State Government Skewing Internet Test to Favor Telecoms

[Note: This article under the same author originally appeared at Washington State Free Press on August 3rd with the same title. The update at the end of the article is with reference to the original publishing date.]

Sadly, even during the pandemic, when thousands of people are trying to telecommute from home, our officials have been slow to really do anything about improving our broadband access in a meaningful way. Most recently, the Washington State Department of Commerce has pretended to “help” by offering a broadband test on their site. However, the site uses a testing tool from Speedtest.net, whose traffic is prioritized by big telecoms and always produces false positives in favor of the big telecoms. Also, it is not a proper network load test, which means the results are virtually worthless. To make matters even worse, the state bases its access speed on the totally inadequate FCC standard of only 25 Mbits down (megabits per second) and 3 up to appease DSL, wireless, and satellite providers who provide overpriced, inadequate service.

Of course, excellent, free, network load testing tools, like Flent for Linux, are available. These tools will show you that the 100 Mbits down and 10 Mbits up speed Comcast claims you are getting for over $100 a month, will perform, at best, at one-sixth of that speed. The speed is even worse during peak times, since Comcast uses shared network hubs. This makes their connections inadequate for most of the work many of us need to do from home during the pandemic. So please ask the Department of Commerce to design a real network load test. It’s not hard, and the tools are available for free.

We have been using Flent to test connections around town. Comcast uses a “speed burst” technology so your speeds always look good at first, and then are throttled way down. This “speed burst” technology is another reason the minute-long test the DoC uses is totally inappropriate. A decent test takes at least 10 minutes. Please notify the Department of Commerce that you are on to their pro big telecom scam, and you don’t appreciate it. We need real solutions, like public fiber to residences. I have spoken to the state about this issue, but they are always careful not to offend the big telecoms, even during a pandemic. Below is the link to their site; please contact the DoC and ask them to design a real load test instead.

Update: 8/6/2020. Bellingham’s Director of Public Works Eric Johnston wrote to a few of us this morning and admitted the bias in the DoC test and its inaccuracy. He then told us he is going to put the test up on the city’s website anyway. In fact, he encouraged us to support him in using this awful test and to spread the word to others about using it. instead, I offered to develop a proper load test. What is the point in knowingly getting a lot of bad data? How can anyone use bad data to make a real plan?

I smell a corporate welfare scheme in the works. I suspect the city and state will use this bad data to justify giving lots of money to companies like Verizon, instead of building real, permanent, scalable, eco-friendly, and cost-effective infrastructure.

https://www.commerce.wa.gov/building-infrastructure/washington-statewide-broadband-act/speedtestsurvey/

About Jon Humphrey

Citizen Journalist • Bellingham • Member since May 23, 2017

Jon Humphrey is currently a music educator in Bellingham and very active in the community. He also has decades of professional IT experience including everything from support to development. He [...]

Comments by Readers

Dick Conoboy

Aug 08, 2020

Jon,

I go through Comcast.  I tried the speed test tool and got Ping 16, Download 201 and Upload 6.  Then I got an immediate pop up window with this:

“Try Speedtest for Windows

For download speeds over 100 Mbps, we recommend downloading the free Speedtest app for the most accurate testing straight from your computer’s desktop.

 
 
What are we to make of this?
 
Dick

 

Read More...

Jon Humphrey

Aug 08, 2020

Yeah Dick, those are pretty skewed results. The telecoms pretty quickly identify speed-test sites and pritiorize them, buy them, etc. so they can pritiorize the traffic. Only a real network load test, like the rrul test in Flent, some script magic with iperf3, etc. will give you a real result. That is unless you plan on limiting all of your Zoom calls, telemedicine appointments, etc. to less than 1 minute. I’ll send you some pics. from the test results of a Comcast connection, we tested by Geneva elementary school. The truth is though, that most of us are reporting problems during the pandemic with most of our connections. To quote an IT tech I know, “How can Greg Baker be offering online real-time classes when even we, as IT people, have trouble staying connected in Bellingham and the County?!” By the way, the $169.9 million the schools took in for tech funding are more than enough to build a county-wide fiber network that would last 100 years. What is the point of giving kids with bad internet connections shiny laptops? You might as well give those kids shiny bricks. It’s the Digital Divide in action. Poor kids go home to inadequate connections or no connection, and rich kids get “good enough” connections. No one gets excellent connections here because we simply don’t have the fiber infrastructure for it. 

Read More...

Dick Conoboy

Aug 08, 2020

 Jon,

Thanks.  Now I also went to the Flent site at the link you provided.  What I saw there was way beyond my capabilities as a computer user.  Is there any way we as the “great unwashed” can gain access to a site with the Flent tool or its capabilities?

 

Read More...

Tip Johnson

Aug 08, 2020

I kept a spreadsheet of my speeds for years, using speedtest.  I got pings from 0.00 to 217 seconds, downloads from 0.00 to 69.96 mbps, and uploads 0.00 and 11.77.  And spotty service.

I spent untold hours on the phione with Comcast service, during years when all the children had homework deadlines.  I would run around rebooting things with anxious kids waiting, and it was not good for me.  Very unhappy.  One day a Comcast guy was on the pole and I mentioned that my service was sketchy.  He looked over and gave it a tug and the cable came free from it’s fitting on the tap.  It had been up there, loose, depending on the wind, temperature and moisture to make a half-assed connection, failing frequently.

He fixed it, but discussiions with Comcast about the service I paid for and what I got went nowhere.

Service is better with a decent fitting, but the speeds are still not as advertised.  When I bring this up with Comcast representatives, they always ask if I am wired directly to the network, or on WifFi.  I advise them that my Mac doesn’t even have a Cat5 connection, and they reply that I should get an adapter.

I haven’t.  And probably won’t.  I use my computer from a number of different locations, only one of which has a direct network connection. Maybe Elon Musk - or some unbridled regulator - will either introduce competition or establish service as a public utility in order to control costs and guarantee some decent standards.

But probably not.

Read More...

Tim Surratt

Aug 09, 2020

Jon,

I am on Comcast and use a MAC.  I don’t have a Linux platform and have not (yet) opted to be able to boot other OS’s.  My experience with speeds anecdotally suggests the need for a long baseline test.  Is there on available for my Mac?

Thanks

Tim Surratt

Read More...

Jon Humphrey

Aug 09, 2020

Thanks, everyone. I’ll try to answer everyone in order. Before I do I want to direct everyone to a great Whatcom Watch article I was interviewed for that goes over the different internet solutions we have access to and compares them. Basically, it urges us to stop letting big telecom, and City staff, and our officials in general, etc. compare apples (fiber) with oranges (DSL and Cable) and pears (Wireless/Satellite). They are fundamentally different in important ways. In short, even if Comcast, CenturyLink, etc. maintained their connections, which they don’t as Tip highlighted, they’d still be inadequate for most. Here’s the link. https://whatcomwatch.org/index.php/article/untangling-the-wires-understanding-tech-recognizing-fiber/

Dick, the best solution would be for the Port, PUD, County and City, or any combination of that, to fund the development of a proper network load tester with a good web interface. I have offered to do this, but obviously it doesn’t have to be me. It’s not expensive. Until that day… I’m assuming we want something free that’s easy to use. There are tons of great command-line utilities for the more savvy like iperf3. Flent for Linux is the best free solution. It is available for Mac, but the install can be tricky on Mac. https://flent.org/

There are lots of paid-for solutions for Windows, but many of them have trial periods so you can get a basic test done in that time. For example, I tried out the tools from Solarwinds, but they are there complex. It’s free for 14 days.  I’ll try to find something simpler. You can run a basic load test by putting a load on the system, ex. streaming something, do a Zoom call, etc. and then testing your connection but the testers like Flent and WAN KIller just make it easier. https://www.solarwinds.com/engineers-toolset/use-cases/traffic-generator-wan-killer 

Of course, you need a destination to test against, so here is a list of public iperf servers. They get busy though. This is why the Port, PUD, County and/or city, or even WWU agreeing to run a test server, would be very helpful. https://iperf.cc/ 

In a perfect world we’d set up at least 5 test servers. Say Seattle, LA, Chicago, New York, PUD3, South Korea, and somewhere in Europe. How about Berlin? It’s important to test to different locations since Bellingham does business with the world, or rather could if we had good networks. So an LA test, for example, is really important as other networks can degrade the signal too along the way. 

Tip, I responded on Nextdoor months ago about the ridiculous claims that Comcast makes about you needing to go totally wired to use their products. Yes, wireless is less efficient than wired, for several reasons but routers long ago exceeded the dismal speeds most Comcast and CenturyLink users get. For example, the Netgear Nighthawk 7000, sold at Costco in 2016 for years, has a wired speed of a Gigabit (1,000 Mbits) symmetrical. However, it’s wireless AC (higher band 5.8 Ghz) can consistently do at least 600 Mbps if you have a good line of sight, the “older” wireless G radio on it can do 54 Mbps. Since most Comcast users, like me, get a load test of about 16 to 25 Mbps down and 1.5 to 3 Mbps up (on Flent) your router has been capable of exceeding Comcast/CenturyLink speeds for a long time now. The problem is their garbage infrastructure in most cases, not the router. That’s just a lie they tell us to keep the “decision-makers” confused and keep them from investing in infrastructure. Wireless G was released in June, 2003. https://www.netgear.com/about/press-releases/2016/NETGEAR%20INTRODUCES%20NEW%20NIGHTHAWK%20ROUTER%20-%20R7000P.aspx 

So yes Tim and everyone, we need better, longer duration, testing :). We should at least be able to upgrade to 2003 standards by now.   

Read More...

Jon Humphrey

Aug 09, 2020

Good news everyone, I found a free Open-Source tool available for Linux, Windows and Mac called packet sender. I still prefer Flent to it, but it runs on everything. I’ll write up a tutorial soon. It’s much lighter weight than the SolarWinds stuff and is free. https://packetsender.com/

Read More...
To comment, Log In or Register