Don't pull your hair out, @Harry! 
We are here to help. 
No, these are completely different methods with completely different clustering/denoising algorithms — you would need to see the original papers for more details, but I would expect different outputs from the different denoising/clustering methods. For one, uparse (as far as I know) is an OTU clustering method (with more stringent built-in QC), whereas dada2 is a denoising method (filters low-quality reads, attempts to correct errors in other reads by modeling the error rate of your sequencing run, dereplicates sequences, filters chimeras and singletons).
Are the other papers using uparse or other OTU clustering methods? In practice these methods usually result in much higher OTU counts than denoising methods like dada2 or deblur. This has been discussed a lot on this forum, e.g., here.
So your observations are pretty consistent with what many others have seen when comparing OTU clustering to denoising results.
I hope that helps!