Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?
The Internet Networking United States Technology

The Fastest ISPs In the US 199

adeelarshad82 writes "PCMag recently put Internet browsing speeds to the test to see which ISP was the fastest. The results were based on a quarter million tests run between May 1, 2009, and April 30, 2010, by more than 6,000 users. The tests were carried out using SurfSpeed, which takes into account the complete, real-world download time of a web page to a browser. According to the results, Verizon's FiOS took the top spot as the nation's fastest ISP, with a SurfSpeed score of 1.23 Mbps. Interestingly though, of all the regions where Verizon's FiOS is available, its dominance is only seen in the northeast and the west, whereas cable service from Cox and Comcast won out in the southern region. Moreover, cable through Cox and Optimum Online beat AT&T's fiber optic service in the nationwide results, with SurfSpeeds of 1.14Mbps, 1.12Mbps, and 1.06Mbps respectively. The worst results mostly consisted of DSL providers, bottoming out at 544 Kbps from Frontier and going up to 882Kbps by Earthlink. Other interesting facts noted in the test were that broadband penetration was highest in Rhode Island and lowest in Mississippi, while the average Internet bill was highest in Delaware and lowest in Arkansas."
This discussion has been archived. No new comments can be posted.

The Fastest ISPs In the US

Comments Filter:
  • Neat, but... (Score:2, Interesting)

    by Chih ( 1284150 )
    I'm more interested in cap numbers these days
  • Mississippi (Score:5, Funny)

    by spike hay ( 534165 ) <> on Saturday June 26, 2010 @01:34PM (#32703264) Homepage

    Is there any metric for which Mississippi is not the worst state?

    • Yes, where Louisianna is last.

      / Louisianna native

    • by Evildonald ( 983517 ) on Saturday June 26, 2010 @01:43PM (#32703312)
      They are the nation's leaders in S's and I's
    • Re: (Score:3, Funny)

      by ffejie ( 779512 )
      U.S. States ranked by similarity of their name to the word "Mississippi":
      1. Mississippi
      2. All other 49 states.
    • by dmomo ( 256005 )

      BBQ. Maybe not the best, but def. not last.

    • Re:Mississippi (Score:5, Interesting)

      by Mitsoid ( 837831 ) on Saturday June 26, 2010 @02:21PM (#32703574)
      Yes, you can buy higher speed and thus invalidate the 'findings' of the article... Which is really just self-promotion with a fancy title to attract attention.

      I was kinda disappointed that the article doesn't address maximum speed, or average speed amongst all "5mbps" connections, instead it lumps in DSL, Cable, and Fiber and says "HEY LoOk! Fiber is usually faster!!"

      What this is really testing is "How much speed do Americans purchase, by region,"... it's just.. almost.. completely useless... except for a few statistical data points that are not frequently mentioned (broadband penetration by state).. We're comparing ISP by what the end-users paid for, as opposed to what end users CAN pay for (i.e. the limit of the technology)... or, as an alternative test, they could have tested Like-speed connections average performance across carriers, but instead they are grouping DSL, Fiber, and Cable..... and ignoring that some people pay $20 for internet while others want to pay $50 (for semi-basic home internet service) and claiming an ISP is "The best" because they have more users that spend more money on internet. (or they have less users but much higher speed to result in the same data skewing of results).

      So yeah, Metrics, IMO, are mostly crap. And Mississippi can pull ahead of every state in this 'survey' simply by spending an extra $5... hell for $10 extra you can probably get speeds 5 times faster then most of the United States!
      • Exactly. Besides, if they didn't care about comparing technical apples and oranges, they could have kept the money constant. Compare the broadband speeds available for $20/mo, $50/mo, $75/mo. Plot those two parameters on a graph, and color code the points by how much competition there is in that market. Now that would have been interesting. Not surprising, but interesting.
      • It's pathetic speed is still a metric, Koreans and Japanese are realistically pulling down 30+ MBPS.

        We're way too far behind to catch up except with better technology (VDSL, Fios, etc).

        In the short term though we need lower latency, the latency is high because of packet sniffing and low consumer demand.
        We do need it though, clearly cell prices aren't dropping and VOIP is the only way out (unless we band together to buy a chunk of spectrum)... so we'll see I suppose!
    • Is there any metric for which Mississippi is not the worst state?

      Canned Possum consumption. Georgia takes it by a country mile.

    • Oklahoma is the worst in state government corruption.

      Thou Louisiana will contest this, naturally...

  • I would rather measure the available bandwidth with, say, google services, network latency and a few round trip timings with known hosts.
    This sounds more serious to me than anything else.
  • by t0qer ( 230538 ) on Saturday June 26, 2010 @01:43PM (#32703322) Homepage Journal

    Even though Speakeasy was slashdot recommended, a lot of my geek friends used it, I had to cut them this year.

    I never got the advertised speed out of them for what I was paying. My business was close to the CO, but when I'd complain their answer would always be "Replace the wire going from the pole into your building"

    Why should I have to do that? I'm old, I hurt when I fall. NO thanks.

    So after 6 years with SE, I called up Comcast. They sent an installer who made sure everything was working right. My speeds were out of sight, 20mbps down and 5mbps up. My bill is $20@mo less too.

    DSL can compete, but they have to give up a little margin for better customer service.

  • Fuck Comcast (Score:2, Offtopic)

    by Aldenissin ( 976329 )

    I came just to say this.

  • by Josh Triplett ( 874994 ) on Saturday June 26, 2010 @01:54PM (#32703410) Homepage

    Despite using bandwidth units (Mbps), their "SurfSpeed" "benchmark" actually depends heavily on latency, as it tries to simulate a web browser fetching resources sequentially from a site as it discovers them.

    Found this report analyzing the article and the benchmark: []

    • Also, what are all the variables they are taking into account? If I have a screaming fast ISP, but the server I'm connecting to is hosted on some dog slow server, then the site is going to slow me down. It's not my ISP's fault, it's the site I'm visiting. Do they account for this and filter out the server's speed?

      • by Josh Triplett ( 874994 ) on Saturday June 26, 2010 @03:59PM (#32704222) Homepage

        According to the article I linked to, they access a grand total of 10 sites:,,,,,,,,, and

        On the one hand, I'd expect none of those sites to have a slower connection than any consumer ISP. (Some sites with large files such as video sites will throttle for bandwidth reasons, but no sane site throttles HTML and similar; better to just serve the files quickly and close the connection.)

        On the other hand, that doesn't look like a particularly representative sample of "top" sites. Who uses mapquest anymore? And how often does the average user visit or (As opposed to or, which seem somewhat more likely for regular visits. Windows Update doesn't count, since *hopefully* that gets much more non-interactive use than interactive use. Similarly for the various Apple services, which don't necessarily live on the same server as

        But in any case, the bandwidth of the server will matter less to SurfSpeed than the latency of responding to each request quickly so it can start the next one.

        The concept of a benchmark for real-world site load times seems perfectly reasonable, but it should not have a misleading unit of "Mbps". A better idea: measure the total number of milliseconds required for page loads during some representative real-world browsing paths (*not* just site front pages either).

        • Actually, is a pretty decent portal for a few things that aren't search. They've got a web-based sort-of repository-like thing (ok, maybe it's more like tucows, but for stuff you'd actually want, and have to pay for.) with various applications that can be downloaded that run on macs, as well as some other things like movie trailers and automator actions, and whatnot.

    • by fermion ( 181285 )
      The thing about real world experience is that the only way to measure it is to make it real world. If we are fetching resources from a web site, part of that process is rendering. For instance, i would expect slashdot to load in a few hundred milliseconds based on data size, with a couple seconds to render. This leads to the three second observed time to load. Pages with complex scripts are going to render even slower. What this means is that for anyone on DSL or cable, one would likely do better to sw
  • This is interesting, considering the ad-hoc testing I did recently. I'm a Comcast customer in northern De, and DSL reports' speed test consistently gives me about 8Mbps down bs 1-2Mbps up.

    My parents, I. Southeast PA, have FIOS. For giggles, I did the same DSL reports test, and got about the same results.

    Do any other slashdotters have similar experiences?

  • by countertrolling ( 1585477 ) on Saturday June 26, 2010 @01:59PM (#32703448) Journal

    I would like to know how much more spam they are getting now. Nice data harvester. I knew the article was a fraud when it said,"...cable and phone companies compete to provide fast connections..." What they possibly compete for are exclusive franchises.

    I bet if you block the ad servers, your speed would double

    • I noticed that web pages were loading the info-line at the bottom of my browser was going out to an ad server. The page would not load hardly anything until the ad server finished. So I installed an ad blocker and the speed to load a page increased quite a bit. There are some web sites I support and I give them a pass and let their ads come through.
  • Oh wait, I already knew th <carrier disconnected>
  • I'm having trouble believing that this test is useful for anything, if I'm understanding their methodology.

    They should be giving TOTAL TIME to download a web page and all its assets, including DNS lookups. That's the only measurement that matters for web browsing.

    Transfer rate is such a small importance to most people -- as an example, their slowest transfer rate (Frontier DSL) would download one of their ~21KB review pages in about 31ms. Their fastest (Verizon FIOS)? About 14ms. The difference is negli

    • It's messed up anyway. Web browsing isn't a useful metric for measuring internet speed on broadband. We need HD quality real time video streams that are large enough to negate any benefit added by speedbooster.

      • They weren't trying to measure all facets of internet speed on broadband. If I misunderstood and that actually was their intention, I agree there are even more holes in their methodology. From TFS:

        put internet browsing speeds to the test

    • Their speeds appear to be based upon not the sustained transfer rate but a normalized rate that incorporates the DNS lookup(s), latency, etc. into the calculation

      On a side note, for those stuck with Crapcast (TM), at least those in the Twin Cities area you can achieve substantially snappier web browsing if you replace their DNS server with something decent. Of which there are plenty of free DNS server providers such as OpenDNS [] among others.

    • by adolf ( 21054 )

      Given the total time to measure a single web page is useless because the sample set would be very small (one web page? really?). Increasing the sample size, however, results in too much data: Suppose it tests 50 web sites. And suppose there are 20 ISPs tested. 50*20=1000, which is way too fucking many data points for layfolk to digest.

      Besides, web sites change. They aren't static things. One day it might be big, the next it might be small. Ads rotate. If they were just measuring time, then the da

  • People are proud of a 1Mb connection? What's the latency? Even here in Germany a provider would be ashamed to show his face if he couldn't do at least four times that for a fucking rural area!

    A year and a half ago, 100 times that speed was considered good [] and in a year and a half from now Korea expects to have ONE THOUSAND times that fucking speed. I know people in US states who can still only connect with a fucking 33.6Kbaud modem.

    • > People are proud of a 1Mb connection?

      They aren't measuring the bandwidth of the connection. They are measuring the average download rate from a bunch of Web sites, including DNS lookups, server bandwidth, etc. Not particularly useful.

    • Sadly, it was a step up for me. From rural Vermont I was getting 0.75Mb down, and 0.2Mb up, on DSL, which was the only thing available. (I was 20 miles from where cable ended, there was no fiber in the state, for the most part.) I moved to the outskirts of a city, and on a limited budget, pay $40/month for 1.5Mb down, and about 0.3Mb up.

      That's the state of broadband in the US. If I was willing to pay $80-$100, I could add phone and TV, and double those speeds. But I'm not willing to do that.

      We're a deca
  • When you have a cap? ( or worse cap + overage charge ) It just means you get there faster.

  • I live on Brazil, a Third-world country. And i have a working 3Mbps download / 1Mbps upload. Houston, the north-americans have a problem!
    • > Houston, the north-americans have a problem!

      Read the article. They did not measure bandwidth. They measured the average rate at which a bunch of Web pages could be dwonloaded, including DNS lookups, latency, waiting for slow ad servers, etc.

    • Ignore the other poster. Even connections just listed as 3Mbps/1Mbps are probably the median or above median number for the US. When friends of ours sprung for fiber listed at 10Mbps/3Mbps, we were all jealous, as that was 2-10x better than what we were getting. Of course, that was costing them over $100 US per month.

      We're a third-world country when it comes to bandwidth, almost.
  • Did we take into account all the bittorrent clients flooding the upstream and choking the downstream? No? O'Rly!
  • by moxley ( 895517 ) on Saturday June 26, 2010 @02:54PM (#32703800)

    As everyone has pointed out, this test in this article really isn't measuring the bandwidth that your ISP is providing; it's like saying "let's see how fast you can run - oh, by the way, you'll be wearing this heavy backpack, dodging traffic." They say it's real world surf performance, but there are so many variables at work here that it really isn;t a very useful metric.

    You can use the JAVA or Flash based speed tests at places like (which is a great site BTW if you aren't familiar for it) those tests are fairly accurate - but not always.

    The best, most accurate way I have found to test whether I am getting the speeds I am supposed to, is to use newsleecher and download a bunch of binaries from my premium newsgroup provider. I use Giganews, and I have been really happy with them, but I assume the other top tier newsgroup providers are similar..... With most premium news providers, you get multiple connections and most of the good ones can max out your connection at anytime, provided you are using multiple connections.

    I'm sure that most people here know this, but if not: - to figure out if you're getting what you're supposed to, once you're as certain as you can be that you are maxing out your connection, take youy average download speed in megabytes and multiply it by 8.

    I live in Philly and have a 22 megabit at home, and 50 megabit at work.

    When downloading at home I get about 2.8 megabytes/sec.....when downloading at work I get about 6.2 megabits per 2.8 x 8 = 22.4 and 6.2 x 8 = 49.6 So all is well...if I notice that something seems to be off, or slow - the first thing I do is queue up some binaries and check....

  • I live in Japan. /wins

    (On a serious note, I get ADSL 50Mbps for about $60 a month in a small Japanese city. I could also get a fiber connection if I wanted to for slightly more)

    • I live in Japan. /wins

      (On a serious note, I get ADSL 50Mbps for about $60 a month in a small Japanese city. I could also get a fiber connection if I wanted to for slightly more)

      Wow, if I could afford it, the most I could get in DSL is 40mbps down and 20 up. This service is double what you pay in Japan. For 50.00 per month, I have 12 down/6 up.

  • ...I had a couple of friends stop by for a taste test. I blindfolded them, laid out a pear, an apple, an orange, a kiwi, a mango, and a banana, and asked each of them which tasted closest to a pear.

    Wouldn't you know it...the pear came out #1! Closely followed by the apple. For some reason, the lowly orange ended up last. I'm thinking about writing a blog article about this. Might even make Slashdot...

  • Qwest might be slightly underrated. I am not sure how recent this survey is or rather how recent PCMag claims. About 10 months ago, Qwest began rolling out FTTN (Fibre To The Neighborhood) in much of the Phoenix and Tucson greater metro areas. Each subdvision has a fibre junction. My ADSL2+ is quite literally 500 wire feet from the fibre box and I get some really good speeds. I hit a download rate of 2.36MB/s. This was incredible for DSL!
  • My answer is No. In Holland much faster median ISP's, the fastest is XMP and has 200 MB/s down- en upload. Compare that with America's fastest of a mere 1.23 Mbps: America's fastest is 163 times slower. And XMP's bandwidth is relatively low considering the architecture. If demand required higher speeds XMP can using its current architecture easily increase to 1000 Mb/s. That's 813 times faster. The average Dutch ISP speed is 16.19 Mbps, still 13 times higher.

    Holland is not the only one, there are more co

You are in a maze of little twisting passages, all different.