|Sunday, 24 August 2008 20:20|
The greatest favor you can do this reporter is to correct errors, whether of fact or analysis.
September 20, 2012 Corrections: Linley Gwennap pointed out I was careless saying "very few of today's chips are produced at less than 40nm." He reminded me Intel is shipping millions of PC processors in leading-edge 22nm technology. While not that many designs are shipping in quantity below 40nm, they include Qualcomm's 28nm MSM8960 processor and chips from Altera, AMD, Nvidia, and Xilinx. In addition, both the iPhone 5 and Samsung's popular Galaxy III use 32nm chips. He adds “Regarding the future of Moore's Law, Intel is already testing its 14nm process, which is due to enter production in 2014. Intel is due to hit 10nm in 2016, with TSMC likely to follow within a year or two. So Samueli's comment that 10nm will carry us another 10 or 20 years surprised me. 10nm represents less than a decade of progress. I think he meant to say 10 years and got a little carried away with the 20-year comment.”
I'll update this as quickly as I can get to it; apologies I am so far behind. Over ten years, I believe the most significant errors I've made have been overestimating the growth rate of the Internet ("Doubling every 120 days") for too long, expecting a more rapid shift to VDSL from ADSL, and an (apparently) mistaken report that Time Warner Cable was ready to sell the New York franchise area. All had senior and reliable sources, but that's not an excuse.
If I discover a mistake in anything I reported or put forth publicly I believe it's my obligation to offer a correction. That's not just errors I make, but also errors in others I quoted or even a projection that proves mistaken years later.
June 28 In the initial web posting of the Alcatel story. I confused Alcatel's 100G long range optical unit with their 100G edge router. Alwan pointed out to me this may be the first time one company had the first units in both categories for a new generation.
Although Cisco is ready to demonstrate a gigabit with AT&T, I don't know if that will occur. I was told they wanted to have something to show AT&T wasn't behind Google, so would make an announcement. It's highly likely that would apply only to very limited builds, and apparently too many reporters are asking questions about that. Even the usually tame D.C. regulars are starting to wonder. The Cisco/AT&T CRS-3 announcement was so over-hyped that holding back is probably a good idea.
I should not have written "CETF, entirely funded by AT&T and Verizon with $60M, is being given control of California's $B share of the broadband stimulus." I should have written something like "CETF, entirely funded by AT&T and Verizon with $60M, is playing a crucial leading role in California's $B request for broadband stimulus."
Apologies to Dennis François of KPN and Steen Garbers Enevoldsen for the mis-spelling.
In calculating AT&T’s DSL figures for the last quarter, I apparently used a number for their U-Verse IPTV, not U-Verse data. I’m probably off by 15K. Thanks to Lawson Key for catching that. Correcting an error is always a favor to a reporter.
Eli Noam points out to me I was offbase saying the 1990's U.S. FCC team drove the belief "competition is job #1" and "deregulation is job #2." Since they are coming back to power in the U.S. (Jules was counsel 1994-1996, Blair, Reed, Kevin Werbach and Don Gips were on the transition team,) I've been reading what they wrote at the time. Kevin's 1997 essay is a very clear presentation of these competition+deregulation = nirvana ideas, but Noam finds many precedents I didn't know. Apparently, that was the general consensus around the Western world around 1990. Two decades later, it's become clear that strong competition is unlikely in many parts of telecom. Where competition succeeds (French broadband) it requires very strong regulation. The normal trend in this business is consolidation, Eli's Columbia colleague Raul Katz points out. The history of the nirvana approach is important today, because I haven't met a policymaker with consistent ideas about what to do when competition is weak. It's a tough question.
Corrections from 2008
Corrections from 2007
Corrections from 2006
Corrections from 2005
"One-tenth of BellSouth customers should have 50Mbps+ service from fiber to the curb," I told DSL Reports. That was because BellSouth had promised the same to Kevin Martin in return for killing unbundling on those lines. AT&T has killed the project now, and Martin hasn't done anything about it. That's a surprise, because many of those lines are in Kevin's home state of North Carolina, where he hopes to run for office. I don't think Kevin wants his opponent to say "100,000 of thousands of you have an Internet connection that's 50-90% slower than it should be because Kevin Martin did nothing when AT&T broke a committment to deliver those speeds." I also said "by late 2006, VDSL2 low profile will be within $10 of the cost of ADSL, most carriers will switch over even for the small improvement." The price was about right, and several did switch, including Deutsche Telekom. But most carriers held off because of issues of power, heat, line card capacity, and interoperability. I made the mistake of believing the CEOs of the chipmakers and others who promised to solve those problems quickly. In 2008, BT and others are making the switch, but some of the issues are still not solved. DSL Forum folks - what happened to the interoperability you expected two years ago?
Corrections before 2005
One of my worst errors of analysis was from 1999 until I met Andrew Odlyzko. I had heard from FCC Chairman Bill Kennard, AT&T Chairman Mike Armstrong, and MCI's John Sidgmore "Internet Traffic is doubling every 120 days." I believed them, although we later discovered it was wildly untrue. I used that figure to extrapolate that by mid-decade the average U.S. home would require over a megabit of committed bandwidth, and present that idea to the DSL Forum in a speech. Odlyzko, a world class mathematician who then worked at AT&T Labs, discovered the numbers everyone was quoting were totally contradictory, as I discovered at a Columbia University CITI event. I went up to him afterwards, and said he couldn't be right, quoting a slew of seemingly informed people including his own company's Chairman. Andrew told me to "just look at the data." I did, he was right, and I was one of the first journalists to report that "everyone" was overestimating Internet growth. I got props from some for predicting the Internet bust that came soon after, but I've always remembered how long and publicly I made the same mistake. Andrew, now a friend, is at the University of Minnesota and burst another myth with his MINTS database. Internet traffic is growing rapidly, but the rate of growth is essentially flat to down since 2002. http://www.dtc.umn.edu/mints/home.php The average speed of the net is going up, not down. Extraordinary amounts of money have been spent suggesting the Internet is collapsing unless the lobbyist get what they want. Analysts at Deloitte, Bernstein, Nemertes and others made what I assume were honest mistakes that they haven't corrected despite overwhelming evidence. Their mistakes have been amplified so effectively that journalists at the London and New York Times, Guardian, and other many regulators (Ed Richards, Debbie Tate) believe some total nonsense. (Thank you, Bruce, Craig, Larry, Bob, Jim, Bob, Tom, and the rest. You're incredibly effective at persuading people.)
E veryone make errors and it's the responsibility of any reporter, scholar or analyst to visibly correct them. In practice, most publications correct obvious errors when pointed out, such as when the Times mis-spelled Dave Burstein. The vast majority of errors aren't that direct, and most overt mistakes even in the most respected media go uncorrected, or simply mentioned in a letter to the editor. I believe that if I've printed something untrue, whether as my reporting or quoted from a source, a correction is the right thing to do, even if discovered five years later and no one had ever complained. I also correct in my publications similar errors I've made in other public settings, such as an industry speech or a newspaper interview. As I moved the site to Joomla, I'm discovering with the help of hindsight many necessary corrections. It's embarrassing but the right thing to do, although before 2005 I'm including only the most important.
we'll be able to implement their targeting and database technology to deliver advertising to the individual user,'
- Latest Issue
- Moffett’s Back: The Most Original Analyst on Wall Street
- 400 Megabit Cable Boxes Shipping
- Australia: Almost Certainly VDSL Instead of Fiber Home
- “Look Out Google Fiber, $35-A-Month Gigabit Internet Comes to Vermont”
- TNO: Top DSL Conference Coming to Dutch Beach
- Times on Wheeler "An Industry Man for the F.C.C."
- Lantiq Sampling 3x3 MIMO for Faster VDSL Gateways
- How it Works: Gigabit Cable Coming in 2013-2014. Downstream only.
- 60 Down, 18 Up as VDSL comes to France
- $10 Qualcomm Quadcores will Push $99 Phones Past iPhone 4
- Confirmed: Wireless Data Growth Rapidly Slowing
- In DC: Are There Any New Ideas For Broadband?
- Dan Artusi: Vectored Chips Shipping this Summer in Volume from Lantiq
- White Spaces in Real Deployment