I’ve seen another live blog on for the Udacity courses on Journal of a Quant by Ilya.
Later this week I’ll be talking about Udacity’s CS101 and CS 373 as well as Coursera’s Model Thinking and Software Engineering for Software as a Servic which have also just gotten started as well. However for this post, I’ve decided to just focus on the course platforms themselves, Coursera and Udacity.
Coursera got it’s start in the fall with ml-class and db-class as Stanford’s Online Learning Initiative. Then, in January it got rebranded as Daphne Koller and Andrew Ng are launching it as its own company. It’s unclear what if any affiliation coursera will maintain with Stanford.
The coursera platform was already pretty solid in the Fall. I never experienced any of the network issues with ml-class that ai-class had every week. Coursera also has some very nice features. I particularly like the ability to play the videos at 1.2X or 1.5X normal speed. The quizzes embeded in the video are very helpful. I’m pretty familiar with most of the material that is being covered, so I will often listen to the lectures in background at 1.2X or 1.5X while completing some other task–such as writing this blog post. When the lecture is stopped with a quiz, I am able to gauge my understanding. If one of the questions takes me by surprise, I know to re-watch the last few minutes. I assume that the Coursera professors are also looking at the quiz/speed/re-watching data to better understand how the students are understanding and digesting the material.
Since the fall classes, coursera has definitely improved. The interface is a bit slicker. The design is a little more professional. There is now an indicator next to the lecture if you have attempted the in lecture quiz. There is also a little hint “Press ? for keyboard shortcuts”. Doing so gives you a list convenient, if largely unecessary shortcuts:Keyboard Controls
- ?: Show / hide shortcuts
- P: Play/Pause
- Up: Increase volume
- Down: Decrease volume
- Left: Rewind
- Right: Fast forward
- F: Toggle fullscreen
- C: Toggle captions
- Esc: Close video
I don’t think that captions had been available in the previous version of Coursera. Howerver, they don’t seem to work that well from my limited look at them. Though that doesn’t bother me as I don’t use captions anyway.
However, the most appreciable improvement? After answering a question you are given a option to see an explanation. I think that is pretty valuable. If I really understand a topic, maybe I don’t want to see an explanation. However even if I got a question correct, I might want to see an explanation on why that answer was correct.
Coursera has definitely improved. My complaints with the new system? There are three small complaints which I think are actually regressions since the first version.
- The lengths of the video are posted next to the name in the list. I distinctly remember ml-class having all of the video lengths posted. I also remember them being mostly ~5min. The new courses have runtimes mostly >10min. I preferred the 5min videos.
- Settings aren’t remembered from one video to another. I start watching a video. I adjust my volume down to ~25% and set my speed to 1.5X. When I finish that video and click Next, I expect that the next video will keep those settings. However, every time the speed resets to 1X , and more annoyingly, the volume jumps back up to 100%. I am split on whether the settings should be kept from one session to another. However, I definitely feel that if I have just adjusted my volume down, the system should definitely remember that for the next video.
- Progress isn’t tracked properly. The original Coursera showed a green checkmark next to any lecture I had already watched. There were some other features around course progress which look to have all disappeared. I think that this is a relatively recent feature loss because I thought that the preview videos for Model Thinking actually had the progress checkmarks. Correction: I refreshed the page and the checkmarks appeared next to the videos that I had seen. The bug isn’t that progress isn’t tracked, it’s that progress isn’t updated until you refresh the page. Not a very big problem at all. However, now that I’ve seen how the progress system works there are some enhancements that I would love.
- A way to mark videos as watched without actually opening it up to watch. There are other ways to watch the video than through the Coursera player, but only watching through the player will update the progress.
- A way to unmark a video as watched. There were times that I wanted to re-watch a video from ml-class to review the information or because I hadn’t been paying 100% attention the first time. A way to remove progress would have been great.
- Remember where in the video I stopped watching. Currently the system marks a video as watched after only watching a portion of it. Now the videos are generally short, so I should be able to watch them in a single sitting. However, there are definitely times that I have started a video and closed the browser, then reopened, forgetting that I hadn’t actually finished the video. To be fair, this is pretty nitpicky. The only site I have ever seen implement a feature like this with any sort of consistency has been Netflix.
Udacity is a good improvement over Know Labs’ freshman effort with Artificial Intelligence. The design has also been spruced up to my liking. The best part was the addition of actual programming exercises. Here come my list of things that I would love to see improved:
- The only way to get to the forum is first by clicking the Announcements and then clicking on the forum link. I’m looking forward to the day that the Discussion link is activated.
- My Udacity login doesn’t work in the forum. I would have hoped they would have had an integrated login.
- The first video starts playing every time you enter Course Content. It was nice the first time, but now it’s just getting in the way.
- Homework answers aren’t saved properly. My programming answers for the homework are saved nicely, however when I go back to review my answers to the other questions, I find the text inputs are empty.
- The questions don’t automatically come up when the video ends. It wasn’t wholly intuitive that I should click the NEXT button to get to the question. This is actually a regression, as it worked for ai-class.
The list of improvements above are really just nitpicks. Also, I am sure many improvements are coming soon. They have already fixed some issues such as an indication of which video is currently in progress which Ilya from Journal of a Quant also mentioned.