Feb 9, 2023·edited Feb 9, 2023Liked by Mike Conlow
If you check the parcel data on that Iowa address in Zillow (the 2nd closest zoom level is the best for checking what the parcel data looks like) you can clearly see that the parcel data BARELY misses the main house so the algorithm goes:
1) Ok, I have 3 rectangles on the parcel for 1276 160th St, Corning, IA 50841 I will choose the most houselike object of the 3.
2) Alright, I found the house on the 1276 160th St, Corning, IA 50841 parcel. Let me move over to the left where there is no defined parcel. This rectangle that is not within a parcel is clearly a very houselike object. I have a record of mail going to 1154 160TH ST
CORNING, IA 50841 so it is very likely that this non parceled extremely houselike object is 1154 160TH ST CORNING, IA 50841.
If you drop the little guy in google street view mode you can actually check their 911 address post showing that only the address 1276 is associated with that driveway.
That's why I think the algorithm favors two separate scenarios:
1) heavily parceled up land with many house-like objects and a history of mailing addresses.
2) poorly parceled up land with a history of mailing addresses and many house-like objects.
Areas that get undercounted would be big well defined parcels that get classified as "single home" but multiple families have put up houses so it kind of 'eenie meenie minie moe's" which house gets to count.
Mike - glad to see that you have all of this data to use for your explaination of exactly what I have been saying down here in Texas. But there is another way to fix this mess if you redefine the terms of accesability to broadband service, and that is to add the terms of "reliability" and "affordability" to the definition. As soon as you do that, most of the non-terrestrial providers get filtered out. Which is sort of the case down here in rural texas - no one is really "unserved" because satellite systems can serve you badly anywhere. And isn't it really affordability and reliability that prevent widespread adoption to broadband? BTW, these are measurable characteristics :) Herb Krasner (google me)
It should not be a surprise that in the best corporatocracy money can rent the data would be skewed. Which NGOs/NPOs are sponsoring OpenStreetMap flavored alternative efforts?
If you check the parcel data on that Iowa address in Zillow (the 2nd closest zoom level is the best for checking what the parcel data looks like) you can clearly see that the parcel data BARELY misses the main house so the algorithm goes:
1) Ok, I have 3 rectangles on the parcel for 1276 160th St, Corning, IA 50841 I will choose the most houselike object of the 3.
2) Alright, I found the house on the 1276 160th St, Corning, IA 50841 parcel. Let me move over to the left where there is no defined parcel. This rectangle that is not within a parcel is clearly a very houselike object. I have a record of mail going to 1154 160TH ST
CORNING, IA 50841 so it is very likely that this non parceled extremely houselike object is 1154 160TH ST CORNING, IA 50841.
If you drop the little guy in google street view mode you can actually check their 911 address post showing that only the address 1276 is associated with that driveway.
That's why I think the algorithm favors two separate scenarios:
1) heavily parceled up land with many house-like objects and a history of mailing addresses.
2) poorly parceled up land with a history of mailing addresses and many house-like objects.
Areas that get undercounted would be big well defined parcels that get classified as "single home" but multiple families have put up houses so it kind of 'eenie meenie minie moe's" which house gets to count.
Great analysis, thank you for sharing
Mike - glad to see that you have all of this data to use for your explaination of exactly what I have been saying down here in Texas. But there is another way to fix this mess if you redefine the terms of accesability to broadband service, and that is to add the terms of "reliability" and "affordability" to the definition. As soon as you do that, most of the non-terrestrial providers get filtered out. Which is sort of the case down here in rural texas - no one is really "unserved" because satellite systems can serve you badly anywhere. And isn't it really affordability and reliability that prevent widespread adoption to broadband? BTW, these are measurable characteristics :) Herb Krasner (google me)
It should not be a surprise that in the best corporatocracy money can rent the data would be skewed. Which NGOs/NPOs are sponsoring OpenStreetMap flavored alternative efforts?