It's a very difficult problem to solve for small independent blogs like this... without handing over your data to Disqus or some 3rd party captcha system. Not everyone is comfortable with using those.

Not that it isn't still a problem and trade-off.

I have a small and independent blog, and zero blog comments spam — I think it's related to requiring manual approval of each new persons first few posts. (The comments software is open source, link in my profile)

Is it more difficult in the English-speaking part of the Internet? My captcha for small sites entirely consists of a decoy <input type=submit style="display:none"> for, like, last ten years.

I have a simple "what is two times three" randomly-generated text question and it's very successful in warding off spam. Someone could trivially automate answering it but it doesn't seem like it's worth the trouble to spammers.

Would the "click all fire hydrants" type be easier to program and add accessibility support?

I hate those picture ones. They all seem to take longer than figuring out the skewed text. But I might be in the minority opinion on this one, idk.

sure but it’s unlikely his site is going to be the target of anyone willing to spend the time to even tell the bot where the text to recognise is. it stops bulk; not target... but he’s unlikely to be a target so it serves its purpose

He even dared including some pictures in the article, ridiculous.