I still find the Neighbors function exasperating. It may not be a particularly important feature, but it’s interesting enough for me to see what other people allegedly have tastes similar to mine. I send out friend requests mostly on this basis.
Here are some things I’ve observed. Bear with me.
ICM claims to determine one’s 25 “Neighbors” via a complex mathematical formula (https://www.icheckmovies.com/help/faq/#how-are-my-neighbors-calculated). And yet, simply visiting another person’s profile page automatically makes him/her a “Neighbor.” Why is this?? This wrecks the whole concept. Who says visiting a person’s profile means you feel an affinity with him? Maybe he just writes stupid comments and you wonder about his background. Another quirk I’ve observed that’s actually good: visiting an established Neighbor’s page makes his Neighbor value (your number of shared films, etc.) get immediately re-calculated. So, that bear this in mind as you continue reading.
OK, every few weeks (?), the Neighbors get re-calculated and you’re given a fresh set of 25 neighbors. I don’t know about other people’s experiences, but this list of neighbors is ALWAYS wrong for me. Every time I’m given a revised set of neighbors, there are five names at the top who don’t belong: users “siriuslymovies,” “Scheffel,” “TotalVertigo,” “Iain_S” and “Vinther1991.” They don’t belong, because I know from experience that visiting these five profiles makes their Neighbor value get re-calculated and, then, their “percentage” compatibility with me drops sharply. Do other people see these names over and over, or is it just me? The “siriuslymovies” factor is especially ridiculous, because ICM repeatedly concludes that I have 100% compatibility with him.
Beyond the problem of these five names who wrongly appear over and over, the rest of the names aren’t correct either. Because I’ve tested this. Since visiting a new profile makes that person become an automatic Neighbor with a corresponding percentage compatibility, I’ve found that some profiles should be in my top 25 and aren’t.
The above gibberish may be confusing, so let me just lay out some hard data.
Yesterday, the 25 Neighbors were re-calculated. And here is whom I was wrongly given. I’m not giving most of my Neighbors’ names, because it’s not necessary.
siriuslymovies: 100%
Scheffel: 51%
TotalVertigo: 46%
Iain_S: 43%
Vinther1991: 41%
User A: (further names and percentages don’t need to be cited)
User B
User C
User D
User E
User F
User G
User H
User I
User J
User K
User L
User M
User N
User O
User P
User Q
User R
User S
User T
Now, if I simply visit those first five profiles, their values are re-calculated and the order immediately becomes this instead.
User A
User B
User C
User D
User E
User F
User G
User H
User I
User J
User K
User L
User M
User N
User O
User P
User Q
User R
User S
User T
Iain_S: 33%
Vinther1991: 28%
Scheffel: 27%
TotalVertigo: 27%
siriuslymovies: 18%
Look at how those percentages plummet! The top five becomes the bottom five, and they’re well below the percentage of User T. The problem seems to mostly lie with the calculation of two people’s shared favorites. For instance, visiting siriuslymovies’s profile and getting his value re-calculated meant my shared favorites with him dropped from 163 to ONE. With TotalVertigo, the shared favorites dropped from 152 to 26. Scheffel drops from 267 to 54. Etc. So…something is very buggy with the Favorites tallies.
Also: In recent times, puzzling experiences led me to explore whether the remaining 20 Neighbors truly deserved to rank. So, I briskly visited the profile pages of the top 1000 ICM users. This means the top 1000 people all were added to my Neighbor list, and ranked accordingly. By doing this, I’ve found some other names who SHOULD be among my top Neighbors. So, I’ve jotted down the names of several users who often get overlooked by the Neighbor calculations. Revisiting those profiles last night meant several names were inserted into my top 25. So NOW, my top 25 changes to this:
User A
User B
User C
User D
User E
User F
User G
User H
User I
User J
***User AA
User K
User L
User M
User N
User O
***User BB
***User CC
***User DD
***User EE
***User FF
***User GG
***User HH
***User II
***User JJ
So, what’s the bottom line? This: The site calculations give me 25 Neighbors and, at most, only 15 of them (Users A through O, above) are correct. I say “at most” because there’s the possibility that someone outside the top 1000 also deserves to rank in my top 25.
Don’t others observe similar problems?