> usually done that by three or so years. What are they doing after that point? This seems like an incredibly optimistic timeline, especially outside of comput

> usually done that by three or so years. What are they doing after that point? This seems like an incredibly optimistic timeline, especially outside of computer science. In some fields, you might finally be able to start collecting data after a year or two of experimental design/setup/optimization, even if things go perfectly. Of course, they often don't: trained animals die, instruments fail, and so on. On top of that, it depends on the "scope" of the PhD. Some of the short PhDs (and postdocs) seem to involve parachuting someone into a functioning research program and having them run a slightly different version of the last study. This certainly gets you /some/ research training, but maybe not enough to actually run a group lab on your own. In fact, I think some of the most useful parts of my (much longer) PhD and postdoc were seeing many things go wrong and then learning when/how/if to fix or salvage them. This necessarily takes time though...