Bless 'em, they're trying to do something constructive with their spare
time, but they need some way of rooting out the rogue data. I think the
little enclave of RG (which is the Reading postcode) in the middle of
Streatham is my favourite bit.

http://dev.openstreetmap.org/~random/postcodes/

John Rowland wrote:

Just like there appears to be an enclave of SK (Stockport) in the middle
of OL (Oldham), and OL4 apparently doesn't exist at all, while there are
two OL15s, and OL2 is apparently centred on the Woodhead pass towards
Penistone.

It's a nice idea, but the execution's... sad.

On 15 Sep, 15:59, Stephen Farrow gmail.com> wrote:

And it seems that a lot of the sea bed has a postcode too.

On Sep 15, 2:24 pm, "John Rowland"
journeyflow.spamspam.demon.co.uk> wrote:

That's the big floaty one which is still visible when you zoom out.

Those seem to be wildly wrong all over the place, but I think the
individual postcodes when you zoom in are more accurate.

Then again ... not much more. Interesting that there's an SE08 as
well as SE8, an E32 next to E3, two E1Ws in different places, an E61
(wow). Not just in the wrong places, but they don't exist.

If it wasn't for the non-existent ones, I'd wonder if it was a simple
programming bug, and they'd all line up if it was fixed.

On Sep 15, 5:26 pm, MIG doreenbird.co.uk> wrote:

It's really just a technology demonstration to show what can be done
with the data - and also to visualise how good the data is! The
particularly odd bits of data most likely derive from random people
misunderstanding how to enter postcode data.

The postcode stuff isn't a core part of OSM, and is going to be far
harder to survey independently than streets themselves, which we're
doing reasonably well at in some places.

MIG wrote:

Not in my part of the world, they're not (see separate post further up
the thread).

--

Stephen

On 15 Sep, 18:31, Abigail Brady gmail.com> wrote:

Abigail Brady wrote:

In article , Stephen Farrow
gmail.com> writes

It also doesn't understand the Cambridgeshire recoding. CBn and CB2n
aren't horribly intermingled. Rather, CBn codes outside the city
boundary are being changed to CB2n codes, sometimes by just adding the 2
and sometimes by more drastic changes.

On Sep 15, 7:06 pm, "John Rowland"
journeyflow.spamspam.demon.co.uk> wrote:

I think this is their ambition.

That's an interesting idea. One thing that may not be immediately
apparent is that the main postcode database source is trying to be PD,
and is therefore based around people locating their postcode on out-of-
copyright Ordnance Survey maps. Which are therefore >50 years old.
Feeding actual openstreetmap tiles (which are creative commons) into
this process might be considered to make the resulting data creative
commons as well. (Perhaps this is hyper-cautious, certainly the OS
are of the opinion that if you find lots of things on one of their
copyrighted maps and make a table of their coordinates, that's a
derived work. Not sure of any UK caselaw.)