There is a list of the top 100 Irish blogs at Justin Mason’s site.  Now this list is old, but unfortunately I don’t know just how old.  However I reckon it dates back to some time in 2008.  For my little exercise, I thought I would test those sites to see just how many are still active.

The vast majority are indeed still there .  Some however came up completely blank or are restricted access.  I decided to test the sites I could access and see how recently they had been updated.

How do you define a redundant blog?  I decided that if it hadn’t been updated in the last 60 days then it was probably dead.  The result is a little disappointing.

Of the 100 blogs, over half [53%] haven’t been updated.  If I use a tighter figure of fourteen days, then 60% are gone.

Now the list claims to give the top 100 sites by Technorati Rank so it is not a comprehensive list of all blogs.  However one would assume that to gain a listing, the blog would have to be fairly popular, and the author would have to be reasonably dedicated, to the 60% death rate is a little surprising.

Maybe they are right.

Maybe Irish blogging is terminally ill?!

I admit I have never used it with an account of my own, but I do know that it can be difficult commenting on Blogger site.  There is one feature of it that I like however and that is the ability to display links along with the title and time of their last post.  It was a feature I envied and finally I found a solution.

I installed WP-Social-Blogroll on Head Rambles.

I had a philosophy of linking to any site that linked to Rambles and this led to a rather long list.  I therefore set the list up on its own page.  At the time, I could not get the damned thing to work in the way I wanted, as it insisted on displaying the links and the icon, but it left out the latest post, the author and the date.   I left it and promised myself I would address it at another time but never did.

Recently it was pointed out to me that the Links Page wasn’t showing any links at all.  I dived in and for the live of me couldn’t understand why it wasn’t working.  I decided to install the plugin again from scratch.

Originally I just set up a normal page and included the Social Blogroll code.  They did recommend that I install it from a template however.  I tried that [I had never used a template page before but it was simple enough] and the links appeared, but still without title, author or date.  I was back to square one again.

I scoured the Internet but could find nothing relevant.  I don’t know how I found the solution, but I did.  I was using the wrong API key from Google!  I set up a new key, using the precise link to the Links Page and inserted it.  It worked perfectly.

The problem then was that I suddenly discovered just how many sites hadn’t been updated for months or even years.  The only thing is to remove all the links that haven’t been updated in [say] the last six months.  That is going to be a big job.

Maybe I should have left the plugin unfixed?!

Last Monday, at 4 in the afternoon, the latest ‘storm’ started.   I call them storms, as the usual pattern is a sudden peak in traffic before the page slides into obscurity again.

This time was different however.  What emerged over the last week is a quite remarkable pattern of traffic.  The Monday evening storm developed into one of the biggest since 2007, and instead of subsiding, it developed into a whole series of cycles.

Hourly traffic over seven days

For some reason, traffic drops to a minimum at around ten in the morning before rising to the next peak.

Daily traffic over thirty days

The decay in traffic is also quite remarkable in that it follows a mathematically precise curve.  I took the snapshots a couple of days ago, but the traffic is still following the same very precise pattern.

Monthly traffic over a full year

As the second illustration shows, the quantity of traffic is quite significant too.  In fact the site is receiving more traffic in one day than it would normally receive in a whole week.   Even more startling is that by the 9th, January’s traffic had already created a new twelve month record.

The traffic is still pretty heavy, but the rate of decay has slowed right down.  Mathematically, it is indicating that I can expect a permanent increase in traffic of around 500%.  However I recognize that is extremely unlikely which probably indicates why I’m not a climatologist?

I know the general principle – you like something, you Stumble it – but it still baffles me.

A couple of years ago, an article in Head Rambles was Stumbled.   The effect was virtually instantaneous and a little alarming.  My hosting company phoned me to say they were on alert because of traffic to the server, it was that bad (or good, depending on your point of view).

Since then, that article has ridden through several Stumble Storms, as I call them.  None has been as powerful as the original, but they still cause massive traffic.

The graph above clearly shows the initial storm on October ‘08.  The majority of the little stalagmite peaks after that are mini-storms.  They appear to occur at random intervals and random intensity.

There is a storm in progress as I write this.

The graph above shows hourly traffic over a seven day period.  The storm started at around eleven last night and is easing off now, but the effect is very clear.  As storms go, it was a very minor one, but I’m not complaining.

What does baffle me is where these storms originate.  I have searched Stumbleupon and can find no mention of Head Rambles.  Presumably though there is a page somewhere that people are seeing?  Has it risen up the ranks again for a brief moment of glory? 

I have a lot to learn about this Internet lark!

Six of them are on my own servers and three have their own hosting.  After my last post, I thought I would try an experiment – to combine several of those blogs under one roof, so simplify maintenance and to streamline things.  Checking nine blogs on a regular basis for updates and upgrades can be tedious.

I set up WordPress MU on a test site.  For the sake of the test, I used a few spare domains I had lying around.

Installation of WPMU is simplicity itself, requiring only the basic knowledge of how to install a database [I used Cpanel] and the ability to change permissions on the server [again using Cpanel].

Having set up the root blog, I set up a subsidiary blog, and decided to try to mirror this site on it.  Once again, setting up the blog could not have been easier, but then my troubles started – how to I import all the features of this site onto the new one?  I want the two sites to be precise mirrors, so this involves copying the theme, the posts, comments, images, tags, categories and links.

Copying the theme is simple.  I just took a copy of the live theme folder, put it in the WPMU installation and activated it.

Copying the posts was relatively simple too.  All I did was to export the XML file from this site and then import into the other.  This gave me the ability to copy all the images as well which was just what I wanted.

I was now left with a problem.  The XML export/import facility does not include links or link categories.  This required a bit of lateral thinking.

In the end, I decided to use the sledgehammer approach. Using PHPMyAdmin, I exported this site’s entire database excluding “wp_options”, “wp_users” and “wp_usermeta”.  I then opened the downloaded file and made a simple change.

To explain the change I made, it is necessary to understand the construction of the WPMU database.  The root blog will contain its records in, for example “wp_posts”.  Anything starting with wp_ pertains to the root, or the global site.  The individual sub blogs contain a numerical prefix, so you will end up with “wp_2_posts”, “wp_3_posts” and so on.

I ascertained which prefix my sub blog was using and then did a simple find and replace on the entire database dump, replacing “wp_” with “wp_2_” or whatever the prefix is.  I then imported this file into the new WPMU database.

It may seem that I have duplicated things here, as I had already imported the posts and comments, but I did that essentially to just import the images.  I overwrote the information to maintain the integrity of all the ‘meta’ tables.

The result is rather effective – you can see it here, though please don’t leave any comments on it, as they will be dumped!

My next problem [and it’s a big one], is how to map my domain to pont as an alias to the new site.  So far, I can only point a domain to it, whereas I want the domain to act as an alias and to mimic precisely the URLs of all the old pages.

Any ideas?