Fair point, but let me clarify. I'm talking specifically about data-analysis tools developed in university laboratories which depend almost entirely on (largely federal, though also private) grants to exist and to pay their members. I feel these should be made available free-to-use to other researchers. In addition, these tools are generally heavily mathematical/statistical, so they should be open-sourced so other users are able to check the calculations performed by the tool if they wish to do so.
Note, however, that these tools are also generally "sold" in the scientific currency of publications. More (good) publications usually leads to more grants. Also, the audience for these tools is likely much too small to serve as a reliable source of revenue anyway.
Yes there are companies that offer privately-developed scientific software, and I am fine with those charging for access. Ideally though, they should still be open-sourced so that others can verify the algorithms, but even then, serious validation is probably enough. In addition, sometimes this software comes bundled with specific instruments and it deals with proprietary data that is very hard to interpret if you are not intensely familiar with the workings of the machine. In that case, it probably doesn't matter either as long as the results can be validated.
Anyway, this is getting just a bit off topic and is completely a matter of opinion. There are lots of nuances and even my opinion isn't completely fixed.
