Microsoft Patent, or is it ?

[Dave]

Microsoft are trying to patent a novel way of coding geo locations. They aim to make it easier for end users to type or spell coordinates. The proposal is to limit the coding to a list of 30 different characters that can be used: 0123456789bcdfghjkmnpqrstvwxyz

So basically all numbers and most consonants. This base-30 number system (anyone remembering the Mayan base-60 calendar?) is then applied to the coordinates via some cunning transformations:

A latitude of 47.64932 converts to the integer value of 18,582,657. A longitude of -122.12926 converts to the integer value 3,906,275. The latitude-based integer value of 18,582,657 converts to a base-N notation of ry7cx. A longitude-based integer value of 3,906,275 converts to 4tp95. The encoded string resulting from the latitude/longitude coordinate would be "ry7cx4tp95".

As a net result you save a few letters, the location is easier to include in URLs, and can easier be shouted over the mobile phone. But is it patentable?

Full details are available here.

[Skippy]

So what the patent is claiming is to have invented a way of encoding a floating point number (WGS84 co-ordinates in this case) using base-30 encoding (invented donkeys years ago) for use in a URL.

This is not novel, base-16 encoding of data for use in a URL is documented by Tim Berners-Lee in RFC 1738 and is part of the HTTP standard.

All this patent has done is changed the base from 16 to 30, and specified that the data to be encoded is a lattitude/longtitude.

This patent is completely without merit, I can't believe what people will try and patent these days. No doubt the Patent Office will grant this patent regardless.

The only amusing thing is that they have used only consonants so that you don't get your lat/long encoded as a rude word. Spoilsports.
_________________
Gone fishing!

[Oldie]

Has anyone tried a few of the common typos on the coded coordinates? When people send me coordinates that they have typed I get a few twiddles and adjacent digits. I suppose a typo will now end up with a location miles from that intended.

Richard
_________________
Various TomToms, Garmin eTrex Legend, GPSMAP 60CSx, Oregon 550t, Forerunner 405

[Bodo]

When somebody sends me a co-ordinate of 47.64932 and -122.12926, then I can visually see where in the world this is. It is immediately clear that this is not in the UK
However, ry7cx4tp95 is totally meaningless - and will remain so, even after practice.
Computer memory, disk space and communications bandwidth is getting so cheap nowadays - who cares about the few saved characters?
I for one would vote against this because it is not user friendly. Every kid at school learns latitude and longitude - it is easy to understand.
Just like the QUERTY keyboard and the 24 hour clock - we are all used to it, don't change it.

[lbendlin]

You mean the QWERTZ keyboard? Or the AZERTY keyboard?

The main reason for this standing-on-your-head-and-twiddeling-your-toes is not the length of the coordinates but to make it easier to type them and say them letter by letter... But I agree, this way they lose their identifiability (uh?) completely.
_________________
Lutz

Report Map Errors here:
TomTom/TeleAtlas NAVTEQ

[Bodo]

QWERTY?? Not QUERTY??
That probably explains my many spelling mistaiks!

hmmm - just thinking - how will Microsoft mark Latitude and Longitude on a world globe? Now that will be creative thinking, wont it?

[Skippy]

I've been thinking about this one a bit. The idea is to shorten the representation of the position, isn't it.

So my position is N47.64932 W122.12926 (20 characters) which converts to ry7cx4tp95 (10 characters). That's half the length - not bad but surely we could do better.

Looking at the map, this location is the Microsoft HQ in Redmond, Washington. It isn't much suprise to see that this position is a densely populated area. Hey, not many people would be interested in a waypoint which is in the middle of the ocean, desert or the barren wasteland of the Tibetan Plateau. People want to share waypoints pointing to locations where there is Interesting Stuff.

Now 70% of the world is covered by ocean. So that means that at, aside from a few good fishing spots, at least 70% of the possible GPS co-ordinates you could come up with are probably not going to point to places which contain Interesting Stuff. So, that leaves 30% of the world which is land and can potentially contain Interesting Stuff.

However, much of that 30% of the earth's surface which is land is sparsely populated, desert, frozen tundra, forest etc. There might be the occasional bit of Interesting Stuff there, but not much.

The fact is that over 90% of the population inhabit less than 5% of the total surface area of the earth. It's this 5% where most of the Interesting Stuff is, the other 95% of the WGS84 datum is very important - but rarely used.

So, what if we could come up with a coding system so that densely populated 5% of the world which contains the Interesting Stuff will have nice concise GPS positions (say 3 or 4 characters in the base 30 encoding scheme) but more obsure positions away from the major population centers require more characters to represent them.

It's a trade off, but on the whole it would be a winner because, on average, the position of Interesting Stuff that a real world user wants to refer to can be much more concisely represented.

You could also use clever tricks when encoding the location so that the accuracy of the position is implied by the resolution of the numbers, rather like you can with OSGB. ie TQ 333 804 (accurate to +/- 100 meters) vs TQ 30033 80403 (accurate to within 1 meter)

A checksum could also be built-in to the string so that the GPS would know straight away that the user had typed in the position correctly. This would cut down on transcription errors and keep people like Oldie happy.

Does this all sound far fetched? Impossible? Have a look at VideoPlus numbers in the TV guide that you use to program your VCR. Certain time slots on certain chanels at prime time have 2, 3 or 4 digit codes. Obsure time slots on out of the way channels have 6, 7 or 8 digit codes.

OK, so a GPS position has a lot more information to encode than a TV program listing, but the same principle could be applied, based on the fact that 95% of the positions are going to refer to positions on <5% of the surface area of the earth.

Should I patent this idea? Or am I completely mad.
_________________
Gone fishing!

[Fuego]

These sausage-machine patents are evil in the extreme. Once the patent is granted, a large corporation can just bully any small or independent developer that does not have a quarter of a million pounds to put on the line.

Base-n encoding is public domain. Application of Base-n encoding to just about anything is common general knowledge and obvious in the extreme. It matters not that a certain sequence of characters are used - the choice of these would also be obvious to anyone skilled in the art of problem analysis.

These types of patents are often thrown out in Europe, although the UK has unfortunately seen a few of them granted. At least the UK is not anywhere near as bad in this respect as the US.

[barryhunter]

Why not just use IARU Locator or QRA-Locator (more)
This is used by Radio Ameters and gives nice short locations like IO34TN
which is accurate to about 4-3 miles, or 12 charactors gives 3m accurcy.
This has the advantage of being based on Lat/Long so IO covers most of UK. (see Gridmaps)

[barryhunter]

Also Skippy's idea of the variable length code reminded me of something similar which is dubbed Universal Address, see
http://www.travelgis.com/
Although I havent fully read about and figured out how it actully works ;-(

[barryhunter]

Sorry to be replying to myself but reading http://www.nacgeo.com/nacsite/ i see this:

February 9, 2005 - Does Microsoft Infringe the Natural Area Coding System?
Tuesday, 08 February 2005 Toronto, Canada - NAC Geographic Products Inc. has noticed that Microsoft has filed a patent "Compact text encoding of latitude/longitude coordinates" that seems an infringement of the Natural Area Coding System that was developed ten years ago by NAC Geographic Products Inc.

So it seems in typical MS style they are copying others and claiming it as their own!

[nej]

Microsoft are also trying to patent the programming operator "IsNot". For non-programmers, that's like trying to patent a punctuation mark or something.

[DavidW]

By IsNot, do you mean something like C's != test - as in:

We see you’re using an ad-blocker. We’re fine with that and won’t stop you visiting the site.
Have you considered making a donation towards website running costs?.
Or you could disable your ad-blocker for this site. We think you’ll find our adverts are not overbearing!