Hacker News Reader

I got a bad taste in my mouth from Udacity when I tried their React Nanodegree. It was oozing with the expert-in-a-week, fake-it-till-you-make it mentality.

Early on, I asked for help for why my code wasn't working, explaining why I thought it should. My question was handled by the actual face of the course -- the ostensible expert, Richard Kalehoff, and yet he had only a superficial understanding. All his answers were some variant of "I don't know why this isn't working, copy the example more closely." The conversation went roughly:

Me: Why isn't this code working?

Him: Obviously, because you called .setState in multiple places. Didn't you listen to the lectures? It's asynchronous, so you can't guarantee that it happens in the expected order.

Me: Yes, I know, but the docs say that if you chain them as callbacks, you can ensure that they happen in a specific order <link>. Why isn't that working?

Him: I don't know, no one does it that way. I think it's being overwritten.

And I paid $800 for that expertise!

If you've signed up for the course, you can see it here:

https://udacity-react.slack.com/messages/C5T90A6TS/convo/C5T...

As much as I love Udacity lectures (see my pro-udacity rant above or below) some of the things they do don't scale.

They can take hours to perfect an explanation and examples and present it to you. That scales.

They cannot give quality feedback to 1000s of students.

I had a similar experience where they switched the sample data in one of the submissions, so the code would not work.

I don't blame the person for not catching that. Had they told me "this should work, but I don't know why it does not", I would have been fine.

Instead they made thing that they thought might work: Use an alternative notation for matrix multiplication which is different in "subtle ways" (it was identical), change the order of the inputs (the functions were commutative, but more importantly they should have known that), and some other nonsensical advice.

I would have respected them simply saying they don't know but actively bad advice is quite damaging.

It's majestically reasonable to expect experts to understand what they're teaching, doubly so when they are the literal face of the course.

That includes knowing how to use the single parameter on the most common React function.

Experts should of course.

But should the person who understands it best be the face? Or should the person who presents it best be the face? The person who understands it best might design the content, or check that it is accurate. The person who is most gifted at intuitive graphics should design the visuals.

I don't think the fact that the face of course is not an expert is a big deal. I do think they should not be answering the advanced questions. However, with 1000s of questions coming in, I don't know how they should solve that issue. They need a quick way of knowing which is the hard question and which is the easy one. (Is that equivalent to the halting problem? I think it might be).

"Why isn't setState doing what the docs say it should" is not an advanced question, and if he's not prepared to answer basic questions, he shouldn't be "faking it until he makes it" and doing office hours and charging $800 for his expertise. Being able to answer my questions is what I'm paying for.

He didn't simply say "I don't know", he confidently faked it until he made it and claimed that he knew the answer, which was false and based on not knowing how a common parameter worked.

I agree with you.

Or rather, it seems you agree with me.

Or perhaps we agreed with each other all along but I am not sure why it sounds like you're disagreeing.

Sorry, I misread your comments; it seemed like you were saying it was unreasonable to expect the kind of answers I was asking for.

I am sorry but nothing comes close to Udacity. (I don't know own stock, know anyone there, or anything).

Their classes as you say are clear but this doesn't do them justice. They are uniquely clear. There is no other place -- iTunes U, Masters courses, etc -- that reaches their level of clarity. And, as you point out, for the toughest subjects.

If I don't take Udemy, I can generally read the documentation or read through umpteen online tutorials and figure out how to get an app up in Rails/Django/React etc.

If I don't take the Udacity couse on (for instance) AI for trading that information would simply not be available short of a specialized course at one of the top universitys and even then it might not be of the same quality. I studied Machine Learning for my graduate degree and the quality is the same as the best professors there (as it should be! They took the clearest lectures and perfected it before they recorded and edited it).

I don't usually get fired up about online course providers but they have earned it.

How can you confidently say that Udacity is clearer than Master's courses in general? How many universities have you taken Master's courses from?

That's a good question.

So I have only taken Master's courses at two schools, but I have taken many courses at these schools, and a few online.

The professors were from famous institutes and from unfamous ones. It was a top 20 (but not top 10) school. So assuming that it was a wide sample, I think I have some sense of how good the courses are.

You are right in that they might have even clearer teaching at the top 4 (Stanford, MIT, CMU and Berkeley), though some of the teachers I worked with had taught at some of those schools and/or got PhDs there.

I agree with the previous poster; there is simply no place that provides you with bleeding-edge real-world projects that are doable within a month. I have a top 10 CS university M.S. degree so I can compare; Udacity also prepares many courses with top 10 university staff (Stanford, MIT, Gatech etc.)

> Udacity Dislikes: They are placing too large of a bet on the nanodegree concept. There isn't a clear ROI for students - and it's fairly clear that most companies don't really give a hoot about online certifications.

My company is really big and really old, and they do pay directly for nanodegrees through Udacity, and nanodegree completion is very much a factor in promotions because management has extremely poor insight into employee performance and this is one of the few verifiable non-trivial boxes you can tick. I don't know how common this is at other companies, but the strategy has worked for Udacity at least once.

The problem is that some of the Nanodegree's are fairly trivial to get as solution code is available on Github.

For me at least, the Nanodegree wasn't a defining feature for me so much as having very high quality produced videos.

You're setting a very high bar for 'trivial' in the enterprise programming world outside of FANG. Half of the developers in my company would learn a nontrivial amount from having to copy and paste solutions from github.

I have also done many MOOCs and I largely agree with you.

Concretely, I found Udacity was simply too expensive so I preferred Coursera as it had great quality at a good price.

Udemy is full of low-quality trash so I don't bother with it.

EdX has some excellent courses that aren't available elsewhere (for example, Plasma Physics) and also has much more reasonable pricing.

I mean, I live in Europe and for the price of an Udacity course I could attend a full brick and mortar course that is much more appreciated by employers.

Udemy's $10 courses are killing content creators that just stop publishing anything - would you spend months preparing quality content to get $400 overall earnings?

Udacity's ND used to be amazing, their Self-driving Car ND was the single greatest online course I've ever taken in my life. Sadly, they now dumbed it down, made it 2-term (original was 3-term). There were some issues with other NDs where marketing was not translating to real experience, such as AI ND, where 2nd term was so dumbed down it could be finished in a weekend, or the late changes to Robotics ND 2nd term that removed a real robot from the plan and instead offered academic discount on Jetson TX2. Also, Deep Reinforcement Learning project's scope changed massively for worse (instead of robot walking it turned into moving a robotic arm). Despite these warts, where else could you get intro to bleeding edge frameworks like TensorFlow/Keras/PyTorch 2 years ago on complex real-world problems (self-driving car, 3D object recognition etc.)? They simply had no competition there, but that market might be too small. Coursera/edX is way behind the curve with their academic courses.

>Coursera/edX is way behind the curve with their academic courses.

Academic courses aren't usually oriented around "bleeding edge" frameworks etc. That's not a judgement, it's just a difference between how university courses are usually constructed and targeted and how a course/book/tutorial/etc. about some current tech is.

I think the fact that many academic CS courses are based around understanding concepts rather than real world implementation makes it a more sensible choice to go with very stable, well proven technologies.

The truth is though...you need both. You need the technical knowledge and the practical skills.

I did a udacity class a few years ago, and it was good content. But with Coursea and EdX in the same space they seemed pretty comparable to the classes I took at those places.

I like Udemy but the content is definitely not quite on the same level. For me it was mostly about getting over the initial learning curve of getting used to something new (interfaces, set-up). As an intro many of the classes are definitely worth the $10-15.

>>Udemy's was $10 for 19 hours of content + responses to any questions I had within hours, and gave me everything I wanted + more.

To add to your point. $10 comes up to some thing 600 rupees in India. Most students can, and do buy courses when its that cheap.

Accessibility increases reach.

Are there any online educational sites that are accredited that aren't ... actual schools?

My company approved a nanodegree for me but I had to convince like 10 ppl that its legit.