UK broadband advertising will quote average speeds from May

[HEXUS]

Some ISPs like Sky are already listing average speeds alongside their ‘up to’ speed claims.

Read more.

[mers]

I've always taken quoted speeds with a pinch of salt and realise it will also be area dependent due to installations Etc. We are with Virgin and live in Surrey , having opted for Virgins 100M package I average about 84 Mbps on downloads and I accept that as good enough and will credit Virgin for having next to no problems over the years , service has been excellent.I'm old enough to remember when we just had dial - up , hours to download a few megabytes and pray your connection didn't fail. We really ought to be thankful for the progress made down the years.

[watercooled]

I don't see how this is anything but a useless statistic being presented as something it's not. It really doesn't give anyone an indication on the speed they will receive - the estimate already does that, but what this sort of poorly thought out nonsense does do is encourage selectively accepting subscribers and speed test manipulation.

Some ISPs already refuse to supply a connection to people with a very low estimate as it drags their stats down, and speed tests are often hosted on-net and prioritised to some degree on some ISPs anyway.

ISPs like Virgin get away with leaving congestion issues to fester for months or years with no resolution, but as long as speedtest.net looks ok for a decent percentage of people, the average still looks better than other ISPs. And they can still drag up the mean and median by introducing speed bumps.

/Useless regulation rant

Edit: oh and one more thing, as for that top 10% nonsense, my package was advertised as 'up to 74Mb' yet my line syncs at 80... Yet more useless regulation faffing.

[saldos]

I was there with the engineer. The first one out of the cabinet with FTTC Superfast a few years back. Engineer then line tested at 59 Mbps, first week or two I downloaded at about 48, now my 'upto' 76 speed is closer to 40.

Line speed and download speed; does anyone get the theoretical maximun?

[watercooled]

I was there with the engineer. The first one out of the cabinet with FTTC Superfast a few years back. Engineer then line tested at 59 Mbps, first week or two I downloaded at about 48, now my 'upto' 76 speed is closer to 40.

Line speed and download speed; does anyone get the theoretical maximun?

I get slightly over. In fact max attainable rate on my line is over 90Mb downstream but Openreach cap it at 80. Theoretical max for the profile used by OR is 100Mb.

They will usually drop you to the 40/55 capped package if your sync is lower than that rather than you paying more needlessly.

Dropping over time can be due to crosstalk if lots of people have subscribed or a deteriorating line. Also check there's no noise on your line, including if you nudge phone cables during the test.

[saldos]

I get slightly over. In fact max attainable rate on my line is over 90Mb downstream but Openreach cap it at 80. Theoretical max for the profile used by OR is 100Mb.

They will usually drop you to the 40/55 capped package if your sync is lower than that rather than you paying more needlessly.

Dropping over time can be due to crosstalk if lots of people have subscribed or a deteriorating line. Also check there's no noise on your line, including if you nudge phone cables during the test.

Ahh there's another project for me - Is there an easy method for checking the noise on my internal cabling?

[watercooled]

I was on a phone earlier so couldn't type anything too lengthy sorry. The best way to do it is dial 17070 and choose the option for a quiet line test - listen for any crackling and try moving about some of the phone cables and see if that changes anything.

Now just something I've experienced myself, I noticed noise during one of these tests (started diagnosing because modem randomly rebooted a number of times) and narrowed it down to the microfilter - unplugging/replugging it a number of times seemed to scratch through the layer of corrosion or whatever it was and made the line quiet again. Just make sure you switch off your modem/router by the mains before faffing causing line noise or it might cause DLM to panic and drop your line rate again.

[konfliqt]

I was there with the engineer. The first one out of the cabinet with FTTC Superfast a few years back. Engineer then line tested at 59 Mbps, first week or two I downloaded at about 48, now my 'upto' 76 speed is closer to 40.

Line speed and download speed; does anyone get the theoretical maximun?

They need to start clearing out the last copper from the green boxes to people#s homes now, even if it just starts with people with extra cash paying a few hundred (their neighbours would then be cheaper to add). Since buying fibre to the premises from companies like Hyperoptic, the 150/150Mbps package actually gives customers 200Mbps down and up. I get 946Mbps down/up on their 1Gbs service. None of that distance degradation rubbish

[jimbouk]

A step in the right direction, and I guess the average Joe or Jane will still be able to understand it.

I wonder if a percentage of sync speed would give an accurate measure of an ISP's actual performance? What you get to the cabinet on a normal line is down to Open Reach generally, but if you can't maintain that speed at peak time then that's the ISP not having large enough tubes.

[watercooled]

They need to start clearing out the last copper from the green boxes to people#s homes now, even if it just starts with people with extra cash paying a few hundred (their neighbours would then be cheaper to add). Since buying fibre to the premises from companies like Hyperoptic, the 150/150Mbps package actually gives customers 200Mbps down and up. I get 946Mbps down/up on their 1Gbs service. None of that distance degradation rubbish

Hyperoptic was a economically viable business plan as early as it was because phycically running fibre to an apartment complex once is very cheap vs what it would take to run it to dozens of houses individually. It's not really comparable.

A step in the right direction, and I guess the average Joe or Jane will still be able to understand it.

I wonder if a percentage of sync speed would give an accurate measure of an ISP's actual performance? What you get to the cabinet on a normal line is down to Open Reach generally, but if you can't maintain that speed at peak time then that's the ISP not having large enough tubes.

IMO it's just yet another random figure that really doesn't have much of a correlation to the reality of the speed you get. Saying an ISP's median speed is 50Mbps because they have a load of people too far from the cabinet to gain full sync speed is a pretty worthless statement if the ISP has sufficient capacity to provide the full 80 to those who do actually attain full sync speed. Likewise, the median being high doesn't mean congestion isn't terrible in your area, or on the contrary you might be in a relatively uncontended area.

And as I also said it also discourages any ISPs from taking on long-line customers who will tarnish their precious 'average' figures. Lies, damn lies, then statistics. It's basically useless to consumers and ISPs will be able to manipulate it to say pretty much what they like, probably at the expense of even more customers. Worse, it could end up making good ISPs look bad (e.g. if they're happy to take on any customer regardless of sync speed and plan capacity well) and bad ISPs look good (e.g. if they prioritise for speed tests and cherry-pick customers).

Edit: The sync speed estimates are a really useful figure. And I agree that something indicating the distance from sync speed e.g. at peak times would be a decent indication of how well the network copes at peak time overall. Though, I'd still be concerned about how exactly this 'speed' was estimated as it's vulnerable to manipulation or prioritisation if it's a single website or type of test. And given the range of speeds now available it would be challenging to make this information useful e.g. xISP's speed being 'on average' 10Mb lower than sync isn't much if it mostly applies to their 10Gbps lines, but it's quite significant if it's mostly affecting their ADSL lines! Similarly, a percentage drop could mean anything between a massive congestion problem to being within the testing error margin depending on which sort of lines it applies to. Overall it would be hard to indicate with a single number. It's like 'average' ISP speeds and how they're effectively useless taken in isolation - if aISP has an average of 58Mb but sells mostly 80Mb lines, and bISP has an average of 54Mb but sells mostly 55Mb lines (or if you have no idea about the numbers regarding sync and package speeds subscribed to), which is better?

[Zhaoman]

Aside from all the points made already, as a general rule of thumb I would prefer to see median rather than average values quoted for this type of thing. Average values are becoming more and more unrepresentative and often tend to overstate what the actual 50% percentile user is getting and this is plain misleading for the average Joe Bloggs who might be reasonably expecting the "average" values to be giving an indication what the average 50% percentile user is getting.

[watercooled]

Mean/median would work better on fixed-rate technologies like Cable, FTTP, etc provided the numbers are separated out into packages e.g. what do people get at peak times on a 350Mb Virgin package - that's of some use. Anything below the advertised rate isn't due to sync speed, but it still leaves things such as WiFi, connection utilisation during tests, hardware performance, etc. skewing results particularly at higher speeds.

As for variable rate technologies like xDSL it's going to be hard to sum it up with one number really, whichever sort of averaging you use. Maybe a graph showing sync speed vs line rate - an ISP that keeps on top of congestion would show a fairly linear line, while one that lets things get out of hand across the board would likely flatten out towards the top. It's still going to have many of the limitations I've already mentioned though.

There are just to many variables to present that sort of information usefully as a single number.

[Xlucine]

Aside from all the points made already, as a general rule of thumb I would prefer to see median rather than average values quoted for this type of thing. Average values are becoming more and more unrepresentative and often tend to overstate what the actual 50% percentile user is getting and this is plain misleading for the average Joe Bloggs who might be reasonably expecting the "average" values to be giving an indication what the average 50% percentile user is getting.

1) Median is average, you need to be more specific
2) Third paragraph:

So how will the new guidance system, which ASA believes will provide “a better indication of the actual speeds,” work? Again it is focussed upon download speeds. This time ISPs must advertise the median downstream speed measured at peak time (8pm to 10pm) for their headline consumer ad claims.

[Fizzl]

Not always the ISPs fault. I live quite close to the exchange but was getting 40/19. I made a lot of fuss, get the engineers out, they tested the speed in the street and 80/20 was achievable. Much badgering later and some digging up the garden, they replaced the wet string with a proper conduit, new cable and gave me a new primary socket. I'm now at full speed.