Switch your computer off for a couple of days and go for a ride on your bike. 
Go out and see if there has been an office block built on a farm near you.
3 Likes
British planning rules are getting a little odd, thatās for sure, although youād probably be more likely to see a housing estate, especially if it was on a flood plain. 
6 Likes
YESSS! I TOTALLY AGREE WITH THAT!
Thatās the Needles lighthouse at Alum Bay on the Isle of Wight. Been there myself!
Great idea.
1 Like
This is me, like on christmas eve, waiting for Santā¦ the update to comeā¦ not yetā¦
And a helicopter to go with it 
1 Like
Ahhh yes. Definitely
Thatās a nice post with a lot of informationā¦howeverā¦since the UKs countryside was absolutely fine before the Japan world update, Iād say blackshark and Asobo havnāt improved the ML system and the time and effort spent so far have just made it worse.
So, while you say there is no other answer than time, I put it to you that unless we complain most bitterly about things getting downgraded, time is not on our side.
3 Likes
Ha ā¦ It would never work!
The whole forum would just be a continuous series of flags although I suppose for some variety we could provide different colours depending upon how grumpy the response was 
2 Likes
We are talking about the other way around procedure already, ie what ever added extra as PG into MSFS2020 to appear back in Bing Maps !
Hmm, I donāt think it was absolutely fine. We had multi story office buildings in the countryside from day 1.
4 Likes
I agree, those office block have always been there. Their size increased and some were more like tower blocks after the Japan update, however there was a community fix and then an official fix for that. But the office block have always been there. Thereās one right near where I live, it should be a farm, it was there on day 1 and still there now. Hopefully, after the update tomorrow it will be a farm!! Got everything crossed.
3 Likes
Well I live in the middle of Europe these days and my house looks like a pigsty rather than a nice office block/apartment building, so careful what you wish for lol 
3 Likes
Talking of farms, Iāve noticed that a friendās house, which is a thatched cottage, is rendered as a flat area of straw. Are there any haystacks, or are they flat too?
I donāt think any of it comes from Bing Maps - they use mapping providers. Whatās interesting though and I think the point you are making - this has clearly gone through Bing Maps team as they have helped with the integration into MSFS but oddly without they themselves incorporating the data into there own service?
I mean the Japanese content put into MSFS recently still isnāt in Bing Maps so I donāt understand maybe they themselves donāt have a budget for that but the inclusion in MSFS comes out of the Xbox pot? Not that I care 2hoots as long as MSFS gets updated 
Maybe they are adding these maps into beta Bings Maps service which hasnāt been published yet as it needs a lot more checks and balances to align with the vector data.
1 Like
Yes, but you get Santa Claus earlier than we do in the UK.
2 Likes
Are we able to form clusters or clubs in here ?
As FlatEarthers, Seasoners, AvionicHeads, VRgeeks, StutterSlayers, WaterMaskers etc. So that we can upvote en bloc to get our wishes at top of the lists.
3 Likes
FlatEarthers!! 
1 Like
Remember to do a full backup tonight.
1 Like
4pm UK time usually.
Itās pretty commonplace for a run to introduce regressions and unexpected results. For something like farmland in Belgium and UK producing wildly different results, well, I didnāt build it, but what is likely is that the data is different in some way between those locations. Noise in the imagery, low resolutions, color-banding, missing or incorrect meta-data, or even the perspective could have thrown off the system.
The ML systemās decision making is in many ways a black-box - itās not really possible to ālook insideā and get a definitive answer about why the system made the decision it did. Itās left to the engineers to infer this from the data, the results, and their specific understanding of the characteristics of ML system they are using.
As for why it would be right, and then suddenly wrong later, well, a run isnāt always going to produce identical results - and the japan update may have introduced other un-reported changes to global data. Itās also possible they could have made a change to the training data or ML configuration that caused the regression.
Not a very satisfying answer I know; ML shares some of the idiosyncrasies that small children have. You can tell a child a dozen different ways ādonāt touch this, itās HOTā, and what do they do? They touch the pan, even though just 10 minutes earlier, they were happily following these instructions re: the fireplace. 