Speedtest.net Opens Up Their Data Free to academics, ranks fastest cities/countries Mike Apgar (formerly of Speakeasy) and the folks over at Speedtest.net this week decided to open up their speed test data for public, government and private research. According to a company statement, that data includes over 1.5 billion tests so far (about a million a day). Academics will have access to the data for free, but businesses and private institutions will need to pay for it. The company compiles a rolling, 30 day average speed ranking by city and country. South Korea (36.5Mbps), Latvia (23.3Mbps), Republic of Moldova (21.5Mbps), Japan (20.3Mbps) and Sweden (19.8Mbps) round out the fastest countries. Meanwhile, in the United States, the top performing states are Delaware (15.6Mbps), Rhode Island (15.2Mbps), Massachusetts (15Mbps), New Jersey (14.1Mbps) and Virginia (13.6Mbps). Alaska checked in as the slowest state with a 2.8 Mbps average download speed. As for the fastest cities globally, Seoul (34.66 Mbps), Riga (27.90 Mbps), Hamburg (26.92 Mbps), Chisinau (24.43 Mbps) and Helskinki (20.60 Mbps) take the top five spots. A U.S. city doesn't appear on the chart until San Jose at spot 18 at 15.03 Mbps. You can take a look at the entire index here.
|
  kapilThe Kapil
join:2000-04-26
Chicago, IL | Cool! If the FCC won't gather and provide us with accurate speed data to gauge the real state of broadband deployment in this country, then we'll take matters into our own hands.
Kudos to Apgar and company.
Speedtest.net ranks high in search results and, consequently, is the default destination for many when they want to run a speed test...so the data has got to be pretty universal and accurate.
As an aside, this move reminds me of the good old speakeasy days....before they went to shit and then got bought out by best buy to seal the deal. It's fascinating to see how much companies embody the spirit and values of their founders...some manage to carry them on after the founders leave, others turn into Speakeasy.
--
»www.VoIPTrunk.com | |
|  | | | |  Romney2012Defeat Obama 2012-Chg we can believe inPremium
join:2002-03-03
USA
kudos:4 | Re: Cool! Which brings up the question of why the FCC needs to build their own bureaucracy to collect speed data on their own. All they need to do is buy this data from OOKLA & M-Labs and put a couple of programmers to work.
--
Are you happy with your rep in Washington, DC? | |
| | | |  MxxCon
join:1999-11-19
Brooklyn, NY | Re: Cool! because fcc's version of the site records more accurate address locations. ookla or mlabs don't ask for your street address. | |
| | | | | Reviews:
 ·magicjack.com
| Re: Cool! said by MxxCon:because fcc's version of the site records more accurate address locations. ookla or mlabs don't ask for your street address. Another problem with random/anonymous data like speedtest's is that it's measuring how cheap customers residents of some states may be compared to other states.
For example, I have the highest Cox service available to me. But, I choose the lowest one (3/256k ?) because that's all I need.
The only way to measure how fast a state is is to know what's available, at how many addresses, for how much money.
The tier they choose is a different measurement.
Mark | |
| | | | | |  morboComplete Your Transaction
join:2002-01-22
00000 | Re: Cool! said by amigo_boy:The only way to measure how fast a state is is to know what's available, at how many addresses, for how much money. You don't need "perfect" data to be able to draw reasonable conclusions. No data to suggest that cheapness differs by state, so cheapness is irrelevant. | |
|
| | | | Romney2012Defeat Obama 2012-Chg we can believe inPremium
join:2002-03-03
USA
kudos:4 | said by MxxCon:because fcc's version of the site records more accurate address locations. ookla or mlabs don't ask for your street address. That is coming soon to the Ookla speedtest.net web site:
--
Are you happy with your rep in Washington, DC? | |
|
| | | | | | MxxCon
join:1999-11-19
Brooklyn, NY | Re: Cool! i heard that blue colored tin foil hats don't work as well as uncolored ones. | |
| | | | | |
| | | My speed is the same as the average speed in Kenya. | |
| | | MxxCon
join:1999-11-19
Brooklyn, NY | Re: Cool! said by Bobcat:My speed is the same as the average speed in Kenya. that's the desired result | |
| | |  sporkmedrop the crantini and move it, sisterPremium,MVM
join:2000-07-01
Morristown, NJ
 ·Optimum Online
| said by Bobcat:My speed is the same as the average speed in Kenya. I hate to do this, but since we're almost neighbors:
 | |
| | | |  joebarnhartPaxio evangelist
join:2005-12-15
Santa Clara, CA | Re: Cool! There's always a faster connection...
(And, yes, I know there's folks out there with even faster home connections.) | |
| | | | joebarnhartPaxio evangelist
join:2005-12-15
Santa Clara, CA | This is fun! Let's try the server over by YOUR house instead of mine...
Not bad for a cross-country trip! | |
| | | | | | Cablevision sucks donkey balls. | |
|
| |
| | | Mean vs. Median More important than averages, which are often skewed by extreme results, is the median speed offered in a country. In other words, what you'll achieve at the 50th percentile. If I recall correctly that demonstrates even more strikingly how much faster countries like Japan and Sweden are than the US. | |
|  BF69Premium
join:2004-07-28
Camden, TN
1 edit | speedtest.net sucks used to be good until a couple of years ago. Results are always way off now. I can 100% gaurantee you my 16 mbps connection is not giving me 4.6 Mbps speeds. Especially when 3 other websites are showing speeds of 13.3, 13.7 and 15.2 mbps. Who would you trust? | |
| | Lazlow
join:2006-08-07
Saint Louis, MO | Re: speedtest.net sucks Assuming you are right, it really does not matter for comparison purposes(this study). If they are off for you, they should be off for everybody. The only thing it would changes is the absolute number and not the relative standings. | |
| | | BF69Premium
join:2004-07-28
Camden, TN | Re: speedtest.net sucks said by Lazlow:Assuming you are right, it really does not matter for comparison purposes(this study). If they are off for you, they should be off for everybody. The only thing it would changes is the absolute number and not the relative standings. That's my point they most likely are off for everyone and this flawed data is what the government is going to rely on? more data is not better if it's bad data. | |
|
| iansltx
join:2007-02-19
Golden, CO
kudos:2 Reviews:
 ·Verizon Online DSL
·Comcast
| Have you tried using a server that "isn't" the closest one to where you are? ay seem counterintuitive but if the nearby server is on a lousy network then you'll get crappy speeds. For me, Speedtest.net has always been accurate, maxing out PowerBoost on my Comcast connection when I test (~20 Mbps down, ~3.5 Mbps up). FOr sustained downloads I just download a test file from somewhere speedy (like SoftLayer). | |
| | | BF69Premium
join:2004-07-28
Camden, TN | Re: speedtest.net sucks said by iansltx:Have you tried using a server that "isn't" the closest one to where you are? ay seem counterintuitive but if the nearby server is on a lousy network then you'll get crappy speeds. 3 closest servers.
5.07
3.30
8.64
10 random servers
Atlanta, Chicago, NY, Dallas, LA, Seattle, Miami, Denver, Cincinatti, Kansas City
13.2, 10.52, 12.43, 11.23, 6.04, 5.85. 1.37, 0.99, 8.84, 9.01
Happy now? Anyone there close to 16 Mbps?
By the way they do get the UPLOAD correct. | |
| | | |  bencPremium
join:2007-06-17
Glen Carbon, IL
·Charter
1 edit | Re: speedtest.net sucks said by BF69:3 closest servers. 5.07 3.30 8.64 10 random servers Atlanta, Chicago, NY, Dallas, LA, Seattle, Miami, Denver, Cincinatti, Kansas City 13.2, 10.52, 12.43, 11.23, 6.04, 5.85. 1.37, 0.99, 8.84, 9.01 Happy now? Anyone there close to 16 Mbps? By the way they do get the UPLOAD correct. Only 11.23 with Dallas? That's odd. I just tested with the Dallas server, and got 21.47.
Dallas:
Chicago:
Edit: Added the following.
I just thought of something. Ever tried the site in different browsers. I find that I get better scores if I use IE instead of FF. It may be due to the fact that I'm running many Firefox add-ons, while I almost never use IE. | |
| | | | | | | Re: speedtest.net sucks Certainly IE returns more accurate than FF, always. And I would like to know why. | |
|
| | | Speedtest.net is one of the few sites that can reliably test my 50/10 connection. | |
| | bencPremium
join:2007-06-17
Glen Carbon, IL Reviews:
·Charter
| said by BF69:used to be good until a couple of years ago. Results are always way off now. I can 100% gaurantee you my 16 mbps connection is not giving me 4.6 Mbps speeds. Especially when 3 other websites are showing speeds of 13.3, 13.7 and 15.2 mbps. Who would you trust? Have you considered that it could be due to Powerboost? Try downloading a large file, such as a Linux ISO. Then look at the transfer rate while it's downloading. | |
| | | BF69Premium
join:2004-07-28
Camden, TN | Re: speedtest.net sucks said by benc:said by BF69:used to be good until a couple of years ago. Results are always way off now. I can 100% gaurantee you my 16 mbps connection is not giving me 4.6 Mbps speeds. Especially when 3 other websites are showing speeds of 13.3, 13.7 and 15.2 mbps. Who would you trust? Have you considered that it could be due to Powerboost? Try downloading a large file, such as a Linux ISO. Then look at the transfer rate while it's downloading. How would powerboost be REDUCING my speeds? Also how come the other sites have the CORRECT speeds? Simple logic says it's speedtest.net. As I said they used to be the best. Then about 2 years ago they made a change and ever since then they've sucked. Coincidence? I think not. | |
| | | | bencPremium
join:2007-06-17
Glen Carbon, IL Reviews:
·Charter
| Re: speedtest.net sucks said by BF69:How would powerboost be REDUCING my speeds? Also how come the other sites have the CORRECT speeds? Simple logic says it's speedtest.net. As I said they used to be the best. Then about 2 years ago they made a change and ever since then they've sucked. Coincidence? I think not. Oops. That's what I get for trying to read your post too quickly. I thought it said something else, I'm not sure why.
As someone else suggested, try using a different server. The closest server to me is the one in St. Louis. However, I get better results with the Chicago server. Oddly enough, I often get higher values with the Dallas server. | |
|
| | The numbers don't add up. How can a state have a higher average throughput than an individual city?
If San Jose has an average of 15MB, then how can Delaware have an average of 15.6MB? Wouldn't there have to be a city in Delaware that surpasses the state (and therefore San Jose's) average? Or do they ignore population centers below a certain number?
Something doesn't add up in the numbers posted in the article. | |
| |  BK
join:2001-09-10
Wheaton, IL | Re: The numbers don't add up. if you look at the data, a "city" for this data means at least 75,000 unique IP addresses
--
-formerly sMoKeNiNja | |
| | | iansltx
join:2007-02-19
Golden, CO
kudos:2 Reviews:
·Verizon Online DSL
·Comcast
| Re: The numbers don't add up. No, that's a country. I can guarantee that Fredericksburg, TX doesn't have 75K IP addresses but it's stil on the map.
In fact, I just downloaded their 250MB (or so) zip file with all the data inside; Fredericksburg, TX today ranges from 250 to 400 tests per day. Even if each test was from a different IP each day (highly unlikely) it would take 300 days to get 75,000 unique IPs  | |
|
| Romney2012Defeat Obama 2012-Chg we can believe inPremium
join:2002-03-03
USA
kudos:4 | said by Squished :
How can a state have a higher average throughput than an individual city?
If San Jose has an average of 15MB, then how can Delaware have an average of 15.6MB? Wouldn't there have to be a city in Delaware that surpasses the state (and therefore San Jose's) average? Or do they ignore population centers below a certain number?
Something doesn't add up in the numbers posted in the article.
Unincorporated suburbs often have higher speeds than the cities they surround. But they won't show up as individual localities in the data.
--
Are you happy with your rep in Washington, DC? | |
| | | iansltx
join:2007-02-19
Golden, CO
kudos:2 Reviews:
·Verizon Online DSL
·Comcast
| Re: The numbers don't add up. Depends on how the IPs are geolocated. They might show up as part of the cities they surrond, or tey might be something totally different.
Also, the Ookla reports don't include test results from really small cities that Speedtest.net (and Ookla) still count. If you want to look at what speeds Po-Dunk can get you'll need to do it on Speedtest.net (and I use Po-Dunk in the most loving way possible). | |
|
iansltx
join:2007-02-19
Golden, CO
kudos:2 Reviews:
·Verizon Online DSL
·Comcast
| Interesting stuff Wonder if the file they're making available for download now is the one they'll restrict to EDU users later. Just pulled it down and all speeds are averages...which is okay, I guess.
Using a couple of grep commands strung together (see the string below) I culled some info from the ginormous city daily speeds CSV; apparently my little town (Fredericksburg, TX) sits around 6M down and 512K up as the average speed. My guess is that (I don't have ISP info to prove this...until they release it later) TWC PowerBoost is helping scores and the WISP in town is hurting them, with a couple DSL results thrown in for good measure.
What would REALLY be nice is a daily list by ISP. Even better: a list that shows the distribution of speed tests; in my town there should be a peak when upload speed equals 512k (standard RoadRunner) and 2M (RoadRunner Turbo), though download PowerBoost wil meld the two together for DL speed. Heck, individual test results would be lovely with unique IDs attached to each IP, but the IPs themselves blocked out (though specificity down to the nearest /24 would be helpful). The thing about narrowing down to the nearest /24 allows me to find, for eample, the local WISP; if I remember correctly they have two /24s and I know which ones they are.
Anyway, the information culling command I sued was:
cat city_daily_speeds.csv | grep "" | grep "" >> customreport.csv
Hope this helps someone  Grep and awk are your friend here; pretty sure any spreadsheet program that you threw this at would choke. | |
|  ansarSearch for HighSpeed
join:2004-12-10
Utica, MS | Dialup and speedtests Heh, do folks on dialup even run speedtests? | |
| | bencPremium
join:2007-06-17
Glen Carbon, IL Reviews:
·Charter
| Re: Dialup and speedtests said by ansar:Heh, do folks on dialup even run speedtests? I've run speed tests on dial-up connections, although the results aren't the least bit surprising. It's how I know that T-Mobile EDGE is about three times faster than dial-up. | |
| | | | I did a few times while on dialup.. It gave some interesting results with PeoplePC accelerated, with the compressed images. I would get over 100k upload speeds. | |
|
 SrsBsns
join:2001-08-30
Oklahoma City, OK | Powerboost How can any of this data be accurate when most of the big names like my provider have Powerboost? According to the speed test I have a 30Mbit line. In reality its 9Mbit. | |
| | ungawd
join:2010-03-18
New Haven, WV | Re: Powerboost It's accurate within reason. Powerboost will affect the first xxMB of data transferred. In my case, and possibly all of Suddenlink's customers, this threshold is 75MB. The average web page is surely smaller than 5MB and well below that threshold. Effectively a user browsing the web will sustain their boosted throughput throughout the duration of their surfing. Online gaming and voice/video communication are restricted more by latency assuming there is sufficient bandwidth. More bandwidth won't make those experiences any more favorable. The only two things that come to mind which will be affected beyond the threshold are substantially large data transfers and streaming HD content with a bit-rate higher than your guaranteed throughput.
Your 9Mbit reality is really just your guaranteed throughput. The 30Mbit you see while testing is a perk you seem to be overlooking. There's no sense in downplaying it because it is very beneficial. As to how much it affects you will be determined by what you predominantly use your internet connection to do. | |
|
 FTTC
join:2008-10-01
Mount Rainier, MD
 ·Verizon FiOS
1 edit | hmm my internet is all ways hit or miss on bandwidth! i hope these studies improves this. verizon blames outside the network. everytime i complain  . as if i call every week for no reason. | |
| | | It all makes sense The data makes perfect sense!
The reason why most attacks come from outside the US is because EVERYONE ELSE IN THE WORLD HAS A BETTER CONNECTION THAN ME.
The actual ironic part is that if we actually, you know, pushed the ISPs we could easily be in the top 10. Instead we allow ISPs to run monopolies while our own government looks the other way and the people get the shaft. What a government.
How many millions do ISPs get in profit every year again? | |
| |  bigfitchPremium
join:2005-06-01
Murrayville, IL | Re: It all makes sense 
Hey don't feel bad. I waited 10+ years for dsl and got it. Your not the only one who is on the bottom of the barrel. | |
|
Mele20Premium
join:2001-06-05
Hilo, HI
kudos:4 | Flash tests are not accurate Flash tests like Speedtest.net are GROSSLY INACCURATE. They always INFLATE the download speed. If researchers want an accurate picture then they should be using Visualware's MySpeed tests. Besides, with crap like PowerBoost that Comcast and Time Warner have there is no way with current speed tests to get an accurate speed. The ONLY accurate test for these cable companies is something like OOL's FTP test. But Visualware's MySpeed provides excellent data on the QOS which is a more important measurement.
--
When governments fear people, there is liberty. When the people fear the government, there is tyranny. Thomas Jefferson | |
| | | Just a matter of time. Since ISPs are using things like "traffic shapers" and other nasty tricks, it's only a matter of time until they are giving priority to things like speedtest.net, and letting the rest of us get by on what's left over.
When they brought out DSL in this area, first it was a "rent your apps online" type of service, with frequently changing internal IPs, PPPOE, and NetWare for all. If the install tech couldn't put the supplied Ethernet card in an ISA slot on your machine, you couldn't get service. A PC at one of the Mac stores must have had it's slots wore out from going home with people for their install day, then back to the shop for the next person. Hundreds of installs on the same PC, because Macs don't have ISA slots.
Later, they decided to offer "High Speed Internet Service". Of course, they were going on about the blazing speed, with big full-page newspaper ads claiming 50 times the speed of dialup. They were also boasting about a dedicated line from their office to your home. One poor client was actually guided by customer support to re-install XP 3 times, and was without service for over 2 weeks, until they discovered that the DSL connection at the "node" didn't have it's ethernet port connected.
More recently, their full-page newspaper ads have changed. It almost sounds the same, but the "50" has been replaced with "15" times the speed of dialup, but don't give up hope, because they have a new "Extreme Speed" package, for only twice the price. Or was that the "Ultra" speed.... marketing confuses me. One of my relatives signed up for their fastest home service, and didn't know that 256Kbps isn't really 15x dialup speeds. Still, they seem to get away with starting the customer fast, and gradually throttling them back until they complain, or leave.
On the Cable side, we've gone from 5Mbps to 15, to a new service level of 30, and a 100, if you have to have the fastest. Oh, and since the phone company started providing TV over IP, the cable company sees it fit to offer phone service too -- too bad they can't get the line current up to handing 1 dial phone plus anything else on the same number, and they don't have anyone who can figure it out. $9.95 a month sounded good for phone service, but since it's more than tripled, VOIP (which is, really, what they are doing anyway) is sounding more and more attractive.
What we really need is something ensuring net neutrality, so that when the customers start running other apps (like VOIP and IPTV), the companies can't just throttle them out of existence, because they offer a competing service. | |
|  Anonymous_AnonymousPremium
join:2004-06-21
127.0.0.1
kudos:2 | any cable co resluts are any cable co results are BOGUS
Power boost easily over inflates the results
so any CABLE co speed test should be NULL/VOID ! | |
|
| |
|
|
·
Thumbs up
