Re: [pydotorg-www] Permission to edit PythonBooks wiki page

[Mats Wichmann]

On 11/12/2017 06:25 AM, dibya wrote:
> Thanks for your response. The bestseller rank indeed means the Amazon
> Bestseller Rank. I understand the reasons you mentioned about why this rank
> might not be the perfect way to rank the books - and that public voting
> might be better. However, I went ahead ahead with this choice for the
> following reasons:

> 2. From my experience, it seems that the Amazon Bestseller Rank, with all
> its faults, is still a very good indicator of quality. How do I know this?
> Well, I visited several online threads about best Python books for
> beginners, intermediate programmers, books on specializations like Data
> Science, Security etc. and made a list of books that people wholeheartedly
> recommend. Then I compared this list with what I get from the Bestseller
> Rank. It's astonishing how well the two lists match up in almost every
> case.  My conclusion from this is the following: you can promote a book all
> you want, but if you want strong sales, quality beats promotion.
> 
> 3. The third point is about books not on Amazon. I do believe this applies
> to only a small minority of Python books. According to my database, Amazon
> currently lists 1005 paperback Python books published since 2009. This is a
> quite complete database and all the popular books are definitely on there.
> I do understand that I might have missed one or two books that are not on
> Amazon, but that's a compromise that I was willing to make (because the
> alternatives were far inferior is terms of cost-benefit).

Not trying to argue, just making personal observations:

I get nearly all of my tech books at the moment electronically directly
from Packt.  At some time in the past I had a company-paid subscription
to Safari and those books arrived electronically as well.  I'm assuming
Amazon was never involved in any way in either set of transactions,
unless there's some kind of API where you can inject a record of a
"sale", though that's a particularly nebulous concept in the case of
subscription services, which I think are a significant factor
particularly in tech book sales - and probably cause "total downloads"
to be not indicative of quality: if an individual book is effectively
free, people grab it and then quickly ignore it if it's bad.  Maybe
Amazon's numbers are still representative on a proportional basis, I
don't know.


Anyway, there's nothing to prevent you from being marked as a wiki
editor, which is now the case.


___
pydotorg-www mailing list
pydotorg-www@python.org
https://mail.python.org/mailman/listinfo/pydotorg-www

Re: [pydotorg-www] Permission to edit PythonBooks wiki page

[dibya]

Thanks for your response. The bestseller rank indeed means the Amazon
Bestseller Rank. I understand the reasons you mentioned about why this rank
might not be the perfect way to rank the books - and that public voting
might be better. However, I went ahead ahead with this choice for the
following reasons:

1. Any site that relies on public voting has to go through an initial
"chicken and egg" problem. The site won't generate value until a critical
mass of voting has taken place. Users won't flock to the site and vote
unless the site delivers value. The only way to solve this problem is for
the creator to have enough clout in the community in order to drive traffic
in the initial phase/ or to do a lot of outreach. Since I don't have any
clout whatsoever and don't have resources for that level of outreach, I
wanted the site to deliver unquestionable value on day 1. Using the Amazon
Bestseller Rank helped me do this - and it worked well too. The site was
popular among reddit and HN users, and I have steady traffic ever since.
But, of course, the long term plan is to involve users as well. Users can
already leave reviews (with stars) for books by using their GitHub account
(here is an example
).
In the long run, the plan is to allow voting and take a weighted score of
votes, stars and the Amazon Bestseller Rank. This might address some of the
issues that you have raised.

2. From my experience, it seems that the Amazon Bestseller Rank, with all
its faults, is still a very good indicator of quality. How do I know this?
Well, I visited several online threads about best Python books for
beginners, intermediate programmers, books on specializations like Data
Science, Security etc. and made a list of books that people wholeheartedly
recommend. Then I compared this list with what I get from the Bestseller
Rank. It's astonishing how well the two lists match up in almost every
case.  My conclusion from this is the following: you can promote a book all
you want, but if you want strong sales, quality beats promotion.

3. The third point is about books not on Amazon. I do believe this applies
to only a small minority of Python books. According to my database, Amazon
currently lists 1005 paperback Python books published since 2009. This is a
quite complete database and all the popular books are definitely on there.
I do understand that I might have missed one or two books that are not on
Amazon, but that's a compromise that I was willing to make (because the
alternatives were far inferior is terms of cost-benefit).

Just to clarify, the ranking used in the website is not 100 % based on the
Amazon Bestseller Rank. For books on specializations, another score is
computed based on what fraction of the book is dedicated to the subject
matter (which I figure out after going through the Amazon preview of the
book). Then a weighted score is assigned. Books that are available for free
get a unconditional score bump (to compensate for reduced sales). Books
with multiple editions that came out in a short span of time also get a
score adjustment (to compensate for fragmentation).

Once again, thanks for your comments and for considering the website for
the "External Sources" section. If you have any more suggestions about how
to improve the website and how to improve its outreach, please let me know.
I will consider them during further upgrades of the site.

Best regards,
Dibya


On Sat, Nov 11, 2017 at 11:09 PM, Mats Wichmann  wrote:

> On 11/11/2017 07:28 AM, dibya wrote:
> > Hi,
> >
> > My name is Dibya and I created and maintain https://pythonbooks.org/. It
> > features a categorized and ranked list of Python books. The site is
> updated
> > every month to include newly published books and reflect new rankings.
> You
> > can see the discussion about the site on reddit
> >  pythonbooks_find_the_best_python_books/>,
> > another discussion on reddit
> >  python_books_categorized_and_ranked/>
> > and on HN .
> >
> > The Wiki page PythonBooks provided inspiration for making this site. I
> > noticed that the lists featured in the wiki are ordered by publication
> > date. I thought users might benefit from popularity based ranking as this
> > helps to quickly find the best books. I implemented this in
> > https://pythonbooks.org/. Also a lot of work was put in to categorize
> > Python books on specializations (ML, GUI etc) and the website currently
> has
> > 23 categories in total (still growing).
> >
> > I have received emails from Python users around the world who have
> > benefited from the site's ranking and categorization. However, while the
> > popularity of the website on reddit and HN has given it a strong start, I
> > believe that there are still a