This is definitely a problem. I'm trying to write a paper now (in applied math), and there's no way I can fit the true details of the implementation into the paper (there's just too much detail in the implementation!). While I can skim over some of the less relevant details in order to get the 'gist' across, it seems ridiculous to me that no-one will ever be reading my code! At least for rigor's sake.
This brings up other interesting questions too, actually: Should I be going out of my way to make my code readily available via a website in spite of publication norms? Are the supposed benefits to the scientific community significant (I'm in a field where the minority of researchers can read the code at all)? What are potential costs and benefits to me, as someone attempting to establish themselves in the field? I can imagine this sort of action improving or deteriorating my credibility as perceived by others.
This has bitten me a few times. CS type journals should totally start requiring source, or at least links to it. I have more than once gotten the source used for a paper, only to find the claims of the authors were nothing like what the source actually did. Worse is that frequently there are leaps of "oh it will work like this" in the papers, which are show to be wrong by the code!
This brings up other interesting questions too, actually: Should I be going out of my way to make my code readily available via a website in spite of publication norms? Are the supposed benefits to the scientific community significant (I'm in a field where the minority of researchers can read the code at all)? What are potential costs and benefits to me, as someone attempting to establish themselves in the field? I can imagine this sort of action improving or deteriorating my credibility as perceived by others.