But I think it might be a lot easier to manage if step names were based on the titles of the history items instead of "data 2" or whatever.

Has this been tried and rejected for some reason?

It's been tried and rejected because dataset names get very long and unwieldy. E.g. "Sam/Bam Alignment Summary Metrics on Tophat on Filter FASTQ on my_rna_seq_reads"

Would a pull request implementing this change be welcomed?

What we imagine would help is a way to easily show/find a dataset's analysis path -- its parents and its decendants -- so that it's possible to trace the datasets/tools used to create a dataset and the tools/datasets subsequently used.

This is something we'd like to do but haven't put much effort into yet. Community contributions in this space would be great.

Best,