Why the IEEE working group attempting to craft the 802.20 Mobile Broadband Wireless Access standard became a bloody train wreck.
"There were a lot of accusations of overt political behavior and misdealings," says Craig Mathias, an IEEE member and a principal with Farpoint Group, a consulting and systems-integration firm. "In this case, someone needed to step in and get everybody back on track."
A revamped working group convened in Dallas in November with a new chairman, vacancies in the other leadership posts and most of its prior decisions invalidated.
Yet the problems and ultimate reboot of the 802.20 committee was no isolated incident. It was just the most recent and dramatic example of an IEEE standards-setting process bogged down by competing commercial interests, lack of vendor-neutral participants and at times ineffective committee leadership, critics say.
Waiting for MIMO
Last year alone, in addition to the 802.20 suspension, the three-year effort to settle on an IEEE standard for Ultra WideBand (UWB) deadlocked and disbanded.
The IEEE working group crafting the eagerly awaited 802.11n standard for 100Mbps-plus wireless LANs was supposed to agree on a standard in 2006. Because of vendor infighting, the projected date for final approval has been pushed back to 2008.
There's so much market pressure for 802.11n products that vendors are forging ahead and shipping pre-N products today. This creates a dilemma for enterprise customers, who have to decide whether to wait for the standard or to deploy prestandard products that might require upgrades a year from now.
One major problem is that IEEE working groups, which once typically counted attendees by the dozen, now frequently attract hundreds of voting members, greatly increasing the chances for gridlock.
"The last committee I chaired had 15 people on it," says Joe Skorupa, a Gartner analyst and IEEE member who served more than eight years on the committee that developed the original Ethernet standards and chaired the 802.3 Network Management Task Force for four years. "Now you see committees where hundreds of people show up. One of the wireless committees had close to 1,000. And it becomes very hard to get business done with that many people."
The sheer size of the working groups can slow the process of drafting, revising and ratifying a final standard, which requires a 75% majority of the member votes. Anyone can attend the plenary and interim meetings of the working groups by paying the registration fee -- typically no more than a few hundred dollars -- and participants earn voting rights based on their meeting attendance.
In the case of 802.11n, the initial draft standard failed to win 75% approval, last May.
All-volunteer army
All participants, including the officers of the IEEE 802 LAN/MAN Standards Committee, commonly known as Project 802, and the executives of the IEEE Standards Board, are volunteers who typically have day jobs as engineers. Some donate their time and effort for the greater good of the industry, but most are supported by their employers who generally underwrite the cost of attending meetings held in locations around the globe.
The 802.11n task force, for example, has at times had more than 500 voting members, and its sluggish progress spawned the creation in late 2005 of the Enhanced Wireless Consortium by Intel, Cisco, Broadcom and 24 other companies to help break the deadlock created by competing factions.
Another time-consuming wrinkle in the process is that the IEEE requires a formal response to the comments that voters submit when they vote against a specific draft or proposal, says Timothy Simcoe, an assistant professor at the Rotman School of Management at the University of Toronto.
"When there are a lot of 'no' votes, that can generate an awful lot of these kind of replies," he says. "So if there are many factions or many members of a faction who issue objections to a particular proposal, that can really create a situation where you get mired down."
In the case of 802.11n, there were 12,000 comments when the initial draft was defeated.
Qualcomm before the storm
Such factiousness among competing vendors -- to which many observers attribute the mushrooming size of the task groups, as companies attempt to stack the deck in their favor -- can make the required consensus harder to reach. Such was the well-publicized case of the 802.20 group, which appeared to be caught in a crossfire between a faction led by Intel and Motorola and an increasing number of voters aligned with Qualcomm and Kyocera.
Qualcomm initially attempted to derail the 802.20 standard, originally based on Flash Orthogonal Frequency Division Multiplexing technology from Flarion Technologies, say observers and working group participants.
Yet even after its $600 million acquisition of Flarion last August, Qualcomm remained at the center of the storm when the 802.20 working group's then-chairman, consultant Jerry Upton, confirmed allegations that he had been a paid consultant to Qualcomm.
"It's no longer developing standards for the technical good," says Mark Fabbi, a Gartner analyst. "There is this political and legal minefield they're having to maneuver through. And certainly they have to think about [the politics] more than they did 15, 20 years ago."
Standard operating procedure
"It's a classic case of too many cooks spoiling the broth," says Frank Dzubeck, a consultant who has followed the network standards-setting process since 1980, before the IEEE Project 802 was launched. Veterans of the standards battles maintain that contentious politics is nothing new, yet most acknowledge that things are getting worse.
"Vendor manipulation of standards processes, including the IEEE, and especially Project 802, is nothing new," insists Bob Metcalfe, the inventor of Ethernet and founder of 3Com, who was instrumental in forging the IEEE's first LAN standards in the late 1970s and early 1980s. "The trick is not to deny and decry vendor motives, but to disclose and align them."
Metcalfe, in an e-mail exchange, recalls the wrangling over the original IEEE 802 standard among IBM, the Ethernet proponents -- Intel, Xerox, 3Com and the former Digital Equipment -- and a group backed by General Motors, each pushing their favored technology.
The sparring became so fierce that his opponents succeeded in getting Metcalfe disqualified from voting, though at the time he was arguably one of the world's most accomplished engineers in the field of computer networking. The two-year stalemate led to a split in the effort into three standards -- 802.3 for Ethernet, 802.4 for GM's Token Bus and 802.5 for the IBM Token Ring.
"This is how 802 got its dots," says Metcalfe, now a general partner in the Boston office of venture capital firm Polaris Ventures.
Steve Mills, chairman of the IEEE Standards Board, agrees, noting that the recent troubles with 802.20 are just the latest in a series of occasional skirmishes among competing vendors that have always been part of the standards-setting process.
"I wouldn't call 802.20 unusual, but I also wouldn't describe it as being anything different than [the battles] back in the early '80s," Mills says. "Occasionally, throughout the course of standards development within 802, there have been contentious situations, 802 Ethernet-Token Ring being an example."
Although he dismisses complaints that the process has grown more contentious in recent years, he agrees with those who attribute last year's trouble to the increasing stakes in multibillion-dollar network markets.
"The higher the economic stakes, the more vested people become, and that makes the goal of consensus in a standard a little tougher to achieve," he says. "But there are many, many examples of standards that have similar characteristics where they are able to move forward and reach a consensus and conclusion."
Of the 50 or so projects that are underway within 802, only two -- 802.20 and the disbanded 802.15a UWB effort -- gained his attention last year, Mills says.
Nonaction items
Yet issues with the Project 802 standards-setting process gained Mills' attention even before last year's difficulties, when its Executive Committee held a brainstorming session on process improvements during its November 2005 five-day plenary meeting in Vancouver.
Although the session was sparsely attended, according to published meeting notes, Mills, Project 802 Chairman Paul Nikolich and Vice Chairman Matthew Sherman participated in the discussion, which addressed such issues as the unwieldy size of working groups, the obstructionist practices of some participants and how the slow pace of the process can contribute to factions becoming entrenched in opposing positions.
"The size of the group is a problem. We have overgrown our process," according to the unsigned notes from the meeting, distributed to the Executive Committee by Mills.
Among the core problems identified in the discussion was that the "down select" process -- in which working groups eliminate competing proposals one at a time until they are down to two -- is too slow, giving competing factions within a group more time to dig in their heels. "We are set up to 'pick a winner,'" the notes read. "It is a win-[lose] proposition."
Another issue identified at the brainstorming meeting -- and also cited by participants and observers as an obstacle in the IEEE standards process -- is the uneven leadership abilities of working group chairs, whose political skills can make the difference between success and gridlock.
"There is a large disparity between chairs in their ability to facilitate consensus," according to the notes of the November 2005 meeting. The session, attended primarily by Project 802 leadership and IEEE Standards Association staff, produced a list of suggested improvements to the process, only a few of which are being implemented. Listed under "Direct Suggestion" were:
* Eliminating the distinction between interim and plenary meetings, by extending decision-making authority to the interim sessions.
* Training and coaching provided for the chairs that teaches how to run a task or working group and to facilitate consensus.
* Providing tools for functions such as electronic recording of meeting attendance, document control, electronic balloting, calendaring, event tracking and conducting virtual meetings.
* Simplified attendance rules that would count attendees as soon as they picked up their meeting badge.
* Formalizing a revenue-based voting system rather than the current one-vote-per-attendee system. (The notes acknowledge that there is de facto revenue-based voting in that larger companies send larger groups of attendees than do smaller companies.)
The lone action item in the meeting notes was for Mills, Nikolich and Sherman to summarize the issues to the Executive Committee, which would then try to prioritize the list and determine subsequent steps. Through a spokeswomen IEEE executives declined to comment on which suggestions are being implemented, calling it "a true brainstorming meeting, which was not intended to produce any binding decisions but to try to put as many issues on the table as possible."
Progress being made
Yet the IEEE has made progress in providing better training and coaching to working group chairs, Sherman says. It has also been actively discussing a new suite of tools based on the suggestions to facilitate the entire process, he says.
"We do have some tools in place, but they are stopgap measures at the moment," he says, citing the 802.11 group, which has more than 500 members and often takes attendance every two hours, which would be impossible without some automation. "We are trying to come up with a suite of tools and it's a very active area, but it will take -- it wouldn't shock me -- about a year."
As with the standards-setting process itself, the challenge has been reaching consensus among the 802 Executive Committee on the right tool set and synchronizing the needs of Project 802 with the broader IEEE standards effort.
"Every meeting that's one of the major topics that we discuss," Sherman says, "and we are trying to do something that couples with IEEE SA, so it will really become more global, not just for 802, but across the board."
Yet most of the other ideas discussed at the November 2005 brainstorming session never gained traction, primarily because of fears that efforts to streamline the process would make it vulnerable to potential abuse. And even the perception of unfairness or abuse of process can inhibit Executive Committee action -- or spur it to action, as it did last year when controversy swirled around the 802.20 working group.
"At this point we're trying to deal with the perceptions that people in the group had," Mills says of the Executive Committee's decision to suspend the 802.20 project and restart it with new leadership. "There were allegations that there were process violations. . . . There were various levels of discomfort within the group," he recalls. "Without getting down to the fundamentals of trying to decide whether all that was true -- my guess is that it wasn't -- some of it was purely perception."
Yet that perception and the subsequent suspension of the working group "caused IEEE to take a close look at our processes and procedures," Sherman says. "And as a result of that I believe you'll see, not just in 802, but across the board within IEEE, modifications or at least more safeguards put in place to avoid abuse of process."
Those modifications are likely to be minor, similar to the tweaks college football makes each year to the much maligned formula that determines who plays for the national championship. And just as college football bowl games have a long history of crowning champions but each year seem more prolonged and controversial, the IEEE standards-setting process may be flawed but in most cases works, proponents and critics agree.
"There are definitely issues, but I think the challenge would be coming up with something that was better," Farpoint Group's Mathias says. "The problem with the process is that it needs to be open. And if it's going to be open, it's going to be open to everything: from new ideas to overt politics. I don't know what they could do to improve it."
Garretson is a freelance writer in Gaithersburg, Md. He can be reached at rgarretson@gmail.com.