One of the claims of the Amnesty report is that, based on their crowd-sourced, AI enhanced survey,
... 1.1 million abusive or problematic tweets were sent to the women in the study across the year – or one every 30 seconds on average.
(setting aside any questions about how the data has been collected) - This sounds like this sort of tweet is very common.
The survey's data is stated as the result of combining both UK and USA information. The report does not make clear what this actually means - are we talking about tweets that originate in the UK and USA about UK and USA women? Or are we talking about abusive tweets originating from anywhere in the world about UK and USA women? Who knows?
Assuming the tweets originate in UK and USA, about 30% of adults are active on twitter. UK adult population is 55 million, USA is 260 million, giving a total of 315 million adults. 30% on twitter means that about 100 million people could be posting on twitter.
In any population (except perhaps North Korea) there will always be a small number of 'nutters' who, no matter what you do to stop them, will always be nutters.
How many nutters are there? I think everyone meets at least 1 nutter, so perhaps 1 in 1000? i.e. 0.1%
For our population of 100 million people tweeting 0.1% represents 100,000 nutters posting at any one time.
Using Amnesty's estimate of 1.1 million tweets per year gives and estimate of less than one offensive tweet per nutter per month.
This does not seem that high to me - it's just that there are so many people in the population that causes the headline figure of one tweet every 30 seconds.
If the report is actually based on tweets from the whole world about women in the USA and UK then the source population is closer to 600 million active users. This increases the nutter pool to 600,000 giving less than two offensive tweets per nutter per year
My point being - unless you have a really oppressive police state, there will always be a significant number of crimes even if the rate of offending is very low