We can predict how a tool might be used by various users. We can also look at how similar tools have been used historically.
If, empirically, certain tools tend to be mis-used in familiar ways, it's just ignorant to say "don't blame the tool". It's a straw-man. When people "blame the tool", it's typically short-hand for arguing that the creators and distributors of the tool share some of the blame for its misuse, along with the abusers.
I.e. there's a long history of tool creators playing dumb / innocent about the predictable and likely abuses of the tools they create. There's an equally long history of these abuses, so any such arguments are to maintain cognitive dissonance, or made out of pure ignorance.
But it's a cost/benefit problem, no? People say this bearing in mind free speech. Sure, free speech makes it ok to say lots of terrible things, but the boon to human rights outweighs by far the evil made possible. People assume that this logic extends to technologies that facilitate free speech. Care to argue that it does not?
Because part of the purpose of these networks is to subvert government control. If you can censor one thing (e.g. child porn), then you can censor anything, which makes it useless if you're an anti-government radical in China, for example.