Featured Wayspot selection algorithm
- Featured Wayspots appear from the level 6 cell you are active in.
- The selection of the Wayspots is based on the Hilbert curve through the level 6 s2 cell, with the three featured Wayspots being the first three Wayspots in sequence on the curve.
Each week three new Wayspots appear as 'Featured Wayspots'. These Wayspots are typically around your play area, but how are they selected? Is it random or is there a pattern?
I had observed for a long period of time that the featured Wayspots in my area came from the far south of the metropolitan area I lived in. I found this curious, because statistically the far greater number of approved Wayspots should be from closer to the city centre. It wasn't until a Reddit thread by Mantolwen (https://www.reddit.com/r/NianticWayfarer/comments/gaccwx/featured_wayspots_location_s6_cell_theory/) which made the observation that Featured Wayspots typically came from a corner of a level 6 s2 cell that I started looking into it further.
I was aware from prior reading that the s2 cells follow a space-filling Hilbert curve. (If you want more technical reading about this, check out http://blog.christianperone.com/2015/08/googles-s2-geometry-on-the-sphere-cells-and-hilbert-curve/ or https://s2geometry.io/devguide/s2cell_hierarchy.html). I was curious whether this observation was in any way related to the Hilbert curve within Level 6 cells.
When you select an area using regioncoverer (http://s2.sidewalklabs.com/regioncoverer/), it lists in the bottom left the IDs of the s2 cells that have been selected, with those cells listed in alphabetical order. For example, these are the 16x level 8 cells that make up my level 6 play area, with the cell IDs superimposed. (To tell which cell is which, I just delete one of the cell IDs and hit enter, then watch which cell disappears from the map.)
Joining the cells in order by cell IDs results in the Hilbert curve. If following this pattern in sequence, this fitted with my observations that all of my featured Wayspots consistently come from the south of the city.
Furthermore, when I looked in detail at the order of the current three featured Wayspots, and found the ID of their level 17 cells the results were:
- Wayspot #1: 2a3279551c
- Wayspot #2: 2a32795fc4
- Wayspot #3: 2a32795fdc
So not only were the three Wayspots from the general area that was suggested by the Hilbert curve, but the actual order of appearance on the website matched their sequence of appearance on this line. (Note: Level 17 cells were selected because Wayspot #2 and #3 were very close and shared a level 16 cell.)
Based on these observations, my hypothesis is:
- Each Wayspot has an internal ID corresponding to some level of s2 cell.
- Each week, all Wayspots approved from a particular level 6 cell are listed in order of s2 cell ID.
- The first three Wayspots from this appear as Featured Wayspots.
There's no significant gameplay effects, this is more of an 'oh well that's interesting' discovery. Although I am curious who will be the first person to use this information to achieve a rick-roll through the titles and/or descriptions of the three featured Wayspots.
In case anyone is curious, there are how this applies to some of places around the world...
I've noticed this since the beginning of the featured wayspot, and I can confirm that your investigation matches to my local area (Melbourne), as per your image. In Melbourne, the featured wayspots are consistently in the south-eastern suburbs, usually around Langwarrin, Clyde, Frankston and Pakenham. Some of these are growth corridors, so there are more estates and areas being built, which brings more people and more potential nominations.
Unfortunately, a lot of the nominations from those areas are very poor in quality, so they tend to get featured. This gives a bad impression to the rest of the reviewing-base as its not an indicator of quality, but rather a random selection of what is often the worst possible nominations to highlight. It's ironically funny.
Holy ****, I spitballed that this was the reason a couple days ago but didn't investigate the traversal algorithm again to verify. Thanks for taking a close look at this!
I just want to say thank you for figuring this out. Really fascinating!! I noticed the featured wayspots all came from the same general geographic area (that is actually pretty far away from me).
Any thoughts on how this might apply to the order in which nominations get voted on? Or edits? That would be significantly more difficult to tell for sure.
Interesting, and your map tallies with results from my locality.
Holy ****, it's amazing how you figured this out! Matches up with the featured wayspots in my area.
This is awesome! I'm really glad my observations inspired more in depth research. It's a shame Niantic have chosen this method without filtering out some lower quality/common wayspots. It would be nice to see properly interesting stuff.
I have no idea why they didn't make it show the three Wayspots with the highest average scores from your L6 cell. That seems like it would do a much better job of highlighting high-quality nominations.
I would have liked to see the high scores too. Would keep the examples much more high quality.
Somewhere a long time I saw the idea floated that they were the highest voted submissions- made me feel confused about why some of the pretty awful ones were somehow scoring better than mine/others in my area. Makes me feel better to know that wasn't the case lol and it makes a WHOLE lot more sense.
So each wayspot has its
S2Point(see https://s2geometry.io/devguide/s2cell_hierarchy.html#s2cellid) and each cell is a range of 64bit integers and thus the selection algorithm is
SELECT TOP 3 S2Point FROM NianticPoI WHERE S2Point > CellMin and S2Point < CellMax and ApprovalDate > last few days
Just adjust CellMin and CellMax for each cell.
Shhh, don't let Niantic know we've figured it out!
Everyone I've shared this with loves the read! Thanks for the investigative work. Now, I wonder what the HP:WU rules are... 🤔
Well that explains why **** like this appears as a featured wayspot in my area.
Works 100% for me this week. However, in the past I have seen new stops from the opposite corner of the cell. This was right at the beginning of wayfarer. (My new stops used to be featured all the time, but sadly not anymore).
Thanks for the detailed work. But how do I know which cell is the starting point and which will be the ending point?
UK player here, I was always wondering why Featured for me are all exact the same town - so thanks for the observation.