I find that the feature marking duplicates in red, blue, etc. is rather annoying. First up it doesn’t really work - very often it marks files that are not duplicates as being duplicates. Furthermore, I quite often have multiple duplicate files (mostly aliases) that I would rather not have marked as duplicates in different colors. Any chance you could include a preference to turn off this ‘feature’? It is very disturbing for my workflow.
Thanks very much - great program otherwise! 
Most users – me among them – find it important to identify duplicates (bold blue font) and replicants (red italic font). Failure to be aware of those distinctions could lead to unintended consequences.
For example, if I duplicate a group, I can modify the contents of that group without affecting it’s ‘partner’. But if I replicate a group, the changes are also made in every instance of that group. So it can be critically important in the consequences of a workflow to recognize this distinction.
By the way, a replicant differs in important ways from an alias.
If in the Finder I create an alias of a file, then decide to delete one of them, it makes a difference whether I delete the file or its alias. If I delete the alias, the file remains unharmed. But if I delete the file, the alias becomes an orphan, without a meaningful reference.
If in DEVONthink I replicate a file, then decide to delete one of the instances, it doesn’t make any difference which one I choose to delete.
Related topic: False Duplicates
I appreciate the need for a way to quickly localize and identify duplicates or replicates. I am sure this is a very important feature for many DTP users, but not for all - myself included. Furthermore, since the feature appears to be very ‘flakey’ at this stage, I would still like to be able to turn it on/off - or even better, be able to adjust the preferences of how the duplicates/replicates are displayed (e.g. be able to change the blue and red colors, etc.).
I’ve looked at some examples of files that were said to be incorrectly marked as duplicates, and concluded that DT was behaving properly.
One set consisted of 3 documents based upon the same template. Each of them contained 71 paragraphs, 477 words, 3026 characters. Each differed from the others by 10 or fewer characters. The similarity score is very high.
The other set consisted of 2 documents that are quite different in content and will not be identified as duplicates when captured as unique files into a database.
The distinctions between normal files, duplicates and replicates can be critical to any user under some circumstances. For example, one can modify a duplicate without affecting the other copy. That’s not true if one were to modify a replicant.