Like most other companies developing professional Joomla! business templates, we have added several features to our templates that help your webpages rank better in the search engines. One of these SEO features is called 'source ordering'. In the Joomla community this feature is often refered to as '2-1-3 source ordering' where the 2 refers to the center column in a three-column layout, the 1 refers to the left column, and the 3 to the right column.
During my recent presentation at the Joomladays in the Netherlands, I was asked the question whether source ordering was still relevant. I have to admit I didn't knew the answer on the spot. Some thinking and discussing with others has given me a pretty clear picture on this though, which I will now try to pass on to you.
I don't think code order matters that much anymore. A couple years ago, when Google was crawling only a portion of a large webpage (for example view this topic on Webmasterworld dating from 2003 that discusses a limit of 100Kb of page size), you'd better make sure your valuable text or links were placed in the first part of your webpage, otherwise it wouldn't get crawled at all! With Google crawling over 100Kb of website content (see for example this article published in 2006 that discusses the crawl limits at that time). If I recall some quote from Matt Cutts correctly he stated that they now index webpages over several MBs in size, as long as they contain enough valuable information.
With Google advancing their detection of the 'visual location' where text and links are placed on a webpage (see #5 on this article by Rand Fishkin), source code ordering will most likely have dropped in value as well. Using CSS styling, we can now order our source code pretty much at will anyway, which has changed it from a valid signal to a 'SEO trick' (just like adding a suffix to the URL has).
Photo credits: Ryan Holst
By 'viewing' (and perhaps manually categorizing or using machine learning) webpages Google will notice patterns in webpage source code. For example: 'that div containing a large amount of links, usually placed in an ul-li, often containing links to 'home' and 'contact' will most likely be your menu. Just like 'that div containing more text than any other div, often starting with a H1 or H2 tag, containing the most images and ending with a call-to-action' link will most likely be your page content area. Thus, Google doesn't 'know' whether a certain part of your source code is your menu, your sidebar or your page content, it deducts it by looking at common patterns.
Perhaps HTML5 tags (such as the nav tag are used as one of the signals in determining which piece of source code is what. Seeing how easily one could manipulate these tags, I don't think it'll be a strong signal though. Of course it can be a good guidance for future web developers to identify pieces of source code!
Google is able to read CSS files (for example to determine if a link is hidden), but I don't think it will parse these files and apply them to the webpage to determine the visual layout of it. I think it would require a great amount of processing power (and time) to actually render a webpage, rather than sort out the pieces based on the source code like I described in my answer above.
Until conclusive evidence has proven otherwise, I think it is still best to use source ordered code. For example screen readers will still view the code by source order, so these visitors still profit from it. Even though the evidence above states that it might have largely lost its value, we cannot be sure enough before deciding to step away from this feature.