The way I understand networking and to be fair I may be wrong but can they just not make this connection a priority to skew the figures? Most connections are on a contention ratio anyway.
The FCC is encouraging netizens to use its internet speed mobile app in an effort to finally get accurate broadband data across the United States. In an announcement on Monday, the telecoms regulator noted that “the app provides a way for consumers to test the performance of their mobile and in-home broadband networks” and “ …
Monday 12th April 2021 23:56 GMT Microchip
To an extent. Virgin Media over here are known for it. However, you can't make a 0.5mbps DSL line go faster than 0.5mbit, no matter how much you want to fudge the numbers.
They could give full wireline speed and prioritise the packets, so if congestion is the restricting factor it could give an unrealistically higher reported speed compared to normal Internet usage, but my understanding is that a lot of people are stuck with poor DSL connections. The cable providers are much more likely to be able to fudge figures with their higher available bandwidth over DOCSIS, AIUI.
Tuesday 13th April 2021 08:14 GMT Martin an gof
can they just not make this connection a priority to skew the figures?
"Normal" speed test websites such as speedtest.net work by connecting to specific single servers and they could easily be prioritised, however I suspect this is not the way the SamKnows app works.
Note that I have no specific information about this app, but I do have a SamKnows whitebox (albeit a very old one) at home (UK), and that gives all sorts of useful information about my connection by testing "real world" usage - it makes connections to sites such as Netflix, YouTube, iPlayer, Fortnite, Facebook, Snapchat, Meet, Zoom, Teams, various CDNs etc. etc. It tests web connections, TCP and UDP routes, DNS, latency, jitter, sustained speeds and much, much more. It tests at regular intervals throughout the day so it will catch issues caused by contention at peak periods, and it is intelligent enough to attempt to make its tests when your own network is quiet. It doesn't have access to your actual data (just port counters - though I think my own ancient box is slightly different) but it will not try to start a test, for example, if it sees that your connection is busy - you might be streaming or video conferencing or downloading the latest Suse ISO.
SamKnows is touting their SDK as an add-on for ISP-provided equipment, but there is no way US carriers are going to submit to that willingly, so perhaps the best solution would be for the FCC to fund the (relatively low) cost of rolling out a couple of hundred thousand stand-alone Whiteboxes to anyone who wants one?
There's a bit more information here but you would do well to browse around the SamKnows website.
While the app is very easy to recommend and persuade people to use, real data would come if people in the US had access to something similar to the Whitebox, which is permanently installed in your network.
Tuesday 13th April 2021 14:02 GMT Timo
Our house had a SamKnows "whitebox" for years (we're left-pondian), happily running speed tests and checking all those endpoints that you mention, and provided a pretty decent dashboard of test results. At one point I was able to use the SamKnows data and charts to show our ISP that their connection was failing and causing retries and terrible performance.
The whitebox that we were given could only handle fast ethernet, so when we upgraded to >100 Mbps cable it couldn't test the full throughput. At that time the response from SamKnows was to remove the box and sign up for a future revision of the whitebox, time TBD. That was probably at least 5 years ago.
Cheers to The Reg because I think I heard about SamKnows here, and signed up for the service. The only downside was that the continuous testing consumed a relatively huge amount of data each month.
Tuesday 13th April 2021 16:09 GMT Timo
Re: Samknows whitebox
in fact I found the email from them (it was back in 2018)
The Whitebox you currently have is one of our older models and it caps at around 92Mbps. Which means that your test results will be skewed and your internet speed will also be limited to 92-94Mbps therefore, we don't recommend using it at this point.
We're not sending new Whiteboxes at the moment, but we hope this change soon. Therefore, we recommend re-registering for a new Whitebox here and we will keep your details on our waiting list and in an orderly fashion.
You don't need to return the Whitebox, we only advise to dispose of it in an environmentally friendly way.
Alternatively, you can use our free SamKnows App which is available on both Google Play and the App Store. Our app measures download, upload, latency, YouTube and web browsing. It also generates monthly graphs that you can share.
We really appreciate all your time with us. We’ll remove your details from our system so you can re-register using the same details again.
Tuesday 13th April 2021 00:28 GMT Shadow Systems
There's a problem with the app...
Specificly it's only for SmartPhones & not desktop computers so the only thing it measures is *mobile* network speeds. I'd love to grab a copy & run it so the FCC could see just how shitty Comcast is, but since I only use a FeaturePhone & a desktop computer, the app is a non-starter.
I went to the FCC site & said as much in the comments section of the form, but I've not heard back yet -- not even the auto robo "Thank you for your submission" reply. I suggested a desktop program to do the same job for those of us with only a desktop by which to access the internet. I'm hoping they create one soon.
Dear Comcast, you can claim I'm getting "nearly 30Mbps" all you like, but when a simultaneous, real world download of a Linux ISO only registers at TwoPointEightFour (2.84) Mbps from the fastest server nearest me, your speed test is dubious at best & utter shite at worst. I've spent the last three days trying to find *any* server that can send me a file as fast as you claim I'm connected, but nobody has been able to pass the 3.5 Mbps mark. So the IE11 download manager is flawed or your speed test app must be. Hummm, I wonder which is more likely?
I hope the FCC grinds your scummy arse into mulch.
Tuesday 13th April 2021 06:33 GMT ghp
Tuesday 13th April 2021 06:45 GMT Sgt_Oddball
Re: There's a problem with the app...
You could use something like bluestacks and run an Android device as a vm. I admit it's a pain, but if you really want to add your stats to the pot it's an option.
It might also be an idea to use something more modern than IE11 so you could take advantage of things like HTTP2 to see if that gives you a performance increase (owing to more efficient data transfer methods).
Tuesday 13th April 2021 18:08 GMT Shadow Systems
At Oddball, re: BlueStacks...
I didn't know they made Android emulators for Windows, so thanks for that. It'll come in handy...
*Hands you a pint & taps rims in toast*
...if the site loads properly in something other than my IE11. I've got Win7 MS Edge (pre switch to pure Chromium) & FF ESR v82(?) I can try, but they're much less user friendly to my screen reader. It might be different if I were using Win10 & even more "modern" versions of those same browsers, but I'm not (I'm on Win7Pro64 for screen reader compatability requirements) so I'm up against a proverbial wall.
*Sighs, smiles, & refills your pint*
Thanks for the tip, I'll try to use a different browser to get the page to render with a download link. =-J
Tuesday 13th April 2021 07:52 GMT Martin an gof
Re: There's a problem with the app...
Specificly it's only for SmartPhones & not desktop computers so the only thing it measures is *mobile* network speeds.
Nope, it will measure whatever network it is connected to. If you are at home and connected to your home WiFi (you could turn off mobile data to be certain) it will measure your home connection's speed. Admittedly there is the slight variability of the quality of your own WiFi, but it should give you an idea.
The information is all there in TOA, along with the link to the FCC's page which states very clearly
You can test the performance of your mobile and in-home broadband networks by downloading the FCC’s Speed Test Application on your mobile device(s)
Tuesday 13th April 2021 09:28 GMT LDS
"it will measure your home connection's speed"
No, it will measure the WiFi connection speed that may be lower than the internet connection speed - depending on your mobile phone wifi, your router wifi and nearby wifi networks. Of course if you have a slow network it may not be an issue - as soon as the speed increase it could become.
The only way to proper measure the internet connection speed is using a PC connected to the router through an Ethernet cable - and the Ethernet has to have a speed equal or higher than the internet claimed speed, of course.
Tuesday 13th April 2021 18:03 GMT DS999
Re: "it will measure your home connection's speed"
Given that this is intended to point to where broadband is deficient I don't see the issue. Unless someone still has an old 802.11b only router it won't limit their speed below the 25 Mbps FCC broadband threshold.
I'll bet there are more than a few people who don't have any wired PCs in their house - my mom comes to mind as the desk she wants her PC isn't where the phone plug is for her DSL so I set it up with a wireless dongle.
Adding a Windows and Mac version of the app (or website link) would seem to be a pretty simple thing for the FCC to do though.
I would hope the app is smart enough to detect whether it is running over cellular and not let people claim that's their Comcast speed or whatever, but given how many crappy apps are out there I guess I wouldn't be shocked if is as dumb as all that.
Thursday 15th April 2021 00:47 GMT Michael Wojcik
Tuesday 13th April 2021 18:21 GMT Shadow Systems
At Martin, re: the app...
1. It requires a SmartPhone. I don't own a SmartPhone. I own a FeaturePhone. It doesn't run apps, therefor I can't run them either. I need a desktop version to run on my Win7Pro64 desktop.
2. I don't have anything other than said FeaturePhone that uses Bluetooth, and I've turned that function of the phone off to save battery & close that security hole.
3. I don't use WiFi at all. Nada, nothing, none, zip, zilch, bupkiss.
So even if I (1) had a SmartPhone to (2) run the app, it would be inherently limited by the maximum speed of (3) the WiFi connection. That is not the same as my physicly connected via CAT5, gigabit LAN port to the computer, gigabit LAN port to my cable modem supposedly ~30Mbps Comcast connection.
*Hands you a pint*
Drink up. It's more fun than dealing with a rabbid badger in your knickers. =-)p
Tuesday 13th April 2021 20:40 GMT Martin an gof
Re: At Martin, re: the app...
Thanks for the pointers. I should have remembered you don't run a smartphone, but to be fair that wasn't actually your complaint. You stated
Specificly it's only for SmartPhones & not desktop computers so the only thing it measures is *mobile* network speeds.You are correct that the app is only for smartphones, but you are not correct that it only measures mobile network speeds - that is, the cellular network. However, that you can measure the mobile network too is a bonus.
As others have pointed out, the fact that it has to measure your wired network over WiFi could limit its usefulness, but only in the case where that connection is faster than your WiFi.
Even "g" WiFi can sustain over 50Mbps, which is faster than you say your Comcast connection is supposed to run, so long as you have a reasonable signal, your neighbour isn't on the same channel and you aren't trying to do the speed test from the end of your back garden when the access point is next to your front door. More modern versions of WiFi can run a lot faster, though keeping up with 1Gbps broadband via WiFi is unlikely, and WiFi will also add some jitter, latency and retransmissions that aren't the fault of the wired network, but it's better than nothing!
The current version of the Samknows Whitebox only works up to 1Gbps and for speeds beyond that they recommend loading their SDK onto the ISP-supplied modem. Apparently BT and Virgin in the UK do do this, but from what I have heard about operators in the US I'd say it's highly unlikely they would agree, or if they did that they would make the data available.
The point is - I think - that the FCC has realised it has an almost complete lack of information with which to point out the flaws in the country's broadband availability and that it is not in the operators' interests to help them get that data. Users downloading and running this app costs the FCC next to nothing and is an excellent example of crowdsourced data. It may need some work doing on the figures to make them robust, but at the very least it will give the FCC good information about where the current problems are so that they can concentrate their meagre resources on those areas first.
This post has been deleted by its author
Wednesday 14th April 2021 10:01 GMT Boothy
Re: There's a problem with the app...
SamKnows also have a browser based one, that works fine on a Desktop. Don't think it's related to the FCC, as they don't provide any background info on the site. But it's a lot less cluttered and cleaner than the speedtest.net equivalent, and seems to produce similar results for me on the same connection.
Tuesday 20th April 2021 20:11 GMT ABehrens
Tuesday 13th April 2021 07:46 GMT Mike 137
"Big Cable has aggressively – and successfully – argued in the past that data provided by users over an app is not sufficiently robust ..."
We have "36Mb" FTTC here in the urban UK, but consistently average 6-8Mb/s actual download speed across a wide variety of services (so it's not down to a low server somewhere). However Openretch classify performance solely by sync rate, which is demonstrably up to speed. the problem (clearly shown in wireshark at this end) is packet corruption. We persuaded an engineer to investigate this and he found a 1970s "dry" crimp full of water in the copper run from the cabinet. he came back and reported that he had fixed around 50 others in the same junction box in a similar condition.
As the copper infrastructure is now deemed "legacy" it's not being properly maintained while we wait to be offered FTTP at a whacked up price. However, as the "broadband" component of our obligatory "all in" connectivity package amounts to only about 8% of the monthly cost, the opportunity for compensation is effectively nil.
Tuesday 13th April 2021 09:12 GMT Kevin Johnston
Excuse my ignorance but...
I am probably about to drown in downvotes but is it not the case where the ISP quoted figure is megabits while the download tracker is megabytes? Not to say this magically brings the observed download speed in synch with the ISP's quoted but it would explain getting around a tenth of the quoted speed.
Tuesday 13th April 2021 09:56 GMT Return To Sender
Tuesday 13th April 2021 13:08 GMT Jellied Eel
Re: Excuse my ignorance but...
Not to say this magically brings the observed download speed in synch with the ISP's quoted but it would explain getting around a tenth of the quoted speed.
It can happen. From a networking perspective, speeds should always be in bits per second, ie 30Mbps. I've seen some get it wrong, ie Mbps instead of MBps when they're talking about bytes. Then life gets more complex sometimes trying to figure out overheads. One standard test is to try a file transfer with FTP, then UDP. UDP should be 'faster' as there's less overhead. Unless there's congestion & packet loss, in which case UDP can be slower. Then there's fun with TCP and latency, where the higher the latency, the more wait time there can be due to SYN/ACK delays.
Tuesday 13th April 2021 10:27 GMT ROT13
Tuesday 13th April 2021 13:11 GMT Jellied Eel
It's better than nothing. Contractually though, contracts for 'up to' really only honor within the ISP's clouds as it's not really possible to guarantee anything beyond that. So speed tests that go beyond that ISP's cloud tend to end up measuring problems with peering/transit links. For ISPs though, that kind of thing is still useful data for managing those links, or trying to.
Tuesday 13th April 2021 23:38 GMT hayzoos
Simple solution to that. Do all your internet in your ISP's cloud then. The only things that will seem slow are, ah wait a minute. . .
Most of what I use is not on my ISP's network. I'm pretty sure I only visit my ISP's network to obtain my "paperless" bill. And, because I use a VPN because I don't trust my ISP's penchant to slurp (and I travel for work), I visit their network from outside their network. My speed tests are skewed because of that on a number of factors.
Tuesday 13th April 2021 18:27 GMT LordHighFixer
Rural and Lucky
I live in the middle of nowhere, population 200, nearest 'town' is a 15 minute drive, nearest 'city' 45 minutes. For me it was satellite or terrestrial microwave for a long time. Welcome to 10Mbps on a good day with between a 90-600ms ping. That is what passes for competition and 'broadband' for most of rural America. Forget reliable cell data service.
Luckily, I happen to live close enough to a highway, that when they laid a new fiber cable, they offered FTTP to everyone along the way. I now have the minimum service that anyone should have 1G SYNCHRONOUS FTTP. And it tests at the theoretical max. 9ms ping. Anything less is unacceptable.
Tuesday 13th April 2021 19:09 GMT Anonymous Coward
"For years, Big Cable has been manipulating the data it is obliged to send the FCC over the speeds and provision of its broadband services ..."
Like so many other industries and businesses in the US, such as pharma and chemical co's, those companies are charged with preparing their own data upon which final judgements are made. Well that's just downright dumb. No balance there.
Wednesday 14th April 2021 03:10 GMT Balding Greybeard
FCC App Gives An Honest View Performance
I live in a fringe urban area and the best internet service package I can get is AT&T 25 Mbs. The service was great for the first four months of the subscription; lately I am experiencing lots of dropped Zoom calls. I run AT&T’s speed test and the speeds reported are at or above what I subscribed to. The FCC app tells another story; this afternoon after a Zoom drop, the FCC app showed 4.26 Mbs with a latency of 44.82 ms.
Finally I have proof to say to AT&T, “liar, liar, pants on fire”.
Thank you FCC!!!
Wednesday 14th April 2021 10:25 GMT Jonjonz
The latest ISP scam
The latest ISP scam is that in the last year or so they all have been adding fine print to the extent that they don't guarantee any usable speed for Wi-Fi.
When 80% of internet users use Wi-Fi between their router and their home device, bragging about 100 mps is totally misleading as the Wi-Fi is now throttled down to 25 or less mps.
The FCC needs to count last leg Wi-Fi in it's measurements.
Thursday 15th April 2021 01:05 GMT Anonymous Coward
Re: The latest ISP scam
I hate to defend scummy ISPs, but you kind of have to throw your hands up with WiFi.
Sure, your average Reg reader may have a router with a dozen antennas and a standards compatibility list that looks like someone spilled alphabet soup. Meanwhile there are people running a 12 year old Netgear router trying to support a dozen iThings of various ages while the neighbors routers are blasting enough signal to pop popcorn.
I'm just old and cranky... in my day, a real speed test was tftping a file from your router to your ISP's router. A Cisco IOS file was a convenient size for a test (and was typically available on the device). If you did that often enough and were observant enough, you could snag an image with a few more bells and whistles than the one you were currently running. Thank you to whoever uploaded the fancier images to the Cable and Wireless routers I happened to frequent.