Discussion:
[I]Another "Quality" Product From Microsoft...
(too old to reply)
Paul Cooke
2005-01-24 17:51:56 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ... 
5. Laugh, Snigger, Cry
Anne Nielsen
2005-01-24 18:12:31 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
That's "impressive"! They obviously want you to take the scenic route.
Or maybe they are sponsored by the ferry company.

-Anne
Matt
2005-01-25 22:20:30 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx> 2. In
the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
That's "impressive"! They obviously want you to take the scenic route. Or
maybe they are sponsored by the ferry company.
Just so long as the ferry doesn't use their software...
--
Matt
John Ewing
2005-01-26 00:34:32 UTC
Permalink
Post by Anne Nielsen
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
That's "impressive"! They obviously want you to take the scenic route.
Or maybe they are sponsored by the ferry company.
After looking at that, I tried using the same start point but a
destination that would require the use of a ferry - Stornoway. For
some reason the software adds an interesting scenic diversion via
Thurso, on the north coast.

Regards,

John
--
John Ewing
Glaschu / Glasgow
Alba / Scotland
John Ewing
2005-01-26 21:20:01 UTC
Permalink
(Following-up to myself, but never mind)
Post by John Ewing
Post by Anne Nielsen
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
That's "impressive"! They obviously want you to take the scenic route.
Or maybe they are sponsored by the ferry company.
After looking at that, I tried using the same start point but a
destination that would require the use of a ferry - Stornoway. For
some reason the software adds an interesting scenic diversion via
Thurso, on the north coast.
Try going from Inverness to Ullapool. The most direct route is shown.
Now try going from Edinburgh to UIlapool. The route shown goes through
Inverness, then detours to the north coast - Thurso - then south-west
to Ullapool. Strange...

Regards,

John
--
John Ewing
Glaschu / Glasgow
Alba / Scotland
Puck
2005-01-24 18:13:30 UTC
Permalink
5. Laugh, Snigger, Cry
I did all three, not necessarily in that order.

Next time someone says Bill Gates can't find his arse with an atlas we will
all know why.


--
Puck (onstage): I am that merry wanderer of the night!
Peaseblossom (in audience): I am that merry wanderer of the night, indeed!
"I am that giggling-dangerous-totally-bloody-psychotic-menance-to-life and
limb, more like."
-Neil Gaiman
Steve James
1970-01-01 00:00:00 UTC
Permalink
*Date:* Mon, 24 Jan 2005 13:13:30 -0500
5. Laugh, Snigger, Cry
I did all three, not necessarily in that order.
Next time someone says Bill Gates can't find his arse with an atlas we will
all know why.
Now we know where GWB learns his knowledge of the geography of the world.


Steve (Steeljam) *BF DAcFD (UU) *
Resident Opsimath in Redivivus Studies
C.C. Baxter
2005-01-24 18:44:35 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
Have you tried the shortest route?
--
Buddy

(Eliminar 17 para responder por correo/Remove 17 to reply by e-mail)
Graycat
2005-01-24 19:09:42 UTC
Permalink
On Mon, 24 Jan 2005 19:44:35 +0100, "C.C. Baxter"
Post by C.C. Baxter
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
Have you tried the shortest route?
Ah yes, the one that thinks you have to go via Newcastle
Upon Tyne to get to Bergen from Haugesund...
--
Elin
The Tale of Westala and Villtin
http://tale.cunobaros.com/
The Oswalds DW casting award - Vote Now!
http://www.student.lu.se/~his02ero/Oswald/index.html
Pudde Fjord
2005-01-25 07:20:05 UTC
Permalink
Post by Graycat
On Mon, 24 Jan 2005 19:44:35 +0100, "C.C. Baxter"
Post by C.C. Baxter
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
Have you tried the shortest route?
Ah yes, the one that thinks you have to go via Newcastle
Upon Tyne to get to Bergen from Haugesund...
That one is fixed now...

Pudde.
C.C. Baxter
2005-01-25 16:33:56 UTC
Permalink
Post by Pudde Fjord
Post by Graycat
On Mon, 24 Jan 2005 19:44:35 +0100, "C.C. Baxter"
Post by C.C. Baxter
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field 3. In the End section, select
"Norway" from the listbox and enter
"Trondheim" into the "City" field 4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
Have you tried the shortest route?
Ah yes, the one that thinks you have to go via Newcastle
Upon Tyne to get to Bergen from Haugesund...
That one is fixed now...
Pudde.
No, it isn't. You have to cross the North Sea twice to go from Haugesund
to Trondheim
--
Buddy

(Eliminar 17 para responder por correo/Remove 17 to reply by e-mail)
Pudde Fjord
2005-01-25 18:35:33 UTC
Permalink
Post by C.C. Baxter
Post by Pudde Fjord
Post by Graycat
Ah yes, the one that thinks you have to go via Newcastle
Upon Tyne to get to Bergen from Haugesund...
That one is fixed now...
Pudde.
No, it isn't. You have to cross the North Sea twice to go from Haugesund
to Trondheim
The route between Haugesund and Bergen was what I commented on. Should
have snipped a few more lines...

Pudde.
Orjan Westin
2005-01-24 18:48:58 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
This has been doing the rounds on a mailing list for The Vikings as an
explanation for the early raids - they were just desperate for a toilet and
a snack and when those snotty monks at Lindisfarne refused them both unless
they converted... Well, who can blame the Norse?

Orjan
--
Get your Tale paperback or CD here:
http://tale.cunobaros.com
Or just read it there, if you don't want the illustrations
Richard Bos
2005-01-24 22:11:06 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
Another quality product of the Open Source community? That is not what
non-breaking spaces are for...
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
...and the logic of doing it for the first line of a sentence, but not
the subsequent part, escapes me completely.
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ... 
5. Laugh, Snigger, Cry
Note that it is not the program itself which is at fault here. The bug
has been found; there is an error in the database. The road from (IIRC)
Haugesund to Oslo is one-way, which it shouldn't be. It's a GIGO case.

Richard
Boyd Bottorff
2005-01-24 23:15:07 UTC
Permalink
Post by Richard Bos
Note that it is not the program itself which is at fault here. The bug
has been found; there is an error in the database. The road from (IIRC)
Haugesund to Oslo is one-way, which it shouldn't be. It's a GIGO case.
Yeah, well, I look at it as a MIMO case.
Richard Bos
2005-01-25 20:31:19 UTC
Permalink
Post by Boyd Bottorff
Post by Richard Bos
Note that it is not the program itself which is at fault here. The bug
has been found; there is an error in the database. The road from (IIRC)
Haugesund to Oslo is one-way, which it shouldn't be. It's a GIGO case.
Yeah, well, I look at it as a MIMO case.
Map In, Map Out? Motorway In, Motorway Out?

Richard
Boyd Bottorff
2005-01-25 22:58:52 UTC
Permalink
Post by Richard Bos
Post by Boyd Bottorff
Post by Richard Bos
Note that it is not the program itself which is at fault here. The bug
has been found; there is an error in the database. The road from
(IIRC) Haugesund to Oslo is one-way, which it shouldn't be. It's a
GIGO case.
Yeah, well, I look at it as a MIMO case.
Map In, Map Out? Motorway In, Motorway Out?
Microsoft in, Microsoft out. It's like garbage in, garbage out, only it
needs to be run through a treatment plant.
Arthur Hagen
2005-01-25 00:18:35 UTC
Permalink
Post by Richard Bos
Note that it is not the program itself which is at fault here. The bug
has been found; there is an error in the database. The road from
(IIRC) Haugesund to Oslo is one-way, which it shouldn't be. It's a
GIGO case.
There must be WAY more than one road that's marked wrong, cause there's
plenty of roads leading from Haugesund to Oslo (or Trondheim, or almost
anywhere east of the mountains). The same thing happes for a lot of
different routes as long as source=Haugesund and destination=(east of the
water divide), no matter how different the resulting routes should have
been.

In this case, I think there must be a combination of garbage data *and*
garbage code.

Regards,
--
*Art
Richard Bos
2005-01-25 20:31:22 UTC
Permalink
Post by Arthur Hagen
Post by Richard Bos
Note that it is not the program itself which is at fault here. The bug
has been found; there is an error in the database. The road from
(IIRC) Haugesund to Oslo is one-way, which it shouldn't be. It's a
GIGO case.
There must be WAY more than one road that's marked wrong, cause there's
plenty of roads leading from Haugesund to Oslo (or Trondheim, or almost
anywhere east of the mountains).
That does not follow. I've yet to see a travel planner that includes all
possible B-roads. All that's needed is for there to be only one road
east out of Haugesund, or one road over a crucial pass, and that one
road to be marked wrong. Basically, by (out of necessity) leaving out
non-viable roads from the start, you've created a single point of
failure.

Richard
Arthur Hagen
2005-01-25 23:54:55 UTC
Permalink
Post by Richard Bos
Post by Arthur Hagen
There must be WAY more than one road that's marked wrong, cause
there's plenty of roads leading from Haugesund to Oslo (or
Trondheim, or almost anywhere east of the mountains).
That does not follow. I've yet to see a travel planner that includes
all possible B-roads. All that's needed is for there to be only one
road east out of Haugesund, or one road over a crucial pass, and that
one road to be marked wrong. Basically, by (out of necessity) leaving
out non-viable roads from the start, you've created a single point of
failure.
In this case, I can't see how there could be a single point of failure that
can be determined, as the problem persists when going from Haugesund to
several geographically remote places, while you can go from Haugesund to
closer towns even after where a single point of failure would have been.

Examples:
Haugesund -> Haukeligrend FAIL
Haugesund -> Sauda WORKS
Haugesund -> Mandal FAIL

If looking at the reverse routes, you'll see that the first two share the
same route for a substantial part on the Haugesund side, while the last one
veers off almost immediately. There must be more than one point of failure.

Regards,
--
*Art
Len Oil
2005-01-25 01:28:47 UTC
Permalink
Post by Richard Bos
Note that it is not the program itself which is at fault here. The bug
has been found; there is an error in the database. The road from (IIRC)
Haugesund to Oslo is one-way, which it shouldn't be. It's a GIGO case.
I once tried to get a route from Sheffield to Edinburgh using both
MapQuest and MultiMap on-line routefinders. One of them told me that I
couldn't do it (there seemed to be a gap, both ways, in the routes
crossing a line roughly equivalent to the Scottish border, when I tried
shorter routes in-between to find the problem area) and the other told
me it was only 100 miles instead of around 200.

(Probably all fixed now, it was a long time ago, but I've not had to
plan that particular route since...)
Arthur Hagen
2005-01-25 00:07:26 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
You don't even have to go to Trondheim; Geilo will suffice.
I'd say Haukeligrend, but then you'd miss two capitals along the route.

Regards,
--
*Art
Flesh-eating Dragon
2005-01-25 01:49:56 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
Messages pointing out this observation are circulating all over the
Internet; my first encounter was here:

http://itre.cis.upenn.edu/~myl/languagelog/archives/001810.html

Adrian.
Pudde Fjord
2005-01-25 07:18:03 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
What I find most interesting is the detailed instructions to which roads
to use, when the underlying fault makes a farce out of it all.

And the fact that the error still exist, even after people have pointed
out where the error lies. I presume that M$ have been notified...

And this is not the first similar mistake I've seen, at a norwegian
route finder, some time ago, they had not connected two pieces of road
in the mountains, so all routes which should go that way were routed
around that point. But they fixed the problem rather quick.

Pudde.
Arthur Hagen
2005-01-25 14:09:53 UTC
Permalink
Post by Pudde Fjord
And this is not the first similar mistake I've seen, at a norwegian
route finder, some time ago, they had not connected two pieces of road
in the mountains, so all routes which should go that way were routed
around that point. But they fixed the problem rather quick.
I have the same problem with Microsoft Streets and Trips. It tries at all
costs to avoid a certain road, which is the preferred road to get to where I
live. It's the same road as I live on, actually. So it tells people to go
through a nearby city. If you block that city off, it will suggest going
through another nearby city. Block that one off too, and it sends you on a
huge 100 km+ detour for a 10 km trip. The interesting thing is that if you
continue to block pieces of road or intersections that obviously are wrong,
it *will* eventually send you the right route, which means the correct route
*is* there, it's just not being used except as a last resort.

Of course, sending corrections to Microsoft for their MapPoint based
products is slightly less productive than not sending it. It's now more
than four years since I complained about a different issue (MS would have me
drive through a building on a non-existing road that hasn't been there for
at least 20 years according to locals), but it still hasn't been fixed.

Regards,
--
*Art
Paul Cooke
2005-01-25 17:12:10 UTC
Permalink
Post by Arthur Hagen
Of course, sending corrections to Microsoft for their MapPoint based
products is slightly less productive than not sending it.  It's now more
than four years since I complained about a different issue (MS would have
me drive through a building on a non-existing road that hasn't been there
for at least 20 years according to locals), but it still hasn't been
fixed.
probably still using the original routing database from when the product was
a DOS product well before Microsoft got their sticky mits on it...

yet another product which has basically stagnated ever since Microsoft drove
out the originators... All they've done since is add pretty graphics and
added some animations to the "searching" wait boxes.

As they say up north... You can polish it as much as you like, but a
polished turd is still a turd...
Bruce Richardson
2005-01-25 23:02:24 UTC
Permalink
Post by Pudde Fjord
What I find most interesting is the detailed instructions to which roads
to use, when the underlying fault makes a farce out of it all.
But of course. Computers have no judgement or sense of context.
Thus they cannot detect the obvious error and carry on. The developers
can add extra rules but until such time as they can program genuine
awareness into the application, these errors will occur. In fact, the
more rules, the greater the complexity and thus, in the continued
absence of judgement, the greater the likelihood of truly bizarre
errors.
Post by Pudde Fjord
And the fact that the error still exist, even after people have pointed
out where the error lies. I presume that M$ have been notified...
Almost certainly. However, they may be having some trouble properly
tuning the application to correct this problem. Even a temporary
measure like hard-coding a special route for this one example is
something that has to be tested - and not on a live system.
--
Bruce

The ice-caps are melting, tra-la-la-la. All the world is drowning,
tra-la-la-la-la. -- Tiny Tim.
Paul Cooke
2005-01-26 17:22:00 UTC
Permalink
Post by Bruce Richardson
Post by Pudde Fjord
What I find most interesting is the detailed instructions to which roads
to use, when the underlying fault makes a farce out of it all.
But of course. Computers have no judgement or sense of context.
Thus they cannot detect the obvious error and carry on. The developers
can add extra rules but until such time as they can program genuine
awareness into the application, these errors will occur. In fact, the
more rules, the greater the complexity and thus, in the continued
absence of judgement, the greater the likelihood of truly bizarre
errors.
Post by Pudde Fjord
And the fact that the error still exist, even after people have pointed
out where the error lies. I presume that M$ have been notified...
Almost certainly. However, they may be having some trouble properly
tuning the application to correct this problem. Even a temporary
measure like hard-coding a special route for this one example is
something that has to be tested - and not on a live system.
there is an error in the road net in the starting town... zoom in close and
you'll notice that a stretch of dual carriageway has a short stretch of
normal road in it. That short stretch is marked as one-way only which is
where the poor thing gets confused... But there is a problem with the
routing routine as it cannot get back to what would be a more obvious route
bypassing this one-way stretch.
Joerg Neidig
2005-01-25 13:28:13 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
Well, you think driving through all of Europe is bad...

I did all the steps and got the following message:
"Routes between North-America and Europe are not supported"

Jörg
Beta Aquilae
2005-01-26 03:12:35 UTC
Permalink
Post by Joerg Neidig
Well, you think driving through all of Europe is bad...
"Routes between North-America and Europe are not supported"
Jörg
Perhaps just as well. The software might suggest the Trans-Siberian
route, with a detour past the various peninsulas (peninsulae?) on the way.
--
"I do not know how many of us it will take. But we must dream it, and if
enough of us dream it, then it will happen...
Dreams change the world."
--Neil Gaiman, A Dream of A Thousand Cats
Elliott Grasett
2005-01-26 03:28:17 UTC
Permalink
1. Go to MapPoint <http://mappoint.msn.com/DirectionsFind.aspx>
2. In the Start section, select "Norway" from the listbox and enter
"Haugesund" into the "City" field
3. In the End section, select "Norway" from the listbox and enter
"Trondheim" into the "City" field
4. Click on "Get Directions" ...
5. Laugh, Snigger, Cry
Will a Eurail pass handle this trip?
k***@gmail.com
2005-01-26 22:32:48 UTC
Permalink
They are getting a lot of free attention over it ;-) I asked a product
manager at Microsoft who responded -

http://www.mp2kmag.com/mappoint/discussion/viewtopic.asp?p=12088#12088
Eric
Darin Johnson
2005-01-27 00:31:23 UTC
Permalink
Post by k***@gmail.com
They are getting a lot of free attention over it ;-) I asked a product
manager at Microsoft who responded -
That explanation doesn't make sense. Even if that road did not exist
at all, the router should be able to find alternate routes that don't
go so far out of the way. And even if there are problems with that
road, why would the router decide to take the ferry to the UK instead
of Denmark or Sweden (given no travel time or distance was given for
the ferry, maybe it concluded that this leg took zero time)? And the
program can correctly get from Trondheim to Haugelund in the reverse
direction. Also it can find a route from Haugesund to Førdesfjord
(about 10km), and it can find a proper route from Førdesfjord to
Trondheim, it just can't tie these two routes together.

So it's not bad map data, but bad routing algorithms causing this
problem. To be fair, routing problems are very difficult to solve.
Even more difficult to solve in the amount of time required for a
quick web lookup. So everybody uses an approximation algorithm.
There is an higher performance (ie, polynomial time) algorithm that
can find a route that is twice the optimal path or better, but even
that probably takes too much time to run for an impatient web user.
So these programs have to take shortcuts. One of these shortcuts
apparently has some interesting worst case behavior.

When these programs were new, and there wasn't a lot of map data, I
got an interesting result when I asked it to calculate directions from
my home to my parent's house who lived in a small town. I got
detailed (but bad) directions about how to get from my house to the
main freeway. And then the instructions read something like "continue
south for 112.3 miles until you reach 333 Your Road, Yourtown, CA".
--
Darin Johnson
My shoes are too tight, and I have forgotten how to dance -- Babylon 5
Pudde Fjord
2005-01-27 21:11:34 UTC
Permalink
Microdofts reply (snipped): "In this case one road segment was
attributed incorrectly in the topology, causing the routing algorithm to
ignore that road and therefore caused the error in calculating the
route. We are currently working a fix, and we expect it to be available
in early February."

(Why didn't you include this, since it's only text...?)
Post by Darin Johnson
That explanation doesn't make sense. Even if that road did not exist
at all, the router should be able to find alternate routes that don't
go so far out of the way. And even if there are problems with that
road, why would the router decide to take the ferry to the UK instead
of Denmark or Sweden (given no travel time or distance was given for
the ferry, maybe it concluded that this leg took zero time)?
I think this is the main problem:

The ferry stretches are credited with no driving (and ferry costs and
time is not taken into account). In addition, no effort is being made o
keep a trip inside one country only, when the starting and end points
are. (This will not always be the *shortest* route, but may save a few
border crossings.)
Post by Darin Johnson
So it's not bad map data, but bad routing algorithms causing this
problem. To be fair, routing problems are very difficult to solve.
Even more difficult to solve in the amount of time required for a
quick web lookup. So everybody uses an approximation algorithm.
There is an higher performance (ie, polynomial time) algorithm that
can find a route that is twice the optimal path or better, but even
that probably takes too much time to run for an impatient web user.
So these programs have to take shortcuts. One of these shortcuts
apparently has some interesting worst case behavior.
There have to be several "shortcuts" in the program to make it this bad.
In addition to bad road/map data.
Post by Darin Johnson
When these programs were new, and there wasn't a lot of map data, I
got an interesting result when I asked it to calculate directions from
my home to my parent's house who lived in a small town. I got
detailed (but bad) directions about how to get from my house to the
main freeway. And then the instructions read something like "continue
south for 112.3 miles until you reach 333 Your Road, Yourtown, CA".
I have seen something like this, but the assumption is, when you are on
the main roads, and don't have to turn off anywhere, additional
instructions are just confusing.

Of course, if the *end* point is *on* the main road, there should be a
more detailed instruction where to turn off the road.

Pudde.
Darin Johnson
2005-01-28 00:59:30 UTC
Permalink
Post by Pudde Fjord
(This will not always be the *shortest* route, but may
save a few border crossings.)
These routing programs seem to have naive assumptions about
driving times. A border crossing should indeed count for
a delay and be a part of the graph. I've seen routes in the
past that were probably the shortest path, but used a lot of
slow residential roads instead of getting to the freeway first.
Post by Pudde Fjord
Post by Darin Johnson
And then the instructions read something like "continue
south for 112.3 miles until you reach 333 Your Road, Yourtown, CA".
I have seen something like this, but the assumption is, when you are
on the main roads, and don't have to turn off anywhere, additional
instructions are just confusing.
Except that the end point was not on that freeway, and all of the
necessary instructions that were important were left out. Even the
map that was drawn just had a perfectly straight line to the
destination without bothering to follow roads or topography.

I looked at the same route yesterday with MapPoint, presumably with
smarter software and it had plenty of confusing instructions.
Everytime the "official" names of roads would change slightly it had
a new instruction. Maps.yahoo.com was far more useful than MapPoint,
and includes exit numbers where appropriate, more common names
instead of county road numbers, and fewer instructions about name
changes.
--
Darin Johnson
Laziness is the father of invention
Arthur Hagen
2005-01-28 03:15:51 UTC
Permalink
Post by Darin Johnson
Post by Pudde Fjord
(This will not always be the *shortest* route, but may
save a few border crossings.)
These routing programs seem to have naive assumptions about
driving times. A border crossing should indeed count for
a delay and be a part of the graph.
Within Europe? No. The Maastricht treaty ensures open borders, and it's
usually not much harder to cross a border within Europe than it is to cross
a state line in the US. Especially so in Scandinavia, where the countries
have enjoyed mutually open borders for centuries (except during war time).
Post by Darin Johnson
I've seen routes in the
past that were probably the shortest path, but used a lot of
slow residential roads instead of getting to the freeway first.
Also keep in mind that the driving patterns in Europe differ between
countries, and are *vastly* different from how people drive in the US.
When a light turns green, you don't leisurely start rolling like in the
US -- the light turns red+yellow before turning green, so you have advance
warning. When it's green, you better step on it.
A New York taxi driver wouldn't last for ten minutes in many European
cities.

As for the speeds, you'll have to look hard and long to find anywhere in
Norway where you can drive faster than 90 km/h (~55 mph), and most highways
are 80 km/h (~50 mph) or lower. Freeways don't exist, and the highways are
*twisty*. For the correct route from Haugesund to Trondheim, you'd have a
hard time keeping an average speed of much more than 60 km/h (~37 mph).
In Germany and Italia, though, you have Autobahns and Autostradas without
speed limits in the speed lanes. Of course, that doesn't mean unlimited
speed, so a map program shouldn't assume it takes 0 minutes.

Regards,
--
*Art
Puck
2005-01-28 04:27:05 UTC
Permalink
Post by Arthur Hagen
A New York taxi driver wouldn't last for ten minutes in many European
cities.
In my expirience, many European cities wouldn't last ten minutes with a New
York taxi driver in them. We are talking of people who, when you tell them
you want to go to Yankee Stadium, will ask "Infield or Outfield?"
Arthur Hagen
2005-01-28 05:11:03 UTC
Permalink
Post by Puck
Post by Arthur Hagen
A New York taxi driver wouldn't last for ten minutes in many European
cities.
In my expirience, many European cities wouldn't last ten minutes with
a New York taxi driver in them. We are talking of people who, when
you tell them you want to go to Yankee Stadium, will ask "Infield or
Outfield?"
Having been to both New York City and several dozen European cities, I have
to say there's no contest if comparing to the worst of Europe. The NYC taxi
driver may be ruthless or even reckless when switching lanes, run yellow
lights and even cut curbs, but they don't accellerate/decellerate/turn like
European drivers do. For one thing, their oversized/underpowered cars
aren't capable of it, and secondly, the rest of the traffic doesn't
cooperate.
--
*Art
Martyn Clapham
2005-01-29 00:12:10 UTC
Permalink
Having been to Norway once a year for the past 3 years some of this
didn't quite ring true so I checked on the web at

http://www.norway.org.uk/policy/europe/schengen/schengen.htm

among other places to make sure I'd got it right.
Post by Arthur Hagen
Post by Darin Johnson
Post by Pudde Fjord
(This will not always be the *shortest* route, but may
save a few border crossings.)
These routing programs seem to have naive assumptions about
driving times. A border crossing should indeed count for
a delay and be a part of the graph.
Within Europe? No. The Maastricht treaty ensures open borders, and it's
usually not much harder to cross a border within Europe than it is to cross
a state line in the US. Especially so in Scandinavia, where the countries
have enjoyed mutually open borders for centuries (except during war time).
The website says the Nordic Passport Union only came into existence in
1954 ( presumably before that there were informal arrangements ) and
when the Schengen expanded to include Denmark, Finland and Sweden,
Norway had to join as well.

So Norway is part of the Schengen zone while the UK isn't therefore
there are passport controls at the ferry ports ( also Customs controls
as we are in the EU and they aren't ).
Post by Arthur Hagen
As for the speeds, you'll have to look hard and long to find anywhere in
Norway where you can drive faster than 90 km/h (~55 mph), and most highways
are 80 km/h (~50 mph) or lower. Freeways don't exist, and the highways are
*twisty*. For the correct route from Haugesund to Trondheim, you'd have a
hard time keeping an average speed of much more than 60 km/h (~37 mph).
In Germany and Italia, though, you have Autobahns and Autostradas without
speed limits in the speed lanes. Of course, that doesn't mean unlimited
speed, so a map program shouldn't assume it takes 0 minutes.
The freeway network is now comparatively extensive, I think it's well
into hundreds of km. I thought we'd got the coach up to the top speed
allowed by the limiter ( 100km/h ) but maybe not. Also a lot of the
twisty roads are being replaced by tunnels, we've seen the progress of
one that will cut off at least 5 miles of road and the Laerdal tunnel (
24km / 15 miles ) which avoids a bad mountain pass has been opened, as
has the tunnel that cuts off the stretch over the mountain at the
Stalheim hotel ( something like 15 hairpins on a 1 in 4 road! ).

I think the average speed for the route above will probably have gone up
by at least 10 kmh and may be even 10 mph.

Mart.
--
Livejournal at http://www.livejournal.com/users/pendlemac
Caroline's afpersonal God of Misunderstandings & afpSlave to CCA.
IM stuff :- ICQ: 246971821 AIM or MSN: pendlemac all via gateways to
***@myjabber.net
Arthur Hagen
2005-01-29 02:16:36 UTC
Permalink
Post by Martyn Clapham
The freeway network is now comparatively extensive, I think it's well
into hundreds of km. I thought we'd got the coach up to the top speed
allowed by the limiter ( 100km/h ) but maybe not. Also a lot of the
twisty roads are being replaced by tunnels, we've seen the progress of
one that will cut off at least 5 miles of road and the Laerdal tunnel
( 24km / 15 miles ) which avoids a bad mountain pass has been opened,
as has the tunnel that cuts off the stretch over the mountain at the
Stalheim hotel ( something like 15 hairpins on a 1 in 4 road! ).
I think the average speed for the route above will probably have gone
up by at least 10 kmh and may be even 10 mph.
Ah - good to know. It's been a long time since I've been to Norway, as you
could tell :-)

Regards,
--
*Art
peachy ashie passion
2005-01-28 03:39:18 UTC
Permalink
Post by Darin Johnson
Post by Pudde Fjord
(This will not always be the *shortest* route, but may
save a few border crossings.)
These routing programs seem to have naive assumptions about
driving times. A border crossing should indeed count for
a delay and be a part of the graph. I've seen routes in the
past that were probably the shortest path, but used a lot of
slow residential roads instead of getting to the freeway first.
oh! I took one of those maps last night.

Spent over an hour wandering through residential neighborhoods to
get to a place that was less than 10 miles from my house. Once I
found the place, it took me less than 10 minutes to get home.

I was NOT a happy camper
--
"Natives who beat drums to drive off evil spirits are objects of
scorn to smart Americans who blow horns to break up traffic jams."
~ Mary Ellen Kelly

AFPslave to Mistress Stacie
***@verizon.net
Puck
2005-01-28 04:30:04 UTC
Permalink
Post by peachy ashie passion
I was NOT a happy camper
It is *purple* to be *squishy* when campers are *dancing*

Bonus points to anyone who knows what classic pc game that is a reference
to.


--
Puck (onstage): I am that merry wanderer of the night!
Peaseblossom (in audience): I am that merry wanderer of the night, indeed!
"I am that giggling-dangerous-totally-bloody-psychotic-menance-to-life and
limb, more like."
-Neil Gaiman
Thomas Zahr
2005-01-28 17:56:32 UTC
Permalink
Pudde Fjord posted:

... navigation software
Post by Pudde Fjord
In addition, no effort is being made o
keep a trip inside one country only, when the starting and
end points are. (This will not always be the *shortest*
route, but may save a few border crossings.)
In some areas (Schengen) that is no longer a concern, and
around Aachen (Aix-la-chappelle) and Maastricht, if you are
not prepared to cross the odd border or three, you are
basically f*cked.
--
Ciao

Thomas =:-)
<http://www.zahr.de>
Richard Bos
2005-01-30 21:23:42 UTC
Permalink
Post by Pudde Fjord
Post by Darin Johnson
That explanation doesn't make sense. Even if that road did not exist
at all, the router should be able to find alternate routes that don't
go so far out of the way. And even if there are problems with that
road, why would the router decide to take the ferry to the UK instead
of Denmark or Sweden (given no travel time or distance was given for
the ferry, maybe it concluded that this leg took zero time)?
The ferry stretches are credited with no driving (and ferry costs and
time is not taken into account). In addition, no effort is being made o
keep a trip inside one country only, when the starting and end points
are. (This will not always be the *shortest* route, but may save a few
border crossings.)
That's still a database error, though. There should be a separate "road"
type for ferries, with a very low, possibly variable, traveling speed,
as well as a data type for border crossings where necessary.
If it isn't easily possible to add these road types, then _that_ is a
defect in the software, of course.

Richard

g***@tpg.com.au
2005-01-28 08:36:02 UTC
Permalink
On Thu, 27 Jan 2005 00:31:23 GMT, Darin Johnson <***@_usa_._net> wrote:

[Route generation]
Post by Darin Johnson
So everybody uses an approximation algorithm.
There is an higher performance (ie, polynomial time) algorithm that
can find a route that is twice the optimal path or better, but even
that probably takes too much time to run for an impatient web user.
So these programs have to take shortcuts. One of these shortcuts
apparently has some interesting worst case behavior.
Weird. Wouldn't it make sense to precalculate route-sections between
significant landmarks along major travel routes (highways, sea and air
routes) etc? Then any request for a route between X and Y would only need
to calculate in real time what the quickest way was to get from X to the
three (or six or whatever) closest pre-mapped landmarks, and ditto from Y,
and then look only at routes that went X-> landmark-close-to-X ->
precalculated shortest path -> landmark-close-to-Y -> Y. Do a quick sort
of the nine or thirty-six resulting route weights (distance, time etc) and
present the best one.

The best part about this is that any computers on this project can spend
their spare processing cycles precalculating finer and finer grids
(setting more and more known landmarks), making subsequent requests faster
and faster.

Of course, if a major route section goes down for whatever reason,
processing power will probably be diverted to recalculating the immedate
area around it, plus any precalculated routes that pass through that
section. Routes passing through that section which would normally be
handed out as shorted paths based on prior calculations will need to be
checked by a longer algorithm, but all other requests should be OK.

Not to mention that the finer the precalculation grid, the more route
sections will need to be updated per given time period, and the more
recalculation this will involve. Although admittedly, fewer sections will
be part of 'backbone' routes.

Plus, once a route section has been calculated, sanity algorithms could
check that it's working properly, by comparing it to previous data about
routes in the area, knowledge of nearby alternative routes sections, and
checking that the numbers for paths which would normally pass through that
section are now not completely out of whack.

For one thing, I'd like paths between major cities checked after any
recalculation of backbone sections in that city. Routes to the city, from
the city, from A to B via the city, and within the city along paths that
would normally include that segment. If the path lengths increase by a
significant amount (an appropriate percentage), they really need to be
manually checked. Another one is that if shortest path A-B plus shortest
path B-C is shorter than shortest path A-C, there's something really
wrong.

Even if a section of highway is washed away, there should be backroads
which can connect the two segments, or alternate highways. Travel between
close towns A and B may now be blown out because they have to go via town
C while the A-B highway is repaired, but travel (time, distance) from
faraway-town D to faraway-town E which normally went via the A-B segment
should not be increased by a huge percentage, even if the preferred route
totally changes (switching to the second of two very similarly weighted
interstate highways, for example).


Yay for unstructured rambling!

-SteveD
--
"Spherical people in frictionless vacuum" - Topi Saavaleinen
Steve Rogers
2005-01-28 18:12:12 UTC
Permalink
Post by g***@tpg.com.au
[Route generation]
Post by Darin Johnson
So everybody uses an approximation algorithm.
There is an higher performance (ie, polynomial time) algorithm that
can find a route that is twice the optimal path or better, but even
that probably takes too much time to run for an impatient web user.
So these programs have to take shortcuts. One of these shortcuts
apparently has some interesting worst case behavior.
Weird. Wouldn't it make sense to precalculate route-sections between
significant landmarks along major travel routes (highways, sea and air
routes) etc? Then any request for a route between X and Y would only need
to calculate in real time what the quickest way was to get from X to the
three (or six or whatever) closest pre-mapped landmarks, and ditto from Y,
and then look only at routes that went X-> landmark-close-to-X ->
precalculated shortest path -> landmark-close-to-Y -> Y. Do a quick sort
of the nine or thirty-six resulting route weights (distance, time etc) and
present the best one.
The best part about this is that any computers on this project can spend
their spare processing cycles precalculating finer and finer grids
(setting more and more known landmarks), making subsequent requests faster
and faster.
Of course, if a major route section goes down for whatever reason,
processing power will probably be diverted to recalculating the immedate
area around it, plus any precalculated routes that pass through that
section. Routes passing through that section which would normally be
handed out as shorted paths based on prior calculations will need to be
checked by a longer algorithm, but all other requests should be OK.
Not to mention that the finer the precalculation grid, the more route
sections will need to be updated per given time period, and the more
recalculation this will involve. Although admittedly, fewer sections will
be part of 'backbone' routes.
Plus, once a route section has been calculated, sanity algorithms could
check that it's working properly, by comparing it to previous data about
routes in the area, knowledge of nearby alternative routes sections, and
checking that the numbers for paths which would normally pass through that
section are now not completely out of whack.
For one thing, I'd like paths between major cities checked after any
recalculation of backbone sections in that city. Routes to the city, from
the city, from A to B via the city, and within the city along paths that
would normally include that segment. If the path lengths increase by a
significant amount (an appropriate percentage), they really need to be
manually checked. Another one is that if shortest path A-B plus shortest
path B-C is shorter than shortest path A-C, there's something really
wrong.
Even if a section of highway is washed away, there should be backroads
which can connect the two segments, or alternate highways. Travel between
close towns A and B may now be blown out because they have to go via town
C while the A-B highway is repaired, but travel (time, distance) from
faraway-town D to faraway-town E which normally went via the A-B segment
should not be increased by a huge percentage, even if the preferred route
totally changes (switching to the second of two very similarly weighted
interstate highways, for example).
As I am working on a route miles calculator at the moment for a friend of
mine who is one of the strange breed known as "Bashers" [1]. All I can say
is that it is a pain in the neck to do especially when waypoints are
included or as in this case all possible diverging route points are
included. The possible route database for the information becomes extremely
large very quickly the greater the distance to be travelled although the
final calculations are simpler in this case as speed is not an issue just
distance and time.

Admittedly what I've been asked to do is easier than for a road journey, but
still all diverging points as well as the start and finsh points are needed
in the correct order (not something that can be gauranteed to be inputted by
a Basher with a laptop on a tour).

Steve
----------------------------------------
Furthering Anglo-American relations ;-)
----------------------------------------

[1] Basher - someone who spends their free time travelling on trains hauled
by a particular class or type of locomotive either as part of a scheduled
service or as an organised tour, getting extremely drunk and doing other
strange things.
Loading...