It really seemed their plan was "James takes almost all the risk and does almost all of the work but James does not get almost all the money."
I dont have enough packt titles to evaluate their general quality, but at least I found their early solr books useful(it was also the only book on solr back then, so not much choice :)
I don't think these books are worth even $5.
Personally, I've found the quality of content and editing in Packt books to be consistently terrible. I won't click on a Packt book in my all you can read Safari Library subscription, I certainly wouldn't pay $5 for any of them.
At the opposite end of the spectrum, I can't recommend No Starch Press  enough.
Still, some of their offerings have been pretty poor (there was at least one Ruby book that was ill-informed) so the best bet is to check reviews for any given book.
Same for O'Reilly. Used to be I would just assume a book from O'Reilly was properly vetted and worth getting, but after buying a few clunkers I learned to do some homework before shelling out or trusting the content.
It was an overall pleasant experience, but I do agree they care more about quantity than quality. Their entire editing process was really sloppy. And review process is crap, at best.
I'm writing my next book through Leanpub.com, that way I will actually make money and can have waaay more control over the final product.
When a book is written for Packt (this is my experience):
1) The author submits his draft chapters to reviewers (there are two or three reviewers per book) - and they provide feedback. The author then has to address all comments from the reviewers, either to correct something (or to make it clearer and understandable) or to reply why it was not corrected. And this goes back and forth until everything is sorted out.
2) After that, technical editors review all chapters and if they feel that something may not be clear enough to the reader or it may have to be double-checked by the author, they will contact him and he'll have to update his drafts again.
3) Then, when the technical editors finish their thing and provide pre-final copies of the chapters, the author has to proofread everything, in order to make sure that everything is as it should be. If the author finds that something is not right he sends feedback to the editors.
You can see that there are many people involved on Packt's side, and (in my case) every one of them cares about quality. At the end of the book the author is even contacted to give an opinion on the quality of the technical editors he was working with.
I know this because I recently authored a book for Packt, named "OAuth 2.0 Identity and Access Management Patterns". It was released few weeks ago.
About the book, I posted an announcement on my blog: http://blog.thisismartin.com/book-oauth-2-dot-0-identity-and...
It is really sad if someone who doesn't know English well accepts an offer from Packt, as I can see in the comments here - there are some cases like that. I hope that they will improve in their selection of authors.
That particular author had a less-than-stellar grasp of English, and the "review" process was much more a "rewrite" process.
I was then asked to be a technical reviewer for something I know nothing about. I declined.
Given those two experiences, I'd never spend money for a Packt product.
Building Machine Learning Systems with Python
Data Visualization: a successful design process
Disclaimer: They gave me a book for free to write an article on once, other than that I am not affiliated.
EDIT: I see many posts about the quality of books published by Packt, the data vis book I recommended is written by an leader in the field of it (Andy Kirk.)
There may be decent stuff in there, but my experience leaves me unwilling to ever try and find them.
I received back an answer with how to proceed, just as if I had accepted to write the book for them, without any specific answer to what I told them.
This didn't seem very encouraging and I didn't try to have additional information.
As the founder and CEO of Packt I want to say thanks to everyone for this feedback. It's really crunchy and specific. We can work on it. We will, and we are.
We'll publish almost over titles in 2013.
We're trying to bring the benefit of structured, organised and reliable content to as many new tools and technologies as we can. We have tended to prioritize quantity, and we have dropped the ball more often than we should. I'm really ashamed about the times we've released content that isn't up to scratch, or we've come across as aggressive or internally focused. I'm accountable for that, and I apologize.
However without being too defensive, I do need to recognize everyone we have worked with on a lot of projects where together we've brought devs a quicker and easier way to share knowledge, quietly, systematically, and reliably. We've also donated over $200k to open source projects as part of our ongoing support.
In terms of specifics, we are working on improving in-house testing and tech editing, the whole tech reviewing process [including who, how and why], how to improve our offer and process with authors to deliver better content and a better experience, how to release updates and corrections faster, improved CRM and discriminating contact. I'll be honest..that's a lot to do while at the same time we keep the titles flowing and the knowledge being shared.
If we're letting you down, or you want us change, let me know. It's my problem. We'll get together and try and work it out.
Happy Christmas to everyone
So far we have:
+Building Machine Learning Systems with Python
+Data Visualization: a successful design process
+Infinispan Data Grid Platform
I read "Building Machine Learning Systems with Python" and it's an interesting and good quality book. I found a small error in the code snippet and added to the book's online errata. Got a reply from Packt and they offered me a free ebook of my choice.
I've read all three of these and found them to be hugely informative.
When Packt get it right, they get it really right. Trouble is, they usually get it really fucking wrong, judging from this and every other thread about Packt on HN.
I have both PostgreSQL 9 Admin Cookbook and PostgreSQL 9.0 High Performance and they are pretty good books.
I got the impression that the quality of any given Packt book would be up to the author, that Packt was OK with releasing sub-par technical works.
Turns out, you get a discount for print books if you own the e-book, I got 50% off after going back to the site once I bought the e-book: €17,50 instead of €34,99
So... €21,88 for both instead of €34,99 (you always get the e-book free if you buy the print version it seems).
1. Apache Wicket Cookbook [DO NOT BUY THE INSTANT TITLE.. THAT IS A WASTE OF MONEY FOR PRINTED SOURCE] [Also if you want a starter book for Wicket go with Wicket in Action/Depth]
2. High performance PostgresSQL 9
3. Google Guava - I absolutely loved this book and how it presented the material. I know that the doc is out there .. but this presented the information in an easier to consume way. [Its worth the $25+ cost of the ebook]
4. Jasperreports for Java Developers - That was the tutorial I needed. I believe it contained a few minor errors but eh.
Now that I see some of the comments here about quantity over quality, I'm feeling better about my decision to turn it down.
That sounds neat. How did you get a part-time dev job?
Earlier this year we reached an agreement where I'd work on RavenDB as a part-time dev job. In particular, I'm porting the RavenDB Management Studio from Silverlight to HTML5, see http://debuggerdotbreak.wordpress.com/2013/09/25/ravendb-stu...
It was a good experience, but they don't provide the resources needed (help and monetarily) for doing good work.
I'd like to do it for career sake. "Hey, I love JS so much, I wrote a book on a popular JS framework."
Just the book i was looking for!
$5 worth the risk?... well can't really go wrong!
All was OK (although the level of English was bad, but I was told not to focus on this) until I got to the REST API section which was atrocious. The author clearly didn't understand REST principals.
I essentially suggested a full rewrite or at least a change of title to something like "simple web api" but I can't comment as to whether that happened yet. I expect not.
For the first two, the communities and versions move so fast that I'd be surprised if a book could be kept up to date. (Storm just hit 0.9; Kafka is now at 0.8. I wouldn't recommend that anyone use older versions at this point.) Following the mailing lists seems to be the way to go in those cases. Anyone disagree?
With regards to D3, it seems like the online examples are prolific; it is hard to imagine a book being needed. Besides, it is more fun to tweak D3 directly.
* Data Visualization with D3.js
* Data Visualization with d3.js
There's no way of reveresing this now but I do feel a bit silly.