how-to block ads
|
Uniqs:
1135 |
Share Topic
 |
 |
|
|
 DogfatherPremium
join:2007-12-26
Laguna Hills, CA
4 edits | Data is inaccurate representation, but shows 17% improvement Every time I ran a test there using my FiOS service (with both 50Mb and 20Mb tiers and my Cox service which typically gives me 20-25Mb results elsewhere) it would report 1/2 of the speeds I typically see even from distant servers and about 1/3 that local high capacity speed test sites report.
And you can't draw any conclusions from this other than their speed test server capacity is sorely lacking.
Meanwhile this time last year »Average U.S. Upstream Speed: 371kbps they reported upload speeds being 371kbps so even if we took these numbers at face value (which we obviously can't) it represents a 17% improvement.
And since when are people using residential upload speeds to send gobs of medical records? You think hospitals have 1500/384 DSL from Verizon? | |
|  | | Re: Data is inaccurate representation, but shows 17% improvement The union has their own special interests in reporting skewed data - like having gov't pour in billions of dollars in taxpayer money to subsidize building out broadband infrastructure, resulting in adding a lot of union jobs. It wouldn't be surprising that the system they set up was designed to lowball the actual numbers.
--
My BLOG .. .. Internet News .. .. My Web Page
Ask yourself one question: 'Do I feel lucky?' Well, do ya punk? | |
| |  en102Canadian, eh?
join:2001-01-26
Valencia, CA | Re: Data is inaccurate representation, but shows 17% improvement Exactly. I'm typically a moderate, however, I see this as a pure push for taxpayer expenditures where it isn't really needed.
- This site takes a representation from users stats. Not all (not many) customers order +10Mbps
- Not many residential customers care whether they have 10Mbps or not
- Most would probably like to see... lower taxes, and lower cost of service. I know that I do.
--
Canada = Hollywood North | |
| |  funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:5 | said by fAcEtIOUs:The union has their own special interests in reporting skewed data Then let's see the "real" numbers, TK, from a source with no interest in skewing them.
(There aren't any.)
And what the CWA is advocating is to pass a law funding the collection of this data so that we can get an accurate picture.
But rather than attack the CWA or attacking the bill, shouldn't we at BroadBandReports be analyzing the numbers? Isn't that our best capability here? Doesn't this site have speedtest data out the ass?
--
Robb Topolski -= funchords.com =- Hillsboro, Oregon
More fun, more features, Join BroadbandReports.com, it's free...
| |
| | | DogfatherPremium
join:2007-12-26
Laguna Hills, CA | Re: Data is inaccurate representation, but shows 17% improvement The CWAs numbers are largely worthless. DSL Reports's samples are limited in scope. Speedtest.net has test servers all over the world and certainly wouldn't be biased. It's not a perfect sample, but would be about as close as we're gonna get for comparing. I would certainly trust their sampling over the CWA who obviously has a vested interest in portraying doom and gloom. | |
| | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:5 | Re: Data is inaccurate representation, but shows 17% improvement said by Dogfather:DSL Reports's samples are limited in scope. Speedtest.net has test servers all over the world and certainly wouldn't be biased. If we're talking about the USA, DSLReports has test servers all over, too.
said by Dogfather:I would certainly trust their sampling over the CWA who obviously has a vested interest in portraying doom and gloom. And they have a vested interest in being accurate. They lobby policymakers. Let's say they were caught lying. Doesn't that devalue their stock in trade?
--
Robb Topolski -= funchords.com =- Hillsboro, Oregon
More fun, more features, Join BroadbandReports.com, it's free...
| |
| | | | | DogfatherPremium
join:2007-12-26
Laguna Hills, CA
1 edit | Re: Data is inaccurate representation, but shows 17% improvement With DSLR the sampling is too small. And with Speedtest.net they're sampling methods are perfect for determining maximum line speed.
Lobbyists caught lying...certainly wouldn't be the first time. Propaganda is their business. | |
|
| | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:5 | Re: Data is inaccurate representation, but shows 17% improvement said by fAcEtIOUs:How about BBR? You can't say they have a bias FOR the ISPs. » /archive/The numbers here show the union info is skewed low. [att=1] You took the top 29 domains and compared it to (what I think CWA says is) a nationwide average.
I haven't read the CWA report -- I guess that I should -- but I don't think the "top 29" list proves or disproves anything.
--
Robb Topolski -= funchords.com =- Hillsboro, Oregon
More fun, more features, Join BroadbandReports.com, it's free...
| |
| | | | | | | Re: Data is inaccurate representation, but shows 17% improvement If you look at the BBR data and only check the biggest ISPs with the MAJORITY of the users, you will see that it comes in higher than the CWA data. | |
| | | | | | DogfatherPremium
join:2007-12-26
Laguna Hills, CA | Re: Data is inaccurate representation, but shows 17% improvement By design or not, the CWA data collection methods are seriously flawed thus any conclusions drawn from them also flawed.
It's not logical to stick with the CWA report simply because there is no other report. Bad data isn't better than no data. I would argue that it's worse since people come to the wrong conclusions and make bad assumptions based on bad data. | |
| | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:5 | Re: Data is inaccurate representation, but shows 17% improvement said by Dogfather:By design or not, the CWA data collection methods are seriously flawed thus any conclusions drawn from them also flawed. You're not being scientific about this. What is the flaw?
said by Dogfather:It's not logical to stick with the CWA report simply because there is no other report. Agreed, but nor is it logical to dismiss a report based solely on its issuer.
Again, I haven't read it. Maybe it's as bad as you say -- but since you've taken a position on it ("the CWA data collection methods are seriously flawed"), you need to explain that!
--
Robb Topolski -= funchords.com =- Hillsboro, Oregon
More fun, more features, Join BroadbandReports.com, it's free...
| |
| | | | | | | | | | Re: Data is inaccurate representation, but shows 17% improvement said by funchords:said by Dogfather:It's not logical to stick with the CWA report simply because there is no other report. Agreed, but nor is it logical to dismiss a report based solely on its issuer. But BBR has dismissed all the data supplied by Bell Canada & other ISPs all the time because it is claimed they are biased. But somehow we should believe CWA data as unbiased. Why?
--
My BLOG .. .. Internet News .. .. My Web Page
Ask yourself one question: 'Do I feel lucky?' Well, do ya punk? | |
| | | | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:5 | Re: Data is inaccurate representation, but shows 17% improvement said by fAcEtIOUs:But BBR has dismissed all the data supplied by Bell Canada & other ISPs all the time because it is claimed they are biased. But somehow we should believe CWA data as unbiased. Why? BBR isn't the standard maker, either. Do two wrongs make a right all of a sudden?
I've now read the CWA report. It's too light on the technical details to be reproducible and I shouldn't have to reverse engineer the speed test to figure out how it works. It is definitely written for the end-user or policymaker who colors with the 8-crayon box.
That said, if we can make the assumption that they used the same method in 2007 that they did this year, the results are still useful for comparison purposes in the aggregate. But even here in "the silicon forest" heavy in tech workers, there were only 10 tests conducted from my zip code.
I wouldn't draw any conclusions from this study alone, including the position that the study itself is skewed.
--
Robb Topolski -= funchords.com =- Hillsboro, Oregon
More fun, more features, Join BroadbandReports.com, it's free...
| |
|
| | | | | | | DogfatherPremium
join:2007-12-26
Laguna Hills, CA
4 edits | Scientific in that the CWA test site reported 33-50% of my actual line speed and it got worse when my higher tier connections like my old 30Mb and 50Mb FiOS tier. It's a big assumption but I'm assuming I'm not the only one getting erroneous results. Plus they're looking at what people are buying, not maximum line speed. If people buy 1Mb service but 10Mb is available to them, in a best case scenario, the CWA would average that user at only 1Mb. That is a reflection of users being "cheap", not the state of broadband (what's available to them).
And it's these flawed results that has me questioning the report. The fact that it's the CWA may be a factor in them not really caring to fix it. It's certainly in their best interest to have the results looking bad.
You can't draw any conclusion from such seriously flawed data. We can argue all day not whether or not these flaws are by design, but it's not arguable that the flaws exist.
Then the question is, where do we get good data, and my argument is that Speedtest.net would be a good source of such data because their data collection practices and computational processes are designed (according to the methodology stated in the FAQ) to compute maximum line speed.
Speedtest.net, has a HUGE sample population and while not perfect, takes only the highest result from any IP, and then takes the 95th percentile to compute the maximum throughput for a provider or locale. So the frugal buyers don't skew the "state of broadband" down because they choose to buy a lower tier.
And if we then trust Speedtest.net's data enough to draw a fair conclusion from it, it states that US broadband isn't horrible. The US is typical for industrialized nations on par with many European nations. | |
| | | | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:5 | Re: Data is inaccurate representation, but shows 17% improvement You bring up some good reasons this time -- thanks, that's what I was looking for.
As for inaccurately testing your super-fast connection, the CWA report uses median speeds so that fault should have no effect. However, using median speeds and having gone through their test myself (very few instructions or other data collection) -- again, perhaps useful in aggregate. I believe the 17% number, that "feels" right. It's plausible. And while someone thinks that is an "improvement," it may simply mean that we're just falling further behind. The number should be around 40% if we're progressing like bandwidth speeds usually do.
--
Robb Topolski -= funchords.com =- Hillsboro, Oregon
More fun, more features, Join BroadbandReports.com, it's free...
| |
| | | | | | | | | | DogfatherPremium
join:2007-12-26
Laguna Hills, CA | Re: Data is inaccurate representation, but shows 17% improvement My point is that the new bad data reflects a 17% improvement over the old bad data. Because the data is flawed no one can make a single assumption or conclusion about the state of broadband from the CWA's numbers.
We have to go to another source and I would prefer an unbiased source. | |
| | | | | | | | | | funchordsHelloPremium,MVM
join:2001-03-11
Yarmouth Port, MA
kudos:5 | Re: Data is inaccurate representation, but shows 17% improvement Me too, and that's the whole point that the CWA is trying to make -- someone needs to collect this data and find out where we are. The 'free market' isn't taking care of it.
This is a communications union, they work on wages, hours, and working conditions. And while this study might help them hold the threat of releasing "bad news" over the bargaining table -- they've released it.
But again, we really don't know much about the test itself. It's valuable for its anecdotal value. Maybe after ten years and correlation to other data points, we will be able to trust it more than we can now. But their point is a policy point:
CWA says they want someone watching the US broadband market. The FCC can't even accurately talk about broadband penetration or speeds. Talking to researchers who cannot get info out of Comcast, AT&T, Verizon, etc., it is something that is sorely needed for them as well.
--
Robb Topolski -= funchords.com =- Hillsboro, Oregon
More fun, more features, Join BroadbandReports.com, it's free...
| |
| | | | | | | | | | DogfatherPremium
join:2007-12-26
Laguna Hills, CA | Re: Data is inaccurate representation, but shows 17% improvement Aside from the bloodsuckers capping, I don't see this big broadband problem people are alluding to. There are much more important things needing some watching over. | |
|
| |
|
