I had a nomination for a horse-riding school rejected by the community after about a week due to “mismatched location” and “other rejection criteria”. I made a schoolboy error by submitting the wayspot at and taking photos of the sign, but the title, description and supporting information just referred to the riding school itself, which is plotted in a different location to the sign on Google Maps.
I corrected my nomination to add explicit reference to the sign in the title and description and explain how that was important in highlighting the school and guiding visitors to it. I made no changes to the photos. However, this time to my surprise the submission got rejected by ML and the reason “rejection criteria” gives me little to go on.
Do others have experience of nominations making it past ML, getting rejected by the community and then being resubmitted and not making it past ML? Does ML often display inconsistency when presented with the same or very similar information, or would it continue to reject now without drastic changes to photos and/or text?
I wanted to check this before posting any specifics in the Nomination Support Category.
The way Machine Learning works is: A certain percent of randomly chosen cases go straight to manual review (community or Niantic). Those results are fed back to the “machine”, for learning. IOW it is incorrect to assume that anything that went to the community passed thru AI first. There would be no learning if AI got everything up front. Well, there would be some learning because AI passes on the cases it’s unsure about - but the logic it’s sure about could never change or be refined.
So let’s assume your first nomination bypassed AI and went straight to the community. Then your second was picked up by AI.
Could be the word “school”.
Or that the sign doesn’t seem to have pedestrian access.
Or that a sign typically isn’t a great place to socialize, explore, or exercise.
I’d resubmit of the actual stables, maybe with the sign as supporting. But of these two - if you’re going to appeal one, I’d choose the first one.
For something like this I would put this as a description:
A established family-run equestrian business that offers riding lessons and hacking for all ages and abilities.
For the supporting information i would put there website and talk about what it promotes and how.
I wouldn’t use the word school as the AI will probably reject it for that so for the title I would put something like Stonehill Riding Centre or Stonehill Horse Riding Centre
In USA, that would be difficult to pass because many ranches are privately owned with a farm house, barn, and armed to the teeth. It would not be safe.
In Hebron, New York in 2023, a 20-year old woman was killed by an overzealous, iron-lickin’ landowner when her friends mistakenly turned to the wrong dark, rural driveway. Kevin went to prison for 25+ years.
Well, the sign is located at the corner of a field, which is adjacent to a public footpath and cycleway along a main road and its junction with a public bridleway. The Riding School is accessed off the same bridleway and whilst the land is private, they clearly want to encourage customers to visit, as that is necessary to sustain their business.
I have never tried wandering around farmland in the USA, but in the UK, legislation allows for public rights of way to become established over time. Once in place, the public has a right to pass and re-pass. Even if a UK rambler inadvertently strayed from a public route onto a private track, they wouldn’t expect to come under a barrage of small arms fire.
Thank you for the advice. I have resubmitted the nomination with the following changes:
included “horse” and replaced “school” with “centre” in title
moved more of the supporting information into the description
increased emphasis on the quality of the countryside and the public right of way network, which merits exploration and included a link to a website with detail about the business
I have kept the wayspot location as the sign for now, because there is no signage at the business itself and there is a dwellinghouse on the site so there is a danger of the ML thinking it is a residential premises.
