NY Attorney General Wants Public To Report Broadband Speeds

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
If you are a resident of New York, the state government wants your help going after large internet service providers that are not delivering advertised speeds.

New York Attorney General Eric Schneiderman invited the public on Sunday to test the speed of their Internet and submit the results online as part of an ongoing probe into whether large providers may be short-changing customers with slower-than-advertised speeds. The office launched an investigation into Verizon Communications Inc, Cablevision Systems Corp and Time Warner Cable Inc in October over the issue.
 
No way this will be massively abused or not understood by the general public.
 
IMO the ridiculous prices, data caps and monopolies are the issue, not the speeds that aren't quite what was advertised.
 
The issue here is testing speeds with nothing else connected. The average person will run over to their local speed test website without acknowledging that they have 2 cell phones, a few tablets, a media server and a bunch of other junk connected to their internet during the test.

If I disconnect everything I will get the advertised 55mbps.. if I leave it all hooked up I see around 30mbps.
 
With all the ISPs I've done business with over the years I only had this problem once. The ISP had an overloaded DSLAM and they fixed the issue when I reported it. They even credited the service on my bill each month until it was fixed.
 
One thing about TWC in my area, is that they give a bit more than they advertise. I'm currently paying for 200 down and rarely get lower than 225 and I've seen as high as 250.
 
Because it is not like ISP's can not set priority to pretty much every speed test site, like i believe they already do....

And if not already you know those ISP are setting up cache systems to provide the fastest possible results to it's users...
 
I can have everything on in the house along with my oldest on her xbox, updating a PC and testing my connection and I get above my advertised speeds of 150 down and 20 up.. ok I may get around 16 or 17 on the up but close enough for me Then again I have a well managed network

AG needs to get with the providers and run surprise tests directly at the cable companies tap outside said house. Disconnect house run test and certify.

Then they can move to the unlimited clauses everyone has
 
auntjemima said:
The issue here is testing speeds with nothing else connected. The average person will run over to their local speed test website without acknowledging that they have 2 cell phones, a few tablets, a media server and a bunch of other junk connected to their internet during the test.

If I disconnect everything I will get the advertised 55mbps.. if I leave it all hooked up I see around 30mbps.
Click to expand...

Bingo! people are going to complain like no other when their 10 kids all try this at the same time while streaming youtube, online gaming and netflix to their homes...
 
auntjemima said:
The issue here is testing speeds with nothing else connected. The average person will run over to their local speed test website without acknowledging that they have 2 cell phones, a few tablets, a media server and a bunch of other junk connected to their internet during the test.

If I disconnect everything I will get the advertised 55mbps.. if I leave it all hooked up I see around 30mbps.
Click to expand...

THIS^^^^ fucking idiots. Have a customer that runs a program that does a speed test of their connection every 5 minutes and then complains that the numbers aren't constant at what they are advertised to be getting. Had to explain that when you have a 20Mbps connection and run 2 Netflix streams and a lot of other traffic you won't see 20Mbps on a speed test during that time.

And this guy works IT. I feel horrible for the people he supports :(
 
You guys all talking crap about the "end user" doing test and not knowing anything, I guess you all probably use speedtest to do your test as well right?

Much more accurate using testmy net to get a more reliable speed rather thent he heavily optimized speedtest that tries its best to give ISP's a more favorable speed then what you might actually be getting.
 
Whether you use a "rigged" testing site or a program or a non-rigged site, having shit tons of bandwidth hogging applications running when doing your test will give you false results.
 
Im a broadband specialist for a cable company where i live, disregarding when there is cable issue's noise / utilization and what not, most slow speeds complaints are user error or fault of there own for not understanding there own equipment, for example wi-fi speeds vary from device to device depending on the device it self, for customers that superscribe to our 150 down 20 up service, many do not understand that you need cat 5e as cat5 is limited to 100mb, as well as the nic in the computer it self, some nics are only 100mb while others are gigabit which you need to pull anything over 100. In my personal home i have a asus AC router, my samsung S5 can pull 145mb over wi-fi on my 5ghz channel, my nvidia shield can only get 95mb on 5ghz, and my old nexus 7 tablet would never pull higher then 35mb /sec even when the link speeds was maxed at 72. like i said it varys from device to device when testing speeds.

BTW the worst offender for speedstest in my experiance is the playstation 3 and 4 built in speed test, its terrible and alwase shows poor results, we get more trouble calls about slow speeds because of the playstation systems then anything thing else
 
i disagree speedtest.net is pretty accurate, using multi streams to maximize bandwidth is no different then newsgroups do to achieve fast download ( which use multiable connections to maximize download speed ) if a website is slow say half your bandwidth that has nothing to do with your connection, go start another download from somewhere else and see if the combined speeds can maximize your connection
 
What Schneiderman (and subsequently his test) don't understand is that the connection you pay for as a customer only covers your connection speed from your device (controlling for wifi and multiple devices, etc) to your ISP's network. Doing a test to servers OUTSIDE of their network they aren't responsible for, and shouldn't be. Having a test that traverses multiple networks or tells you that your speed to Google or some other server outside of the ISP is useless.
 
Stiler said:
You guys all talking crap about the "end user" doing test and not knowing anything, I guess you all probably use speedtest to do your test as well right?

Much more accurate using testmy net to get a more reliable speed rather thent he heavily optimized speedtest that tries its best to give ISP's a more favorable speed then what you might actually be getting.
Click to expand...

Point being most people have never heard of the site you mentioned (heck i never have) and will be going to speedtest.net for their testing..

You forget most end users are not [H] type Computer people what so ever, you give far too much credit to the general public's I.T knowledge.

Sure if you surveyed random people and asked them to do a speedtest they would reply with
"a what... you want me to try and run as fast as i can?
 
auntjemima said:
The issue here is testing speeds with nothing else connected. The average person will run over to their local speed test website without acknowledging that they have 2 cell phones, a few tablets, a media server and a bunch of other junk connected to their internet during the test.

If I disconnect everything I will get the advertised 55mbps.. if I leave it all hooked up I see around 30mbps.
Click to expand...

Also an issue can be the other end of the test. Some of the speedtest sites are crap, others are great. The faster your connection is, the more of them that can't actually test it accurately.
 
Sycraft said:
Also an issue can be the other end of the test. Some of the speedtest sites are crap, others are great. The faster your connection is, the more of them that can't actually test it accurately.
Click to expand...

Completely true. Our sales guy tries to use speedtest.net to show customers how fast their fiber to the home is and I want to slap him for it as it can't handle fast speeds with low latency.
 
Do they provide a link to a speedtest sight that doesn't automatically favor and give priority to, i.e. Ookla?
 
Dregan73 said:
speedtest.net works just fine for giving accurate results

https://youtu.be/_Y9rrxzG3Mw
Click to expand...

Depends on the server used and the speed of your connection for example here are two from my connection:

4916211779.png


4916216302.png


Clearly one of those is wrong, in particular, the second one. That Speedtest server just has inadequate bandwidth to test my connection. Either their server itself has low bandwidth or is overloaded, or there's not a good link between them and my ISP. Either way, it isn't a reliable result.

Speedtest is useful, but you have to know something about expected results and what servers are good to use with a given ISP.

Others can be ever worse. That testmy.net place that another guy talked about is crap. They have very poor bandwidth to any place I've tested (like my work which has a massive connection) so the results are artificially low.

DSLreports new test seems to be the most accurate for high bandwidth connections I've yet found, but you still have to be careful.
 
Sycraft said:
Depends on the server used and the speed of your connection for example here are two from my connection:

4916211779.png


4916216302.png


Clearly one of those is wrong, in particular, the second one. That Speedtest server just has inadequate bandwidth to test my connection. Either their server itself has low bandwidth or is overloaded, or there's not a good link between them and my ISP. Either way, it isn't a reliable result.

Speedtest is useful, but you have to know something about expected results and what servers are good to use with a given ISP.

Others can be ever worse. That testmy.net place that another guy talked about is crap. They have very poor bandwidth to any place I've tested (like my work which has a massive connection) so the results are artificially low.

DSLreports new test seems to be the most accurate for high bandwidth connections I've yet found, but you still have to be careful.
Click to expand...

which one of those two does speedtest run by default?
 
mnewxcv said:
which one of those two does speedtest run by default?
Click to expand...

My guess would be the first one as it finds the one with the lowest ping. That is where I notice my issues. When you get down to 1 or 2 ms ping time to the server it seems to have trouble with processing so quickly.
 
If you think the web-based speedtest.net test is giving you accurate results, you may want to give the command-line tool a try.

P.S.: Traffic-shaping. Know what that is? It's how ISPs choose what to throttle... and what not to throttle.
 
:eek: I personally just see what kind of download speed I get from Steam, the stupid update is quite good saturating your connection to the point where you can't do anything.

I get about 4.8MB/s right now on a 30mb TWC connection. I miss the 40-45MB/s I was getting with my 400mbps Grande before I moved.
 
The ISP's have two words on their side. It's in every contract:

"Up To"

Not taking their side, but short of flagrant lying on the part of an ISP, this sounds like a big waste of taxpayer dollars.
 
Stiler said:
You guys all talking crap about the "end user" doing test and not knowing anything, I guess you all probably use speedtest to do your test as well right?

Much more accurate using testmy net to get a more reliable speed rather thent he heavily optimized speedtest that tries its best to give ISP's a more favorable speed then what you might actually be getting.
Click to expand...

Going to disagree with you, just tried testmy net and it was off by a fair bit (as in much slower). Speed test is way more accurate compared to what I actually get for download speeds from things like steam or blizzard games.
 
auntjemima said:
The issue here is testing speeds with nothing else connected. The average person will run over to their local speed test website without acknowledging that they have 2 cell phones, a few tablets, a media server and a bunch of other junk connected to their internet during the test.

If I disconnect everything I will get the advertised 55mbps.. if I leave it all hooked up I see around 30mbps.
Click to expand...

Its 55mbps per household, not per device. Also keep in mind, your router could also be a problem with that many devices connected.

As an aside, I've never had a major problem with Cablevision. Pay for 50 mbps, get 57 most of the time. Only major issues are that, because we are buried, one of the cable lines sometimes gets flooded, resulting in latency spikes. Isn't a problem except when we get heavy snowfall though; though I wish they could find whatever connector isn't waterproofed right.
 
evilsofa said:
If you think the web-based speedtest.net test is giving you accurate results, you may want to give the command-line tool a try.

P.S.: Traffic-shaping. Know what that is? It's how ISPs choose what to throttle... and what not to throttle.
Click to expand...

I'd love to see the packet sniff results of that test. Plus, it'd be nice to see a like for like comparison with the results for the web interface. Also, should have selected a range of distant ends to control for link utilization variability.
 
sfsuphysics said:
You sure about that? Ookla sights are well known to be given priority access on major ISPs so that you feel good about yourself about how fast your internet is, yet it doesn't seem to always be the case everywhere else on the internet.
Click to expand...

I am going to play devil's advocate here for a moment as somebody who's job it is to run the customer facing side of an ISP. I will agree we do have our own speedtest server onsite and normally use that to give customers better number since they are on our network and gives a better representation of their speed from the standpoint of not having lose from other networks, but will agree it doesn't show what you will get while out on the actual internet. However ISPs can only do so much.

I ensure that my network doesn't have any bottlenecks for customers. We have a connection to a carrier hotel with a connection that is over 5 times what our customer usage is and hand most of our traffic off to one of the top tier ISPs while peering with Google, Netflix and a few others. I can't help what happens once my customers leave my network. Therefor if I configure a 20mbps customer for 20mbps, and they have 20Mbps through my entire network till I hand them off I can't help if they get to youtube and only get 15 or they get to some popular server farm and only get 5. I know that youtube used to have a cap of 10Mbps, I know because I tried to max out a 75Mbps connection with youtube and it capped at 10. So my 4k video I was trying to play buffered due to the low bandwidth. I can't help that they had a limit. Just like when you try to download from certain sites and they limit the bandwidth of a single person.

outside of the very large guys where you never leave their network as an ISP you can't promise quality of service once your customers leave your network. You can do your best on your network but that is it.
 
mnewxcv said:
which one of those two does speedtest run by default?
Click to expand...

Neither. Both are ones I chose to be off-network, for more realistic tests. It actually doesn't choose a great one by default since it chooses one in the same city as me, but not only is that ISP not great but the way Cox handles the provisioning we are provisioned through a nearby city so an ISP there actually gives a lower ping.
 
Ragenrok said:
Going to disagree with you, just tried testmy net and it was off by a fair bit (as in much slower). Speed test is way more accurate compared to what I actually get for download speeds from things like steam or blizzard games.
Click to expand...

Did you run single or multithread test? DId you select from different servers?


This explains the main differences and reasons why it's so varied compared to speedtest
http://testmy.net/ipb/topic/28902-why-do-my-results-differ-from-speedtestnet-ookla-speed-tests/
 
All this arguing/debating and there's a simple solution to it all. We know people try to do silly stuff like speed tests on wifi, tests to CA when they live in NY expecting super-fast "I pay for this and expect this all the time" speeds and other silly crap like that. Route prioritization to mask issues is probably all part of it as well for some popular speed test sites.

PEBKAC aside, here's the bottom line and the simple solution: Customers of an ISP should be able to see load graphs on routes they take on nearly EVERY HOP. All load graphs up to their ISP's peering point that the ISP is *responsible* for should be at hand when an issue is reported so proper diagnoses can be performed. This is not usually at hand for call center reps either, I might add.

I've worked at TWC (L3) and know that some routes can definitely hit 99% utilization in one or both directions at some times of the day. Sometimes all it takes is a configuration change or perhaps a upgrade of hardware in some fashion. It happened once when I worked there and I got it escalated and fixed by myself and even THAT was a struggle.

The point is that if I didn't have that graph to back me up, nothing would be done (without hours on the phone and possible tech visits and ticket escalations that many don't waste time with... it's pulling teeth) and I would just be labeled another one of those pesky dumb residential customers complaining.

The whole reason I'm dropping this here is lately my TWC connection in (Hudson Valley) NY has been completely inconsistent for *UPLOAD* speed... which is shocking to me as it's usually the download throughput being garbage. We're talking wild fluctuations from my 6 Mbps rated speeds that dip down to 1-3 Mbps - including stalls and more which interrupt my 3500 kbps (CBR) stream to Twitch at night.

Now I only notice this because I broadcast online but I could see this impacting others with demanding online games as well. I suspect TWC has a congested peering point someplace, but there's no way to tell and no one will look further into it to prove that it *ISN'T TWC* which is all I'm looking for.

Frustrating. (I'm sure I missed something in this mild rant, but there you have it.)
 
Stiler said:
Did you run single or multithread test? DId you select from different servers?


This explains the main differences and reasons why it's so varied compared to speedtest
http://testmy.net/ipb/topic/28902-why-do-my-results-differ-from-speedtestnet-ookla-speed-tests/
Click to expand...

Honestly not sure on single, just picked test download and let it go. Testmy net came back with 78Mbps avg after 10 tests to a server in Texas, then i ran speed test to Texas and got 102Mbps a few times in a row then downloaded a game for good measure and achieved roughly 103Mbps steady after 500mb worth of download. I'll play some more tonight as im actually curious about this now but so far speed test has been more accurate.
 
You must log in or register to reply here.
Back
Top