitors of its print connection. The site includes Globe content but has its own distinct identity. |
"We cast a wide net," says Manning. "We’re reaching a lot of different audiences."
The second floor is home to marketing and staffers who produce commercial World Wide Web sites and online-advertising campaigns. The group’s recent TechNews Best Practices Award, bestowed for its homegrown approach to managing traffic statistics (TechNews, January/February 2000, p. 7), hangs in the reception area.
A quick elevator ride brings us to ad sales, finance and senior-management offices. A totem-like "sales god," quite possibly wearing the only other tie in the building, sits on a table near space being cleared for an expanded online-classified telemarketing staff. Turning a corner into the technology group’s area, a stuffed animal flies from one cubicle to the next.
Of course, no Internet operation would be complete without odd playthings, and Boston.com is no exception. Life-sized cutouts of Xena: Warrior Princess, inflatable neon furniture, double-parked bicycles and gaudy hubcaps inhabit various corners, as does an Eiffel Tower-shaped lamp that could charitably be described as "quaint."
"That used to be mine," Shore admits with a shrug.
But one prop–a four-foot tall, blow-up penguin standing guard over the warren of cubicles in Boston.com’s technical area–hints at more than sheer kitsch. Named Walter, the penguin is, as all good tech diehards know, the quasi-official mascot of the Linux movement that’s changing the way Boston.com is served up to the public.
PENGUIN POWER
Boston.com’s gradual shift away from what Reppucci calls the "cookie-cutter" approach to site infrastructure was, in truth, more evolution than revolution.
Preparing for its Halloween 1995 launch, Boston.com contracted with what was then BBN, which dictated out-of-the-box solutions, including Netscape server software and hardware from Sun Microsystems Inc. As part of a managed-solutions agreement, BBN maintained the servers off-site, an arrangement that lasted about four years with varying degrees of frustration.
"To keep a site of this volume running requires a lot of tinkering, and when you’re dealing with a managed-solution [arrangement], they want to do the tinkering," Reppucci explains.
 |
| About 10 Boston.com staffers produce content ranging from arts-and-entertainment listings to regional technology news for the site. They use Allaire Homesite to design Web pages. |
Boston.com staffers gradually wrestled control of server administration, if not hardware, from BBN. BBN was then bought out by GTE, and Reppucci, originally hired as a temporary contractor for Boston.com’s launch, became the site’s chief techie. By then, "we wanted to control everything," he says. "We wanted someone to drop a big, fat Internet pipe into our cage and let us do the rest."
About a year and a half ago, the site moved to Boston-based HarvardNet, which provides what hosting salespeople call "paint, pipe and power"–an Internet connection and electricity, in other words.
Today, about two dozen servers secured behind a chain-link fence at HarvardNet power Boston.com. Exact numbers change almost weekly, but during our visit 10 Linux boxes running open-source Apache server software sit next to 12 Solaris systems, two network file servers and an Arrowpoint switch maintained by Cisco.
Backers of Linux, a free, Unix-like operating system maintained by a fervent community of open-source programmers, contend that it gives less expensive hardware the same firepower as top-of-the-line workstations (TechNews, January/February 1999, p. 25). "For graphics servers, I could have bought a Sun server," Reppucci explains. "For the same price, I could buy four or five Linux servers that do three times the work."
What’s more, Apache supports mod_perl, Boston.com’s programming language of choice. Perl powers the brunt of the site’s custom-built widgets–tools ranging from online-classified software (a temporary fix that, at least for now, has become permanent) to chat and messaging features.
Consider the site’s arts-and-entertainment calendar. "Vendors were throwing around $25,000 in annual licensing fees," Reppucci says. "One guy built it in three weeks. We did it quickly, and we have something we can extend as we see fit."
Not all of the software used to serve up Boston.com is Linux compatible, so the Solaris boxes continue churning away. But Reppucci claims that the care and feeding of two separate operating systems by his staff of 10 isn’t onerous, as techies tend to be voracious hobbyists. "Just about every system administrator I know who runs Sun at work has a Linux box at home," he says.
 |
| Boston.com’s offices share space in this Fort Point warehouse with several other technology companies, including another dot-com. |
And that’s another open-source benefit, Reppucci argues, noting that the requisite tweaking "appeals to really good programmers. [Recruiting] has been very easy for me."
Which brings us back to Walter, still keeping watch from the far wall. Not only a mascot, he’s also the e-mail contact for Boston.com’s own help-wanted ads (walter@boston.com). Why? To thwart a common headhunter trick–luring away the very person who’s placing the ad, a constant threat in Boston’s hypercompetitive job market.
THE MISSING PIECE
Throughout our visit to the Congress Street offices, people keep referring to "the pod," where the Globe’s print stories are made Web-friendly for Boston.com. It’s at the far end of a 30-minute cab ride through rush-hour traffic snarled by Boston’s ongoing "big dig," an ambitious attempt to stuff downtown throughways into tunnels.
Nestled between photography and sports in the Globe’s sprawling print newsroom, the pod is just that–a semicircular workstation crammed with about a half-dozen computers "and even more people at times," says Eric Bauer, Boston.com’s executive news editor. He pauses to show a visitor the Globe’s long display of Pulitzer Prizes, something not many dot-coms have on their walls.
By design as well as by geography, Boston.com remains distinct from the print Globe, and the pod serves as the bridge linking the two. Its staffers run homegrown scripts that coax copy from the Globe’s Atex editorial front end into Microsoft Word, where macros create data fields to simplify HTML coding. That’s handled by still more homegrown scripting, which also generates index pages of all the day’s stories and different versions of each one (for high- and low-bandwidth users, for printing, and for a service used by the blind).
Sounds pretty automated, right? Well, it isn’t.
"We still have a long way to go," Bauer admits. "The old Atex system does not give up its information easily. Like most older Web sites, we really had nothing to start with. Every tool you made yourself or adapted to fix something."
It takes two full-time staffers all night to coax the entire print edition onto the Web by 5 a.m., after which they go back to add photos and links, and then continue tweaking. Day-side staffers work with print editors and the wire to update the site throughout the day.
"It’s easy to think of it as spit and baling wire, but the reality is that it’s lasted for five years," Bauer says.
 |
| Boston.com’s "pod" in The Boston Globe newsroom posts all print content to the site. It also provides breaking-news updates throughout the day. |
Intermediate fixes are now on the way. A homegrown system that pulls stories from the Globe’s "more reliable" print-library feed and deals somewhat in XML is in the works. Content producers in the Globe newsroom and at Boston.com now handle site design with Netwon, Mass.-based Allaire Corp.’s HomeSite 4.0, off-the-shelf Web-authoring software that generates HTML but lacks robust site-management tools. So staffers still manage Boston.com pages the same way we all keep track of things on our PCs–by manually juggling files and folders. Completed Web pages are sent to a staging server, where they sit until the folks at the pod decide to update the live site.
Sounds like the perfect scenario for a content-management system, but after researching the experiences of other large newspaper Web sites, Boston.com discovered that "lots of people aren’t happy with what’s out there," says Manning.
Indeed, the industry consensus is that such systems aren’t quite ready for prime time. Here’s Bauer’s theory why: Whether big or small, newspapers’ print work flows remain similar, while Web sites are "vastly different in how they do things." Indeed, at Bauer’s former job at Community Newspaper Co.’s town online.com, staffers processed a jaw-popping 10,000 stories a week.
But the tide may be turning. "There are starting to be some good CMSes for the Web that can handle the quantities newspaper sites deal with," Bauer says. Indeed, Boston.com parent New York Times Digital is now experimenting with a next-generation content-management system at its flagship site. If successful, that solution could find its way to other New York Times Co. properties.
"Some things it doesn’t make sense to do yourself," Bauer says.
In the meantime, Boston.com’s staffers work hard to maintain the revolutionary zeal.
"We have a lot of the startup feel here," says Reppucci. Some
sites are "a newspaper that has a Web presence. We’re a Web
site with a newspaper. Our attitude is more dot-com than newspaper."
Toner is TechNews editor. E-mail, tonem@naa.org.