Skip Navigation


Why Ofcom changed the way 2 Mbps postcodes are defined
Tuesday 21 May 2013 11:47:31 by Andrew Ferguson

The data Ofcom has used to arrive at the figure stating that 95.3% of the UK has access to broadband of faster than 2 Mbps was published as part of their 2012 Infrastructure Report, and we have taken the liberty of plotting the data, based on how the USC was measured in 2012 and how Ofcom is now measuring it. The red spots plotted are those postcodes that fail the USC test.

Differences between how Ofcom measure 2 Mbps USC in 2012 and 2013
(click image for larger version)
A full size image (3952x2704 pixels) is also available.

Statistically speaking there is some sense in adopting the new method, where only postcodes that have a median or average speed of 2 Mbps or slower are counted as slow spots. The previous definition of the postcode was plotted if any property in a postcode had a slow speed could have meant that people with poor extension wiring were contributing to the picture, or those with old legacy products.

The big problem though is that the new metric makes the situation look actually pretty good and if taken as gospel could lead to many people being overlooked in the rush to publish statistics in 2015 that show that 99.9% of postcodes have an average above 2 Mbps. The advantage to the new method is that it identifies the worst postcodes, and if you have limited funds to solve the problem it tells you where to start spending the money.

Yes we know Northern Ireland is missing from our map, alas we do not have the geo data to plot those postcodes at this time.

The original Digital Britain report recommended that in addition to spending money on infrastructure improvements, that some money be spent to try and improve speeds on existing technology that is in the ground, and while the volume of people with poor telephone wiring in the home has decreased we still see constant stream of people where their broadband speeds are improved with some help, sometimes allowing people to watch catch-up TV online for the first time.

Comments

Posted by herdwick over 4 years ago
Any chance of a black on white version ? It took me 5 minutes to realise there are dots on the right hand side
Posted by mikejp over 4 years ago
"could have meant that people with poor extension wiring were contributing to the picture, or those with old legacy products." - problem surely avoided if the correct data - is sync speed,is used? Not sure what a 'legacy product' is but if you mean dial-up. most dial-uppers are there because.....................
Posted by herdwick over 4 years ago
Someone on Home 500 (A legacy product, they are out there) would record a sync speed of 576 so a single one of them would condemn a postcode under the old approach.

Rubbish home wiring could have someone sat at 1M or less in a location that would give them 6M if they sorted it. I fixed one at 160k to deliver 12M. These outliers and anomalies should not mess up the stats.
Posted by Dixinormous over 4 years ago
Where's the raw data lurking?
Posted by mikejp over 4 years ago
"Rubbish home wiring could have someone sat at 1M or less in a location that would give them 6M if they sorted it" - would that not be ignored by the router sync speed?
Posted by csimon over 4 years ago
I too am having trouble seeing *any* dots at all on the right hand map! I'm colour blind, can you provide a larger version or change the colours? I was looking for my postcode area, which contains 22 houses. There are at least 5 of my neighbours who like me are well below 2mbps (I'm currently on 300kbps), therefore statistically as stated above we should still be one of the WORST postcode areas under the new rules. what constitutes a postcode for the purpsoes of these rules? The complete postcode or a truncated version?
Posted by andrew (Favicon staff member) over 4 years ago
http://www.coolwebhome.co.uk/postcode/checker.php

Lets you check a specific postcode, and it is using the full one.

In terms of specific areas, we will plot smaller areas so people see more detail in due course.
Posted by herdwick over 4 years ago
@mikejp - no, wiring problems affect the sync speed. Some might also affect speedtests if they generate errors, but domestic wiring issues primarily hit the sync speed.
Posted by mervl over 4 years ago
Tried it for a few rural hamlets (around 50 houses or so) where friends live (and the odd one I thought of moving to). Whilst it works well in my village/suburban location, for those rural locations it shows "insufficient premises" for data which I suppose could be a useful way of meeting the targets!
Posted by herdwick over 4 years ago
Data at http://d2a9983j4okwzn.cloudfront.net/downloads/ofcom-uk-fixed-broadband-postcode-level-data-2012.zip
- last years ?
Posted by Kushan over 4 years ago
"The big problem though is that the new metric makes the situation look actually pretty good and if taken as gospel could lead to many people being overlooked in the rush to publish statistics in 2015 that should that 99.9% of postcodes have an average above 2 Mbps."

You might want to read over that, it doesn't make any sense ("that should that").
Posted by csimon over 4 years ago
Thanks for providing the link to a direct postcode. LOL, my result is:

Average Speed: 0.5 Media Speed: 0.5 Maximum Speed 0:9

Can't see my dot on the map though!
Posted by csimon over 4 years ago
I think you're probably right that the new rules should tell you where to start spending the money. I mean, even the Superfast Wales scheme got it wrong a couple of months ago, rolling out a fibre cabinet first to the centre of a postcode area which has the following stats:

Average: 8.5Mbps
Median: 8.6Mbps
Maximum: 15.3Mbps

Obviously, they've just been guessing up to now as to where to invest the money, now they've got the information to get it right! (oh, the irony...)
Posted by mikejp over 4 years ago
Can someone clear my confusion please? OfCom have produced speed data (from ISPs) for each postcode. I see a max of 21mb in my post code, 4km+ from an ADSL2+ exchange. I find that difficult to believe.

How are these figures derived? If sync speed/IP profile is no guarantee due to 'internal wiring issues', how are these figures produced by the ISP? Can a punter establish his or her actual speed? My modem/router is fed from the master socket via a splitter plate, so I am stuck on how internal wiring can affect my sync speed which is what I am told (above).
Posted by camieabz over 4 years ago
Great visual aid (albeit, difficult to spot the right-hand stuff, but that's the point).

Perhaps an outline of the UK if done again?
Posted by herdwick over 4 years ago
http://maps.ofcom.org.uk/broadband/broadband-data/ might help mikejp

"Average modem sync speed (Mbit/s)
The average maximum speeds of existing broadband connections. Speeds achieved in the home will be slower."
Posted by Dixinormous over 4 years ago
Ah I see it's the same data but using the median not minimum speed.

Ah well at least this postcode stands proudly at 1.5Mb median.
Posted by nerdmeuk over 4 years ago
Hmm, using that tool (thanks for the line) I can see that the average for my postcode is 16.4 Mbps, we get < 3 Mbps :-(. The exchange is not part of the current FTTC roll-out so these are ADSL2+ connections.

I'm not surprised as all the 1950's housing that surrounds us is supplied over-ground by relatively new copper, our house and a few others built in 1980 are supplied by old under-ground aluminum.

Seriously considering FTTP on-demand assuming our cabinet is eventually FTTC enabled.
Posted by andrew (Favicon staff member) over 4 years ago
Re-plots are underway, and a new map too. If it comes out looking reasonable.
Posted by mdar5 over 4 years ago
Insufficient data for me
Mind you there are only 2 houses in the postcode!

Some extra high "silly" result for max sync may be due to a single property buying a commercial solution such as multiple line bonding to get umpteen tens of Mbps
Posted by Fellwalker over 4 years ago
What a fiddle! If half the people in a postcode can afford ADSL2 or fibre, the ones on traditional broadband are out of it and will never be counted as not having 2Mbps. I've two friends nearby who each get just under 2, whereas many pay for Infinity. They are both on TalkTalk because it is the package they can afford. Sometimes they can watch iPlayer, but often it stutters.
Posted by herdwick over 4 years ago
ADSL2 doesn't come at a price premium, and besides it's an availability map not a personal choice of what to spend money on map. Infinity is £5/month more.

I don't want to pay more tax to subsidise your friends, sorry.
Posted by andrew (Favicon staff member) over 4 years ago
mdar5 Bonding will not boost this dataset, it is the sync of individual lines.
Posted by WWWombat over 4 years ago
I guess we have two separate issues here:

Ofcom's measuring, and the BDUK targets of 100% 2Mbps coverage.

Is BDUK target 100% of premises? Or when Ofcom's chosen measure reaches 100%? I think premises.

Which of the two Ofcom methods leads to the best summary of what work is actually needed on the ground? I can see that ignoring outliers is a good move, but using median might not be.

Rusty mathematics reminds me that using +/- 2 standard deviations was a good method for normal distributions. Would that be better?
Posted by WWWombat over 4 years ago
Better images now. Ta!
Posted by chrysalis over 4 years ago
these new maps look much better.
Posted by WWWombat over 4 years ago
With the new images, my eyes tell me that the densest-looking areas (in the 2013-method map) are East Lancs/West Yorks. Does it look this way to others?

I wonder if they were further behind on FTTC rollout last summer.
Posted by mdar5 over 4 years ago
Andrew:
My area does not have fibre FTTC or FTTP - only ADSL2+
However in the next door post code to me which is further up the road from me the data set says the max sync is given as >30Mbps.

Can you explain how this is?
Posted by andrew (Favicon staff member) over 4 years ago
Hopefully the addition of a higher resolution map will help people.

If there is demand, then happy to plot postcode area maps also e.g. GU postcodes only
Posted by New_Londoner over 4 years ago
New maps much better, thank you!

The methodology will hopefully ignore problem home wiring which is good, but could still highlight an area as slow because people are opting to keep on slow ADSL lines when FTTC/P is available. IMHO there ought to be a distinction between those areas where other options are available and those where the speeds plotted are the best available.
Posted by andrew (Favicon staff member) over 4 years ago
This is not the only map I am producing, assuming people make use of them.
Posted by mikejp over 4 years ago
" IMHO there ought to be a distinction between those areas where other options are available and those where the speeds plotted are the best available. " - not forgetting that FTTC/P may not be available to all premises in a particular postcode.
Posted by New_Londoner over 4 years ago
@Mikejp ^^^^
Agreed, but where it is then highlighting a low average speed is irrelevant as it presumably means the residents are choosing to spend their money on something else. It's their choice so fair enough, not a matter of concern to others.
Posted by camieabz over 4 years ago
@AF

Better map, that white one. No sea-blue ink in the MS Paint palette? :p
Posted by andrew (Favicon staff member) over 4 years ago
MS Paint has not been near those images :-)
Posted by bbluefoot over 4 years ago
I can't speak for the rest of the country, but practically the entirety of North Norfolk (outside towns - which is most postcodes) should be black. I know this for a fact. If the rest of the country's map plots are as misrepresentative as this, the real situation is immeasurably worse than even the old data method.
Posted by mikejp over 4 years ago
Thank you, bbluefoot - exactly the point I raised in the thread "Coverage of 2 Mbps broadband increases to over 95%" (post #4) which everyone appears to be ignoring with a shrug of the shoulders. Surely this misrepresentation of actual speeds needs urgent addressing and there are obvious implications for the glowing report BT will no doubt give itself in 2016 - and this will doubtless be gladly 'accepted' by HMG and the Local Authorities in a blaze of "haven't we done well".
Posted by WWWombat over 4 years ago
@mikejp
Your comment over there (with one example of where the estimator was wrong) was not ignored. There was a lot of discussion pointing out that the data used in this report doesn't come from BT's speed estimator.

@bbluefoot doesn't seem to raise any question at all about the estimator. So doesn't seem to be the same point either.

There are undoubtedly questions about the validity of the actual speed measurements, but that's a different point to the one you made in post #4.
Posted by WWWombat over 4 years ago
@bbluefoot -
Won't most postcodes be in towns & cities, where the majority of premises are?

In rural areas, the postcodes may be a bigger area, but there won't be more of them.
Posted by WWWombat over 4 years ago
@Andrew
When you plot a "low-speed" postcode on the map, do you paint the whole area represented by the postcode? Or do you paint a fixed-sized "dot" at the postcode's geo-location?

Fixed-sized dots give a reasonable representation of the number of premises affected, as postcodes tend to cover similar numbers of properties

Painting the full area of the postcode would give a reasonable representation of the landmass affected, but would appear too black in uninhabited portions of the country.

I think a fixed-size dot is best - and might explain why @bbluefoot doesn't see a black county.
Posted by mikejp over 4 years ago
WWWombat - I have not seen any clear indication of where the data comes from - have you? All I have seen is "data from ISPs" - a little indeterminate?

By the way, the "21mb max speed" for my postcode quoted by OfCom has been admitted by them to be "erroneous".

Posted by andrew (Favicon staff member) over 4 years ago
It is a fixed sized dot, a single pixel in the maps here.

The Norfolk story is same as elsewhere if near exchange good speeds are possible, if two villages away then its slower.

We have some other data from speed tester that I hope to share soon.
Posted by WWWombat over 4 years ago
@mikejp
This Ofcom reports states "The underlying data file on broadband speeds provided information on the number of residential and small-business lines with a broadband connection in each postcode."

The 2012 infrastructure report states: "For each postcode, this notes whether there are any sub-2Mbit/s broadband lines, the average sync speed, the median sync speed and the
maximum sync speed."

I think it is clear that they are talking actual sync speeds, and they even note it is of existing connections.
...cont>
Posted by WWWombat over 4 years ago
<cont...
They further go on to state:
"It should be noted that the sync speed information published here is not directly
comparable with speed test data measured on an end-to-end basis by companies such as SamKnows. In our latest report using data collected by SamKnows we measured an average speed of 9.0Mbit/s"

They also describe the differences between the Samknows-managed speed tests (end-end tests) and the sync speeds values.
...cont>
Posted by WWWombat over 4 years ago
<cont...
I can't quote the whole lot in 600 chars, but they specifically say two things...
"End-to-end measurements determine the network performance from the home to a typical web-site. They take into account not just the performance of the access network, but also... [a list of things]"

"The sync-speed provides a direct means of determining the performance of the access network, as reported by the modems which maintain each broadband connection."

And what we want to know, in this context, is the performance of the access network.

It seems clear to me.
Posted by mikejp over 4 years ago
Thanks WWWombat - I have not downloaded the 'paid-for' data from Point Topic so I missed the bit about sync speeds. Hopefully my '21mb' was the 'erroneous data' in that provided by BT.................
Posted by WWWombat over 4 years ago
I only got thse from Ofcom's reports on the dataset, but I agree - when you find a mistake in someplace you know about, it throws your confidence in the full set of data.

The government are going to rely on these results, whether we like it or not, so the best thing we can do is make them better reflect reality.
Posted by mikejp over 4 years ago
From OfCom
"The data was based on information provided by the largest of the ISPs."
Posted by gah789 over 4 years ago
The figures cannot be taken seriously as a basis for any policy analysis. In my area and others I have checked the postcode checker shows No Data, thus glossing over large areas with minimal DSL service. On the other hand, anyone can get a 20 Mbps satellite connection if they are willing to pay and can live with its characteristics. So the targets just become an empty exercise in manipulating definitions.
Posted by WWWombat over 4 years ago
@gah789
Satellite does get you around the targets, except for 2 (or 3) simple problems:
- It is too expensive, so doesn't meet the affordability requirement that are also part of the BDUK targets (basic BB: <£100 install, <£25pm).

- Low capacity, so even in 2014 there will only be 300,000 connections to serve the UK. Not enough satellites or transponders pointing at us.

- Even supporting only 1% of the UK, there are very low usage allowances too, with high costs to use more.

BDUK expect to use between 100k-200k satellite connections.
Posted by BBSlowcoach over 4 years ago
At 21:40 the wind must have stopped blowing as the power died as I was trying to post my comment!
I was attempting to say that I hope my Post Code IP30 is one of the black dots as a speed check at around 21:25 revealed 0.896Mbps. At 'good times' it can be as high as 7.0Mbps momentarily! With lower being more normal it underlines how stats can be used against those parched in the BB desert.
Posted by cyberdoyle over 3 years ago
"I love statistics. I can make them say anything I like" Winston Churchill.
You must be logged in to post comments. Click here to login.