Poll: What book should I write?

"When Programming Language Examples Fail" (with pictures!)

Caveat: At the time of adding this comment there are 2 points for "12 Programming Languages" and 3 for "Data Science Starter", so they are the preferred options at present.

For the 12 languages if you are looking to create some revenue from this it could be seen as direct competition to "7 languages in 7 weeks" (https://pragprog.com/titles/btlang/seven-languages-in-seven-...). So you may need to expand on what your book would provide that extends / adds to the work done in this previous book (other than comparing 12 languages instead of 7).

To be honest, some of these ideas have a fairly saturated market. If you write another python tricks & tips book or another data science starter examples book, it might sell some copies and make some money. But it probably won't contribute a lot unless you have some extremely good and new ideas and can execute them consistently.

Medium-complexity books are much more interesting but also harder to write.

"in pictures" books are also great; I feel we don't have enough of them. But they are also not very simple, because if you can't use animation, some phenomena are kind of hard to convey. If you can pull that off for graph theory or statistics, awesome!

Perhaps you should take one hard example from each domain and think about how you would write it.

I.e. How would you explain a gibbs sampler in pictures?

Since you seem to have a broad knowledge of programming languages a good book on 'how to choose a programming languge' (i.e. to which requirements fit a language (including ecosystem)). Though 'statistics in pictures' sounds nice too.

When Machine Learning Fails. I'm especially interested when reinforcement learning doesn't work, what do I do then? How do I debug?

If you're going to write "When Machine Learning Fails", then the following spreadsheet might be helpful: https://docs.google.com/spreadsheets/u/1/d/e/2PACX-1vRPiprOa...

To "When Machine Learning Fails" add a chapter or two on "Sabotaging Machine Learning Models".

"When Machine Learning Fails" and commit to using a subtitle generated with a (flawed) piece of code from the book.

I'm the author of four nonfiction books, none about programming and none self-published, but I think some of my experience still holds.

You're doing market research. That's great, and it's a step that a lot of people neglect to do before embarking on a book-length project. A traditional publisher would expect you to do a decent amount of this in your proposal, and the thing they're looking for is, "How is this different enough from everything that's already out there that it fills an unmet need, but similar enough to books that have sold well that we can expect it to sell well too?"

The other thing that a lot of would-be authors underestimate the importance of is their platform. Anybody can come up with a great book idea, far fewer can actually write it well... but the real differentiator is how big of an existing audience you have and how well you can sell it to them.

Finally, you've thrown out a grab bag of possible titles, but which of those are you truly passionate about? If you do commit to writing a book, it's like a marriage. It's going to be misery if you aren't genuinely interested in your partner.

Suggestion: Python + ML

"When Machine Learning Fails" reminds me a lot to Matt Parker's "Humble Pi"[1]to me

[1]:https://www.amazon.com/Humble-Pi-Comedy-Maths-Errors/dp/0241...

[Serious] I prefer video. Not gonna sit down and read. Statistics in pictures sounds like a good idea for an animated video series. Not sure how to pay you for it though.

If you don't already know the answer to which book you want to write then the question you're asking is "which book do you think will sell".

The answer to that question is "who cares". No one wants to read yet another tech book written by someone who is not both qualified in and passionate about the subject.

I am another self-published author. I am of the belief that the best poll you can conduct is to announce that something exists and then try to get people to invest something—time, attention, money—in it.

For example, you can announce you're writing a certain book and ask people to sign up for a mailing list to be informed about updates. Or follow the LeanPub model and offer the unfinished book for sale at a steep discount, with early adopters/patrons getting all updates.

I count amongst my biggest successes a small list of books I DIDN'T write because although people told me that they were interested in the books, mocking up a fake cover and attempting to sign people up for emails told me that the market wasn't actually receptive.

The key is that asking people to make even a small investment gives you a far truer signal than any poll.

Don't write anything if you don't know where your heart lies. The world could do without endless seas of copy pasta making it to the bookstores.

Write a book about how to choose a topic for a book.

My first programming book was one that introduced several different languages at once. I forget what it was called but I remember TCL, Perl, PHP, C, maybe Bash.

It was successful in igniting the spark inside me that lead to more books and more learning.

Often, and especially these days, all people need to get started is an introduction to the environment setup, and where the docs are.

I'm not sure if these are "book shaped" ideas, or perhaps more blog/webpage/Instagram shaped.

If you go for any of the "in Pictures" ones, I'll gladly help. I've been working out on a similar thing.

Whatever you write, see if you can imbue it with a sense of personality. I think readers are more open to this now, and it's a great way to stand out in a world of often dull, derivative books. I wish more of the technical books I read also shared the author's opinions, experiences and gut feelings (even if just in sidebars/breakouts) as well as the facts. The real world experiences of the author could add a lot of depth to a technical book.

“When Machine Learning Fails: in Pictures” could be a bestseller, with the right pictures.

I have a question.

When you say that you are going to write a book titled "When Machine Learning Fails", is your book going to based on the boundary, the limits of machine learning or is it going to talk about obvious limitations and examples? Will it have interesting, insightful case studies, or the regular toxic, baseless anti-ML stuff?

          ° ° ° | °  °   °
     ° °     °  | °   °
      °    °    |   °
        °       |      °
       °   °    |   °  ^
       °        |      |B°
                ^
                | A
     ML Applies      ML Fails

Are we going to get stuff about points in A or trivial stuff like B?

Will this be a lesson for MBAs so they can steer away from use of ML in projects, or the limitations of ML in an ML 101 course?

OR

Will this be on ML theory and practice deep enough to teach ML/DL practitioners new lessons and insights?