similar to 'Related nodes' ?

Excellent question! I will research...

I used Related Nodes for a long time on an experimental site and I got excited when I read about this Similar By Terms module! So maybe I can answer your question.

yes and no. Similar By Terms and Related Nodes aim at the same thing, but the means are different.
Related Nodes counts similarity by the text, Similar By Terms, well, the name explains it all.

My conclusion: On a large site calcutation based on full text search might take while, whereas the calculation based on terms can be pretty fast. Another difference might be that similarity is based on all words of the text. So similarity in writing style gets into the calculation: Articles of people using the same "filling words" become similar although the subject is completely different. On a small site where people aren't used to tag, Related Nodes might be a good choice.

There is another one called Similar Entries, that research contents not terms.
http://drupal.org/project/similar

Yes, I saw that one and thought that it looked way too heavy-weight for me. This module is a much more light-weight solution designed not to bog down the server with complex queries.

I played with all. It seems "Similar By Terms" is easier to install and better performance.

Thanks for the module.

I checked related_nodes just because I was curious. (Didn't check 'Similar Entries'.)

It seems to do almost exactly the same as similarterms. There's only a few differences in configuration:

* In related_nodes, you can specify one or multiple vocabularies (to search for terms), inside the block configuration.
In similarterms, you do not have that option in the block configuration, but you can choose to turn on/of several blocks (one for each vocabulary, and one for 'any vocabulary').
Configuration of similarterms seems more intuitive to me, and usually more flexible. (Though you can't do _everything_ with similarterms that you could do with related_nodes, i.e select multiple-but-not-all vocabularies to use for comparison in the one 'related nodes' block.)

* there's an extra config option inside the related_nodes block: you can restrict the node types to search & display.
Maybe that's handy, but not very often I guess, since you have already assigned specific node types to the vocabularies.

----

Further:

* the SQL in related_nodes seems more complicated than it should be, which slows it down.
* ...after all this, I can't get the related_nodes block to actually DISPLAY :( I don't feel like looking further and will just use similarterms.

Oh yeah, what this issue is about:

Sooo... I think you can just add a line to your readme for interested users...
...saying that this module has exactly the same concept as related_nodes, with only some differences in configuration (and minor details in their possibilities). And that similarterms is perceived to work better by most users.

Does this matter any more? Eaton seems to have abandoned Related Nodes and is recommending Similar By Terms instead.

I'm going to mark this as "won't fix". If anyone feels that the differences should still be documented, please post some documentation.