Ask HN: What do you want to learn in 2017?

Sounds like exposure vs actually learning something deeply.

However, both activities are very valuable. Without looking at many things, how else can we know what is worth spending time to learn or do well? I currently feel that alternating between the two is good for a while. Perhaps later in life I will know enough to confidently work on one thing for many years :)

I've also found the tree trunk of knowledge model to be very powerful (I believe I heard of it on waitbutwhy.com). The best learning and understanding comes when we build it up in a tree like fashion, where each leaf or branch is supported by a stronger, more fundamental conceptual branch. At the core is the trunk & roots, which are the deep, underlying principles supporting the entire tree of concepts/knowledge/ideas.

Without a strong trunk to build off of, concepts and bits of knowledge float alone, ungrounded, and can wither or rot more easily.

Learn the first 1006 kanji characters used in Japan (AKA kyoiku kanji).

Counterpoint: ReactJS solves a lot of engineers' problems in a clean and elegant way. Even if it goes away, the ideas behind it will continue to be good ones.

There's nothing wrong with Haskell, but I think that a lot of people mythologize it because it's not used by many people and has an air of academia. A lot of people do the opposite to ReactJS because it's a popular framework that a lot of people use. It has no mythos. Of course, mythos is orthogonal to actually having practical, useful ideas.

I want to understand the principles of a signals and signal processing, as they form info theory, and the theoretical underpinnings to learning. Then I want those insights to enrich my deep dive into machine learning, esp deep NNs, in particular to extract ineffable features from images and other complex signals, and finally, how to build a mind.

And I want to learn French.

Start slow, and work your way up. I started with docker early, so I can see how its ecosystem can seem intimidating now, with all the different "tools" and "workflows" to know about. So my recommendation would be to go bottom up. Start with just plain old docker - learn what containers are, what makes them work, then see how to manage containers in a single host with compose. Then move on to clustering with swarm, and then move on to other cluster-mgmnt projects like k8s. Try to make your containers more space-efficient by basing your images on Alpine. Make your host more robust by using CoreOS/Rancher, and so on. You won't need all the steps (or even in the order) I've listed here. But once you've started with a base, you'll have some idea as to where you want to proceed next.

If you need any help, you are free to ping me with any queries at the email in my profile. My authority on the subject: having written a book [1] and a Udemy course [2]

[1]: https://www.packtpub.com/virtualization-and-cloud/orchestrat... [2]: https://www.udemy.com/mastering-docker/