Hi, if you estimate this was an incorrect rejection, you can appeal or resubmit it. You have been able to give your feedback here and receive some from the community on how to present it differently. It does not follow that staff is obligated to revert this decision, there is no need to tag everyone.
1 Like
@Xenopus can you please follow the other ambassador’s advice?
But also, can you be more helpful, instead of being defensive, please?
Yes, I thought about re-submitting but I’m far and I won’t have access to the location in the near future. I also don’t have any more pictures. If I resubmit with the same pictures, even if cropped, the ML - that has proven flaws - would most likely reject it AGAIN. So, let’s strike that out.
Yes, I thought about appealing, but appeals are scare and should only be used, in my opinion, when negligence from Niantic isn’t in question. In this case, it is. This was insta-rejected because the ML had a bad call. Using an appeal with a valid submission just because the system they put in placed messed up, is wrong. They should manually review it, approve or send it to the reviewers. I’m sure you’d agree if you were wearing a different hat.
As for tagging, I think I’m allowed to, right? It’s been 10 days since the last update, no word, not acknowledgement, nothing. Maybe they’ve been busy so I’ve tagged them ONCE.
Thank you.
Our learned ambassador @Xenopus was definitely being helpful. They provided you with the correct information.
We are unable to reverse the decisions of the ML models or make any exceptions to be fair to other explorers. You can choose to appeal this rejection if you do not agree with the decision.
4 Likes
You have done it in the past. Is there a new guideline?
So, just to be clear, if the system/ML you’ve put in place fails, submitters should use an appeal. Correct?
It doesn’t seem a fair usage of a resource that is scarce. Feels like it’s thrown out because of something you launched isn’t working as expected.
ML is an ever evolving process. Like any other system, it has its pros and cons and we need to accept both.
By your words, your ML is not perfect. So, even if it failed miserably, as it did in this case, you will not revert the decision through the support channels.
Submitters facing a wrong ML decision can only resubmit (using other photos, going again to the location, eventually, etc) or appeal (the rarest resource available to them).
Doesn’t seem fair, but I’m just a pawn.
ML may not be perfect but there are multiple upsides of it including fast turn around times. I’d take the win and not bother about occasional misses.
What is the win? I’m not fussed about fast turn around times. I’m sitting on +250 unused upgrades. There are no wins here, quite the opposite.
I beg to differ. I believe ML was a huge QOL improvement. Nevertheless, you’re entitled to your opinion. Cheers🥂
Pre ML here, submissions took months. So I have found it to be good, although not in every case, the overall effect has been positive
Using @Mystogan5097’s words: “you’re entitled to your opinion”.
In this case, it failed. Niantic had the chance to fix it, they won’t. The only solution is to use an appeal so they can have a look at it again. Using an appeal to fix the failures of a live system doesn’t seem fair.
@s3w2 You have had an answer from the team. It is not the answer you were hoping for but it is an answer.
A lot of discussion here has revolved around trying to improve the nomination as there was a general sense that it was not currently well enough presented and improvements would help.
I understand too well the frustration of a rejection when I am not in a position to resubmit. And appeals are a precious resource. Some have slipped down the priority pile.
In the end it is your decision. Appeals can go either way so you need to decide if what you have submitted plus appeal text would be likely to pass. or whether to wait for the opportunity to revisit and make a fresh submission.
Frustrating though it feels my view is that it is time to move on.
2 Likes
Agree. Frustration is usually what brings me here.
1 Like
Previously, you guys specifically requested that we use the forum to highlight cases where machine-learning was in error.
We were invited to do this for the betterment of your system, and for the chance of a reversal by a real human being, which could be a real win-win.
Before today, I have not seen a statement that you do not want this feedback and are happy to leave bad machine decisions as they stand.
Do you no longer value external input?
My point exactly. In the past, we were asked to report issues with the ML. I have done it quite a few times. I was getting insta-rejections all the time, reported them in the previous forum (too bad the archive was deleted). Niantic acknowledged the errors and reverted the decisions. There was no need to push for an appeal.
I was getting ML rejections all the time, such as this one:
ML thought the submission below was a duplicate of the one accepted above… I had a batch of these, all getting flagged as duplicates. No one from Niantic said “yeah, sucks, use an appeal”
Same thing with these below (I can’t pull the screenshot of these in detail because my contributions were wiped out from my account…). These are unique coat of arms also flagged by AI in August 2023 – 3 of them, out of 27.
Another example of contributions being flagged as duplicates by ML, back in 2023. Feedback was appreciated and decisions were reverted. I worked with an Ambassador back then and helped to build a spreadsheet with all these instances. Niantic fixed those.
Seems things have changed now, from @NianticAaron’s words. They publicly acknowledge that it’s not working as intented, that I should be using an appeal and ambassadors have been tremendously helpful, too…
Maybe that’s how Scopely manages things now.
Hi @s3w2
I think the main problem is that we don’t have an established process for “wrong ML decisions”. And I think it could be hard to set up.
Sometimes when user present their case the community feedback is clearly “yes that’s wrong, hope that nia will reverse” and sometimes the feedback is to improve your nomination.
I think about a thread how we would want to have dealed that in future as a community.
What would you suggest? Do you think an appeal section for ML rejections would be the right thing? Don’t know if they have enough employees for that.
1 Like
Could you point me to where this was said?
1 Like
In the past, we were invited to share these issues in the forum or work with the ambassadors. I’ve done both. It’s the first time I’m getting the door shut - resubmit (not possible) or appeal.
It’s highly likely that my submission would have been approved if it were reviewed by humans.
If resubmitting was possible (which isn’t), there is no guarantee that ML wouldn’t pick it up again and insta-reject it. From my past experience, even with different crops and text, didn’t help, I was still getting insta-rejections - overturned by Niantic employees after reporting in the forum/ambassador.
What would you suggest?
Not shutting down the door. Niantic seen this particular case, as many others in the past and have done something about it. Suggest using appeals, the rarest resource in the whole Wayfarer system to solve a failure from their own system is an incredibly unfair suggestion.
Do I need to mention how appeals worked out for me recently? (funny enough, decision reverted after bringing them to the forum! So unfair…)
I think we are talking about things discussed on the old Vanilla forum, so I don’t think I’m going to find a quote, even if my memory provides a rare flash of verbatim accuracy.
@NianticTintino acknowledged that the machine-learning could miss badly, asked for our help flagging things ‘for the team to revisit’ and indicated that this feedback was valued.
Subsequently, various submissions (including some by explorers who went on to become ambassadors here) received recognition from staff, and had their status changed to Accepted.
More recently, explorers who posted ML errors reported the status changes, themselves, without any word from staff.
Are you claiming this was mass halucination? Or simply that it was not a contractually binding obligation?
1 Like
Can you point me to an ML decision that was overturned on this forum? I only know of some clearly incorrect appeal decisions made by humans being overturned.
I do remember one instance early on where they did reevaluate ML rejections after tweaking it. I was affected, but had already resubmitted mine by then. If I can find a reference to that, I will edit it in here.
[found it! ML Review Hiccup and Automatic Appeals for Rejected Nominations ]