site Search:


 
   
story category
PC Magazine Again Ranks Fastest ISPs By Web Browsing
And again people don't entirely trust their methodology...
by Karl Bode Thursday 24-Jun-2010 tags: competition · business · alternatives · bandwidth · consumers
PC Magazine has unveiled the latest data on broadband ISPs collected by their surfspeed application, which, unlike most speed tests measures the browsing speed of an ISP connection. When the application first launched you'll recall many of you took issue with PC Mag's data and their methodology, which they've since claim has improved and includes 236,739 tests from 6,449 users (methodology is here). The results are broken down by region (Northeast, West, Midwest South) and aren't too surprising, with Verizon FiOS taking top honors in the Northeast and West, while Time Warner Cable nabs the Midwest and Cox wins the South.

Click for full size
But before PC Magazine could even affix the crown to the top of Verizon's head for the "fastest ISP in the nation" honors, critics again began poking holes in PC Magazine's methodology. Speedtest outfit Ookla, which knows a thing or two about ISP tests, quickly blogged about how PC Magazine's test still fails in a number of major ways, and doesn't represent the performance of an ISP -- or even a modern web browsing experience:

the test happens in serial using two connections at a time, but modern web browsers go well beyond this old HTTP limitation. For example Firefox 3 and Internet Explorer 8 default to six parallel connections to a host when retrieving content. PC Magazine claims that they are trying to represent the speed of an actual web browsing experience, but apparently it is not a modern one. Ignoring that issue the test could still be used to compare rudimentary web browsing performance, but does that really represent the performance of an ISP? No.

In fact, Ookla goes so far as to note that if you include a larger data set, Verizon's ranking as an ISP starts to look downright pedestrian:

The fastest ISP according to PC Magazine is Verizon FiOS, but this was based on test results from less than 6500 users. Looking at Speedtest.net data and only factoring in major ISPs (ones with over 1000 unique IP addresses having taken tests), Verizon doesn't even break into the top 50. Decreasing the requirement to 100 unique IP addresses drops them out of the top 200. A much larger data sample certainly makes things more interesting.

Of course Ookla's factoring in a huge swath of users on Verizon's DSL network (of which the vast majority of Verizon customers still are). So depending how you look at it, Verizon is either the fastest ISP in the United States, or doesn't even really rank due to their significant number of slower DSL users.

view: topics flat text 
Post a:
fiberguy
My views are my own.
Premium
join:2005-05-20
kudos:3

WOW!

I can agree that Verizon would have the best in the east, but Time Warner in the MidWest? Really?

I'd think Cox or Comcast would... they both have deployed DS 3.0 and faster tiers where as TWC has a relatively more antiquated networks as far as HSI goes and isn't upping their speeds like the others.

Selenia
I love Debian
Premium
join:2006-09-22
Lanesboro, MA
kudos:2
Reviews:
·Verizon Online DSL
·AT&T Wireless Br..
·Verizon Wireless..

Re: WOW!

said by fiberguy:

I can agree that Verizon would have the best in the east, but Time Warner in the MidWest? Really?
Not everywhere. Verizon speeds suck in this area, but TWC is about 5x as fast sustained(as in not including Powerboost). Pings are way better too. I think TWC kills Verizon in most non-FIOS areas, of which there are more of than areas actually served by FIOS. Cablevision is way faster than TWC, but has a limited footprint up here. Both leave VZ in the dust in most of their footprints.
--
The new Sony rootkit-Using the ability to remove features you paid for. What's next? Boycott Sony products »[Rant] ps3 update = no more Linux

NYR 56
Premium
join:2000-12-05
Smithtown, NY

Re: WOW!

The problem is the critics are lumping Verizon DSL with Verizon Fios, which is retarded. Just because both are run by the same company doesn't mean they shouldn't be separated. Why don't they throw in dialup users too? I'm sure Verizon still has plenty of those too.
fiberguy
My views are my own.
Premium
join:2005-05-20
kudos:3

Re: WOW!

Because this is about "what company" is offering the fastest speeds in what area.. not what product. Verizon is a company.. AT&T is a company.. they're not going to break down U-Verse from DSL for AT&T... most ISP/Companies offer one service.. and as for dialup, it's about broadband.. not narrow-band to which is equal across the board... slow-as-snot for this day and age.

Verizon Fios & DSL ... doesn't matter.. There is no such thing as "Verizon" and "FiOS".. it's just Verizon.

NYR 56
Premium
join:2000-12-05
Smithtown, NY

Re: WOW!

I'm not saying you're wrong, but I just think it's pointless to be comparing such a statistic. I fail to see how it's relevant to group companies against each other, when their product offering, territory and size are all drastically different. I know 56k is no 'broadband' but to me, the difference between 56k and DSL is practically the same as DSL and Fios.
fiberguy
My views are my own.
Premium
join:2005-05-20
kudos:3

Re: WOW!

They're looking at the company.. this particular group/study chose to do that. Verizon, being they are one company, needs to be taken into account. They didn't split the office of the president at VZ, and they are the same corporation. So therefore, VERIZON offers services as low as 768K on up to 50+, right?

This wasn't a full on broadband map.. the way I read this is they are looking at which company offers the fasted ISP in regions... not specific locals.

NYR 56
Premium
join:2000-12-05
Smithtown, NY

Re: WOW!

I understand what and why they did it, I simply think it's absolutely pointless to do it this way.
fiberguy
My views are my own.
Premium
join:2005-05-20
kudos:3
Verizon IS the ISP for both the DSL product AND the FiOS product.. it wasn't ranking the product, rather, it was ranking the ISP.

The ranking was accurate as it was disclaimed and the intent was stated in this case.
majortom1029

join:2006-10-19
Lindenhurst, NY
kudos:1

1 edit

ugh

This is a little biased. most people who have say optimum online ultra would not read pc magazine therefore use their speed test.

These ratings usually dont work for smaller isp's like cablevision.

dvd536
as Mr. Pink as they come
Premium
join:2001-04-27
Phoenix, AZ
kudos:4

Re: ugh

said by majortom1029:

This is a little biased. most people who have say optimum online ultra would not read pc magazine therefore use their speed test.
Even most people on AOL wouldn't have anything to do with that rag.
the one who pays the most gets the best rating. lol
wake me when consumer reports has one.
--
The shortest distance between 2 points adds 1.5 stars to T. want $25? solve »coord.info/GC20A37 for me
dfxmatt

join:2007-08-21
Evanston, IL
from the get-go it appears these results aren't reliable at all.

If these results were for 600 thousand individual connections instead of 6400 then we might be talking about something worthwhile.

This is an enormous stretch for a nationwide test to be considered conclusive off of 6400 users. thats not much more than 100 per state. I would be skeptical of this even if it was 6400 users for a single state, honestly.

amonynous

@bellsouth.net

Not a good test.

How do these test factor in the people that do not subscribe to the fastest tier offered by their ISP?

I do not want or need anything faster than what I have - 3 meg tier. If they count me in their test, then wouldn't I bring down my ISP's average, even though I'm getting exactly what I pay for?

What they should test is actual speed to advertised speed to get a quality of service grade.
JoelC707
Premium
join:2002-07-09
Stone Mountain, GA
kudos:4

Re: Not a good test.

Actually the way they do their test will in essence cancel out people like you, me and everyone else. Lets face facts, do we NEED 50mbps or faster to browse websites? Not at all and that's why you and I don't have it. Sure the person with the faster tier might get their page 0.001 seconds faster than us but that's an insignificant margin. If they did this test correctly then yes people with lower speed tiers would bring down the speed average for that ISP (though it would have to be a lot of users to bring it down significantly).

In essence the way they do their test takes those people out of the equation. That's not to say it isn't tested correctly still but it is better than a straight raw download speed test because people like us would lower that average. If anything the straight raw speed test is only good for showing where people tend to subscribe to the faster tiers because in those areas that ISP will be faster than others.

I think your quality of service test idea is a good one though. Of course that requires people to know what they subscribed to which sadly some don't. Any ISP no matter where or who can be flaky. In Atlanta, every time I tried Bellsouth it was horrible and most of the people I knew in the area that had them didn't like it either. Comcast on the other hand was leaps and bounds better all around. Out here in Tucson I'm on Comcast as well and its not as stable at it was in Atlanta but its still a faster speed than what Qwest can offer me (never tried Qwest).

Of course things affecting speed do NOT have to be something the ISP has control over either. Out here everwhere I've seen the coax cables are literally run across the roof and in my personal case not a single run is continuous, every one of them has a barrel in it creating at least one splice per run. That combined with the fact they are left to bake and dry rot on the roof meant the signal getting to my modem/TV was shit. Replaced with a fresh, unbroken run and all is well again.
Hellrazor

join:2002-02-02
Abyss, PA

Silly...

Comparing FIOS to DSL is like comparing a dragster to a ford escort. Up Next - comparing Windows 7 to OS/2 Warp.
jbwhite99

join:2005-03-22
Raleigh, NC

Re: Silly...

yeah, it would be silly. OS/2 is the dragster, Windows is the Escort. Remember, OS/2 had better plumbing underneath, was more crash resistant, etc - just didn't have the ISP support and the graphics front end. It had toolbars before Windows, etc.

No Limit
Just Say No To Social Promotion

join:2000-10-07
Daly City, CA

Average internet bill in California is ~$37...

I know the article say average cost, but I would love to see any Comcast bill from California where the cost is near $37 just for Internet. Sure it's that price when you sign up for a year or 2 contract, but it definitely not that cheap after the contract is up. I wish I could pay that much for my comcast bill, but my internet connection is "fast" so I guess I can't complain too much.

BHNtechXpert
BHN Staff
Premium,VIP
join:2006-02-16
Saint Petersburg, FL
kudos:32
Reviews:
·Clearwire Wireless

Verizon will no doubt attempt to exploit the results...

Despite clear data from Ookla indicating extremely skewed results as presented by PC Magazine, Verizon is no doubt tripping over itself cooking up a way to exploit the results as presented by PC Magazine. The only thing missing from the article was a well placed Verizon FiOS ad.

FBGuy
yippee ki yay
Premium
join:2005-03-19

i about cried

when I started to read the comments for the pcmag.com article. those people don't know what they are talking about. at least the reaffirms what the fcc said.

kados
Hail Odin
Premium
join:2003-03-14
Watertown, SD
kudos:1

not even close

I am in South Dakota, and they don't even list the 2 prominent ISPs, Midcontinent and Knology

I'm with midcontinent and I pay $20 a month for 20/2 connection

I just did a speedtest.net and I had 19.6/1.9

flawed survey
--
Bla Bla Bla

benc
Premium
join:2007-06-17
Glen Carbon, IL
Reviews:
·Charter

Re: not even close

said by kados:

I am in South Dakota, and they don't even list the 2 prominent ISPs, Midcontinent and Knology

I'm with midcontinent and I pay $20 a month for 20/2 connection

I just did a speedtest.net and I had 19.6/1.9

flawed survey
     Seriously?  That's really cheap.  It sounds like a special or promotion that you were lucky to get.

kados
Hail Odin
Premium
join:2003-03-14
Watertown, SD
kudos:1

Re: not even close

yeah, It's normally 34.99 or 39.99 I got a deal for 2 years
--
Bla Bla Bla
davidl

join:2008-07-11
Vaudreuil-Dorion, QC

Legacy Media

I quit reading PC Magazine back in the early 90's

ARGONAUT
got ping?
Premium
join:2006-01-24
New Albany, IN

Node business like overselling business

I wonder how ISPs rank on overselling nodes?
tmc8080

join:2004-04-24
Brooklyn, NY
Reviews:
·Optimum Online
·Verizon FiOS

Re: Node business like overselling business

said by ARGONAUT:

I wonder how ISPs rank on overselling nodes?
That depends if you include the isps in Canada and around the world for that matter... but strictly speaking US.. Comcast will be at the top since they have the largest US footprint-- there is huge financial incentive to "SCREW" many non-competitive areas.

hayabusa3303
Over 200 mph
Premium
join:2005-06-29
kudos:1

i didnt know

that pc magazine was still around.

sivran
Back to Opera again
Premium
join:2003-09-15
Arlington, TX
kudos:1
Reviews:
·RoadRunner Cable

Ookla's Glass House

"Wha! Your speed test is inaccurate!" cried the pervasive purveyor of inaccurate flash-based speedtests that can't even measure my cable speed right.

Pot, meet Kettle.
--
So I'm pretty much disappointed with both political parties. Yeah.
tmc8080

join:2004-04-24
Brooklyn, NY
Reviews:
·Optimum Online
·Verizon FiOS

browned results

»www.youtube.com/watch?v=RO2xi0uLnj8

Monday, 04-Jun 22:24:02 Terms of Use & Privacy | feedback | contact | Hosting by nac.net - DSL,Hosting & Co-lo
over 12.5 years online © 1999-2012 dslreports.com.