Talk:Drupal/Archives/2011/June

Distributions
Should there be a separate entry for each distribution (ex. OpenPublic) or should they be covered within Drupal? While it is exciting to see more emerge, it also warrants thinking about how to scale the Drupal article to accommodate them. — Preceding unsigned comment added by Elizabeth N2 (talk • contribs) 11:41, 13 March 2011 (UTC)


 * Pressflow is not an install profile! It's a performance hack/fork of Drupal core 5 and 6. It also happens to be 100% compatible with Drupal modules, and I would recommend using it over any standard Drupal 6 core. ( Written by a drupal professional, who is too lazy to edit the article himself. )

Inaccurate Usage Statistics
First of all, apologies if I get something in the etiquette for this wrong, I've never posted to a discussion page before so please just let me know if I've gone about something in the wrong way.

The first paragraph states "It is used as a back-end system for at least 1.5% of all websites worldwide[5][6]". There are two references given. The first [5], is: http://trends.builtwith.com/cms/Drupal

Looking at this, the stats only show usage statistics for the top 10k, 100k and one million websites. The 1.5% stat was obviously taken from one of these, although the quoted sentence doesn't state which. It is obviously not the top one million as that is currently only at 1.31%. The others are both over 1.5%, so obviously the stat is also old.

After the graph on that page, it states "We know of 279,440 sites using Drupal". Looking at the BuiltWith FAQ (http://trends.builtwith.com/faq.aspx), we find the following: "We've indexed about 90 million distinct websites". Putting these two figures together, we find that (279,440 / 90,000,000) x 100 = 0.31. That is, Drupal is used "for at least" 0.3% "of all websites" indexed by BuiltWith and 1.31% of the top one million.

The second reference [6] is: http://w3techs.com/technologies/overview/content_management/all

This one does use a figure of 1.5%, so perhaps that is where the article ultimately got its figure from. However, looking at their FAQ (http://w3techs.com/faq), they only look at the top one million sites "For the surveys, we count the top 1 million websites according to Alexa". Obviously the top million sites is in no way "all websites worldwide", so this statement just becomes really inaccurate.

I would suggest changing the statement to something similar to one of these options:

1. "It is used as a back-end system for more than 270,000 websites[5]"

2. "It is used as a back-end system for approximately 0.3% of websites worldwide[5]"

3. "It is used as a back-end system for approximately 1.5% of the one million most popular websites[6]"

My preference is for #1, as it is demonstrably accurate - that is, that number of websites has been tested and found to be using Drupal. As soon as you get into percentages (#2 and #3), you have to start considering how many websites have been indexed, how many websites exist in total and whether or not it is accurate to use a percentage of indexed websites as an assumed percentage for all websites.

That said, some people tend to find percentages more useful, in which case #2 would be significantly more accurate than the current statement.

I think that #3 is problematic as there is no good reason to believe that readers only care about popular websites. Additionally, the article immediately continues with "ranging from personal blogs to...", which implies that it is used for non-popular websites (most personal blogs aren't all that popular), so a statistic based on popular websites isn't really a particularly useful statistic. — Preceding unsigned comment added by Steve Belac (talk • contribs) 02:57, 13 April 2011 (UTC)


 * I am Sam Soltano, webmster at W3Techs. Obviously I am biased in my opinion which reference is most appropriate, so I cannot participate in any voting. I will try to contribute by providing some thoughts: Any attempts to include "all" websites in a usage statistic is problematic for several reasons. First, there are many millions of websites of little value: link farms, inactive sites, parked domains, etc. One of the reasons why the Netcraft server statistics sometimes fluctuate wildly is because somewhere, someone registers thousands of domains, all with identical nonsense content, which disappear a few months later. If these sites happen to use some CMS, that doesn't tell us much about the popularity of that system. Second, to know some number "of all websites indexed by BuiltWith" is of little value without knowing how a website qualifies to be indexed by BuiltWith. Third, it's even not clear what a "website" is. Are all the *.wordpress.com sites separate websites, or just subdomains of the same site? These are the main reasons why we, at W3Techs, use only percentages, and why we ue a clear and independent criterion how we select an appropriate sample of sites. You made a valid point by stating that the top million sites are not necessarily representative for all the web, but I would argue that this is more representative and certainly more useful than including millions of link farms and parked domins. SamSoltano (talk) 18:21, 14 April 2011 (UTC)

In the Security section, it says: "When compared to three other well-known open source CMS platforms covered by the MITRE CVE database, Drupal ranked second - after Plone but before WordPress and Joomla." What does this mean? Does this mean Drupal is the second most secure by this ranking, or the second least secure? — Preceding unsigned comment added by 64.183.74.243 (talk) 22:34, 29 June 2011 (UTC)