[talk-au] The Paradox of Postcodes (Was Re: Victorian Vicmap Address Import Proposal - Suburb and Postcode discussion)

[Ewen Hill]

Hi all,
    We have been using the ABS 2016 postcode boundaries under intense
scrutiny compared to the "current" Auspost and found very few
inconsistencies in Victoria and I would suggest that this is mere legal
jargon to avoid any commercial litigation or that this was required under
the release terms to the ABS. The issues we have seen are

   - The new postcode of 3336 for Deanside, Aintree and Fraser Rise -
   https://auspost.com.au/postcode/3336
   - Another new postcode or altered postcode boundaries around Tallangatta
   or Albury (can't remember what the specifics of this one were)
   - The population of 3066 of Derrimut and Laverton North of "72" when it
   was  industrial / farm land has now ballooned into the thousands
   - Some minor anomalies where roads have been rerouted (Geelong by-pass
   from memory)
   - Some park land / national park differences

As there is no formal process by Australia Post that I can see from
announcing changes, then I see Andrew's approach is solid and will require
minimal upkeep - and who sends a letter nowadays ;)

Ewen



On Sat, 19 Jun 2021 at 03:58, stevea <steveaOSM at softworkers.com> wrote:

> On Jun 17, 2021, at 2:14 AM, Andrew Harvey via Talk-au <
> talk-au at openstreetmap.org> wrote:
> >> It's a fair point that Vicmap's own postcode field shouldn't be taken
> as 100% correct, it looks like it might have been assigned based on
> postcode boundaries so might still suffer issues because of this, but where
> addr:postcode is not already mapped, most of the time the Vicmap one will
> be correct.
>
> To be clear, I'm 100% OK with postcodes on nodes with addresses, such
> things belong together (as that tag on that node):  it is indeed "the
> correct way to go."  (IMHO).
>
> I'm not terribly excited (dejected) to see a suggestion that ABS'
> described "imprecise process" (of conflating postcodes with geographic
> boundaries) is glibly said as "we can still have postal_codes on admin
> boundaries where the vast majority of addresses within that boundary have
> that postcode."
>
> In the USA (in OSM) we say rather bluntly "ZIP codes are not boundaries."
> (ZIP codes are USA postcodes).  It seems ABS agrees.  Putting them on
> entire admin boundaries, especially where they are not 100% correct (all of
> them?) adds noise to our data, which I am identifying and say "in the USA,
> we just don't do this" (as they are simply not the same).
>
> Though, postcode tags on address nodes, sure.  Good way to do it, correct
> way to go, et cetera.
>
> In the USA, OSM imported mid-2000s national census data to "lay down a
> road grid."  We continue to unravel and fully "TIGER Review" these data, 15
> years later.  They are "noisily (though that gets better over time, with
> effort) mostly correct" today, but.
>
> There is a wide distribution / spectrum of such (postal) data scattered
> around OSM in various jurisdictions.  I'm saying that at this level of
> conversation, pave the road smarter, rather than glibly or easily.  Good
> planning makes better maps.
>
> Thank you for saying "fair point," too.  I hope I haven't beaten it up too
> much, so thank you to all for patience reading.
> _______________________________________________
> Talk-au mailing list
> Talk-au at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-au
>


-- 
Warm Regards

Ewen Hill
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-au/attachments/20210619/318cc582/attachment.htm>