#1

Hi, I’m writing because I’m having some trouble with my API access and communication related to its restriction. I have a developer application that tweets out images from a NASA mission (@PersevereImgBot), and on Monday NASA released a lot of images, so the automated bot was tweeting a lot, to the point it started getting rate limited. At that point I stopped the automated posting, and later resumed with only posting 50 tweets every 30 minutes, to stay under both the semi-hourly and daily quotas.

This was working fine (no API errors) for several hours, and then suddenly I received a 261 error stating “Application cannot perform write actions. Contact Twitter Platform Operations through [link]”

I did submit an appeal with the information above, but where the developer portal noted my access is currently “limited”, it also shows me the message “An email has been sent to [my email address] with details.” However I haven’t received any such email with details (I did also check spam), or any response to my appeal submitted 48 hours ago. The app ID is 20167427.

As this is a timely educational bot, I’m really hoping to get some kind of help getting this looked at since the lack of any emails has me concerned it’s not being triaged (I understand this type of issue generally won’t be resolved via this forum), and I don’t want to submit duplicate appeals that would harm my chances of getting this resolved. I really appreciate whatever assistance you can provide, even if just to verify that my appeal has been received. And if so, is there an associated case number I can use for follow-ups?

Where can I found my submitted ticket?
#2

I should add that I also suspect that it being a relatively new account and suddenly posting a lot seems to have caused it to get flagged as spam because certain users reported seeing “This tweet is unavailable” so I’m also wondering if this is the larger cause of what happened and if it’s possible to get any anti-spam measures cleared. I’ve been running the very similar @InsightImageBot for a pretty long time if that helps bolster the case that I’m truly trying to have a benevolent educational bot here that’s just sharing NASA images and nothing more :slight_smile: Any help you can provide is greatly appreciated!

#3

Yeah that happened to me before too - and i also didn’t get an email, but had to fill in the form Platform “My application has been suspended or restricted from performing write actions” manually

It’s definitely anti spam flagging it.

1 Like
#4

Thanks for the tip @IgorBrigadir, that’s a bit reassuring! I submitted the form on Monday so just need to keep waiting, I guess. Do you recall how long it took for you to get a response back and have write access restored?

#5

oh it took a while - like 2 or 3 weeks or something?

1 Like
#6

That’s really quite a penalty for a new bot doing…exactly what the initial application said it would be doing :stuck_out_tongue_closed_eyes: I get the desire to curb abuse but this really seems rather excessive, especially when the behavior was adjusted before it became a problem. In any case, I really appreciate the info!

It actually looks like the shadowban expired because I got a flood of new followers, I just can’t produce any new content for them until this gets resolved :roll_eyes:

1 Like
#7

I noticed your images and therefore tweets, are extremely similar - for an anti spam system this look suspect. I guess it sort of defeats the purpose of the bot, but maybe you can put in a filter if it tries to tweet something identical twice? I know this is a bit harder to do with images - but even a rudimentary image similarity GitHub - up42/image-similarity-measures: Implementation of eight evaluation metrics to access the similarity between two images. The eight metrics are as follows: RMSE, PSNR, SSIM, ISSM, FSIM, SRE, SAM, and UIQ. with a high threshold in a time window of a day or so will probably do.

1 Like
#8

Thanks for the tip, I was thinking of a similar angle with adding some variation to the text itself. Since NASA had a timestamp bug where many images erroneously had the exact same timestamp, a lot of the tweets ended up with truly identical text. So that may have really increased the odds of the tweets looking like they were spamming the same thing repeatedly. Perhaps by changing up the format a bit I can break that pattern across tweets.

Thanks for digging into potential solutions a little more @IgorBrigadir!

1 Like
#9

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.