1. Quantum Mechanics and Quantum Computation (on edX, from UC Berkeley: https://www.edx.org/course/quantum-mechanics-quantum-computa...), taught by Umesh Vazirani. Intro to quantum computing that made clear key ideas in quantum mechanics, almost in passing. The first of over 70 MOOCs I completed, not available at the moment.
2. Astrophysics (on edX from Australian National University, 4-part series: https://www.edx.org/xseries/astrophysics) taught by Brian Schmidt and Paul Francis. Delightful. Plenty of math but mostly at undergrad level. A grand tour of current topics.
3. First Nights - Handel's Messiah and Baroque Oratorio (on edX from Harvard: https://www.edx.org/course/first-nights-messiah-harvardx-mus...) taught by Thomas Forrest Kelly. Historical perspective and structure of the music. I was hooked from the first lecture. One of a series of 5 outstanding courses in the "First Nights" series, this is my favorite.
I started using SageMathCloud a couple years ago while taking courses from edX and Coursera in physics and data science. I liked it enough that I joined the project as volunteer contributor. At 60+ MOOCs completed, I still use SMC for online learning. Most valuable features to me: 1) not having to install tons of software locally; 2) consolidation of work from different courses & schools; 3) Sage units of measurement and symbolic expressions (good for physics); 4) Jupyter notebooks and sage worksheets, so work from a study session reads like a lab notebook rather than a jumble of scratch paper notes.
The internet is a large, evolving cognitive system. Present information flow, riddled as it is with disinformation, faulty reasoning, and category errors, might be comparable to disorganization in the minds of individual creatures during early development.