6/19
Report
Test automation process.
Finished improving my automation code.
The process is quite simple.
1. Do 3 random actions. (No like / follow / tweet.)
2. Follow 1 account.
3. If total followed today is not 10(set amount), go back to 1.
It repeats 1,2,3 for follow, likes, and PM until it hits the set amount.
PM Block.
following, liking is not a problem in twitter.
But it seems that Twitter has a PM block.
This is the log of a single account.
If you see in this log, the first pm was a success, and the second pm failed.
This is the log of multiple accounts.
Some accounts succeed once, some succeed up to 5 times.
So, I am suspecting 3 things for the PM limitations.
1. Is twitter detecting automated clicks instead of pyautogui when sending PM?
2. Is there a problem on account creation when I ditched creating by phone?
3. The accounts are not warned up or aged enough?
1. automated clicks.
I was testing only selenium clicks at the beginning.
So selenium clicks were succesfull up to 1~5 times, as you can see in the logs.
So I have tested manually if the PMs fail.
You can get blocked on the 1th try manually.
You can get blocked on the 6th try manually.
Sometimes you pm to specific account, and is not PM blocked.
I think this is not relevant, the account auto PMs you when going into the PM screen itself.
So this seems to be a business account, which is irrelevant.
Selenium clicks fails randomly after 0~5 times.
Manual clicks fails randomly after 0~5 times.
The results are same, so I think this is not the cause.
2. Is there a problem when I just ditched creating accounts by phone?
There is this ban screen that I get after I created accounts with my browser.
If you solve a single captcha, you can simply get away with it.
I did not get this captcha if I register by phone.
But I changed to creating with browsers. (which is fast, but gets this captcha)
The accounts tested today both had phone created accounts and browser created accounts.
The time I switched from phone creation -> browser creation is '2023-06-08 00:00:00'
The account tested today had both the accounts.
Did not have a significant difference after the '2023-06-08' mark.
So I suspect this is not the problem either.
3. Warming up, age issue.
I think I need to change the test factors a bit to check if this is the cause.
I was thinking of doing PM tests separately every week.
After 7 days of randomly browsing, liking, and following,
I will PM 6 accounts for every account. (no account passed 6 today)
May, or may not be the problem. I do not know.
Proxy traffic limits
Everything else seems to work fine except PM.
But I still had some additional problems, that I need to think about.
It seems that I don't have enough proxy to test all the accounts.
The time spent per account was about 10 minutes. (10 actions, test set 1)
I cannot handle this much traffics running tests this slow.
I will need to improve the speed of the actions or add another proxy.
I have 2 proxy sims (excluding my own phone)
Consumed proxy min per day
100 acc X 10 action (10 min) = 1000 min
100 acc X 20 action (20 min) = 2000 min
100 acc X 30 action (30 min) = 3000 min
total = 6000 min
Proxy handle limit
day (1440 min) * 2 proxies = 2880 min
Proxy handle limits are too low, nowhere near close.
Solutions
1. I can add 2 more proxies,
2. Or I can extend the test actions to run per 2 days.
I am more leaning towards
option 2.
Proxies in my country is expensive.

So each account will follow/like, and wait 2 days until the next action.
You will have to tolerate me about this one. Sorry about that.
What to do next
There was quite a lot of changes I had to make. Quite a bumpy ride today.
But I had fun, and I think that matters a lot.
Will begin tests tomorrow.
I was thinking of giving out log statistic charts.
That would be cool.
Adios amigos.