This is a problem, because of another very important fact regarding these courses. People taking them probably are because of the flexibility, money-light (if any at all) requirements, and less-than-ideal circumstances. In various ways they are probably disadvantaged and don't control their own time. That flexibility itself may be perceived both an advantage and a great hindrance, as the common academic and chattering-class-member commentator goes about self-paced/directed learning schemes that the large degree of autonomy permits laxity for lack of consequences and the threats imposed by traditional structure, but since a lack of discipline and drive will harm someone in life regardless I think we can ignore them, and instead focus on that people relying on a flexible course are already time-constrained, and lacking direct interaction with knowledgeable instructors for assistance, their only option for hindrances encountered, such as app engine not running, is research; since they probably have time constraints and have already made marginal adjustments to life just to make doing these possible within small windows of time, however, they probably can't spend ten hours at a time researching problems in running Google Apps, why the recommended python downgrade didn't work, PATH variable settings and etc., and remember it all without significantly more investment of time, despite their doing so being ideal.
Maybe my concerns in this regard only apply to a minority of users, but having had a unique set of experiences, these things matter to me and if adjusted for are things that benefit everyone, I think: designing things with the most troubled in mind turns-out, I believe, work with a great amount of benefit to all--greater than just designing with a narrow majoritarian range. My complaint is not research, either: I am an obsessive-compulsive learner, meaning if nothing else is in the way--perhaps including food needs which I may just ignore--I'll spend a 100 hours straight-through researching a topic, related material, and everything relevant to become expert in it, and have been since effects of chemotherapy on my brain during cancer, but...life doesn't permit this, and compared to many other people I have more free time, have been forcing myself to sleep just 4 hours/night again so I can start doing these courses consistently and thus gain even more time, and becoming semi anti-social to make people leave me alone to get even more time and...life still makes it difficult to keep regularly at these things.
One way to alleviate that problem, then, while reducing R&D&I ([re-]Implementation) would be provision of recommendation readings, instructions ultra-clear to solve problems, as well as further background material to understand what is in front of users. It's one thing for them to flail in the dark looking for solution, another to offer that by doing certain reading they can find those answers if they are only willing to apply that knowledge and work-out the solution, which also demonstrates how telling how and actually following how (doing) are completely different. I don't actually think that someone should expect every issue possible to be solved for them, but if a course isn't at least hinting what direction to take, then why take the course? (At least, that's my thought on paid classes, but even in a "free" one like this, that applies because if something required to continue cannot be accomplished for whatever reason, a library updates on a system or the person doesn't know how to define PATH variables or all about unix file structure and permission for program execution...we can again ask, what's the point?)
Something not acknowledged much in modern education yet understood traditionally in classical approached and surfacing in research findings in the works of the brain and memory is that things are not understood of themselves, but by relation to priors understood: think of Bayesian approaches to estimating, but realize it applies also to acquisition. I can sit and read, despite having "forgotten" most things, a paper from biology and understand either immediately or pretty quickly what another would not get in two years of study, because I studied biology and related material. Likewise, someone without a background with mathematics or even simple experience using symbols and mathematical conventions to display a problem...may not get the very simple lay-outs of problems and material shown in some of these course: thus providing extra enrichment material--not like that found in textbooks to assuage politicians and interest groups who brutally ruin educational materials while on federal and state committees, but rather related to or underlying the material at hand, would I believe be a major contributor to successes of users.
Finally, apologies for what may seem as unfocused or too "fuzzy", but I do "fuzzy" when various things seem together to constitute solutions to a need or to specific concerns which are material, and which involve difficulties and complexities. Addressing those things, fuzzy works to creative ends. : )