In days of yore, some badly set up mail servers wouldn’t necessarily look for an MX record first, but would go straight to the A record if it was present. If you point an A record for the domain at your web server, you get rogue mailservers banging on the door trying/failing to deliver mail. This may be irritating, or problematic, depending on the situation — and if you’re the tax office, probably the latter.

I suspect these badly configured mail servers have all been dead for 10 or more years, but it wouldn’t surprise me if there were still people out there setting up DNS zones who still consider this an issue, and leave out the A record for the bare domain to protect against lost mail.

Knowledge – as @nevali suggested, finding somebody who knows what the change actually means is problematic. Its not just a public sector issue – I’ve come across a huge number of commercial/personal websites that have identical issues. Generally people in charge of managing domains don’t understand the fundamentals of DNS – so they don’t realise that you need an A record for both the domain root and for the www (not totally accurate I know, but for the purposes of this comment it’ll do).

Change Control – otherwise known as ridiculous bureaucracy to ensure that any changes will not break anything. A DNS change like this is probably seen as too risky (due to lack of understanding).

Priority – as you mentioned, it is probably sat in someone’s queue to be dealt with. People are still getting to the site, so I imagine it isn’t seen as important.

Suppliers – some DNS services make it easy to change these things, others are much more difficult (I recently had to deal with one by sending a fax!). Even after you take that into account, chances are the supplier suffers from all of the above problems as well. Recently I requested an old supplier alter the DNS records for a legacy domain name – they only changed it for the www, not the root. This has happened on more than one occasion.

From a technical point of view (and my own personal opinion) there is no reasonable reason for http://hmrc.gov.uk to not work. Unfortunately, it doesn’t work out like that in reality.

http://coi.gov.uk/blogs/digigov/2010/01/avoiding-sub-domains-in-web-addresses/

Glad someone out there is really thinking about these things.

“www.” succeeds where http://ftp., gopher., and ns. all fail. Geeks see the Internet as a hierarchy – services like the web are attached to domain names which are attached to organisations, companies and individuals, but we fill the gap in between “www.” and our browser.

Non-geeks are picking up the pieces of what DNS is – they don’t see hierarchies, they see 1-to-1 matches between “label X” and a virtual object. Someone’s e-mail is one string and one string only – “person@example.com” – so why should “www.example.com” be the same webpage as “example.com”? Doesn’t that lead to confusion and questions about whether you need to add the “www.” or not? The branding is a legacy.

OK, that’s a lot of focus on 4 characters, but totally agree with you that someone needs to fill the space for usability of all technology. We need to realise technology is not a brand, but a tool. We need arguments for why this kind of user-unfriendliness and unjoined-up thinking is bad for productivity and risk management. Not because they’re annoying, but because it’s a symbol of how much we understand the technology we’re pimping and relying on so much.

Some of our clients (and curiously, they are some of the smallest ones, not the the major corporations whose names I won’t disclose but I can guarantee you will have heard of them) insist on running their own DNS. Despite the fact that pointing a domain at our DNS servers doesn’t prevent them from pointing them somewhere else [without our involvement] at a later date, they still feel that running their own DNS is some sort of safety-net.

Except they screw it up. Every single time. They don’t actually know what any of it means, so even if we find the person responsible for it to as them to fix it, they don’t actually see that there’s a problem in the first place, nor why they should alter something which has always worked fine in the past (as far as they know), and anyway, they don’t know what it is you’re asking them to do and what negative effects it might have—somebody needs to perform a proper analysis and that’s costly and time-consuming.

And those are the small clients. The big ones usually ask us to buy a domain for them, look after it from end to end and send them the invoice each year. The smart big sprawling organisations delegate. The likes of Capita, who take on these projects, are experts on nothing beyond consultancy and project management: it’s like going to a supermarket instead of a master pastrychef. Sure, they have resources and a specific skillset, but for almost everything, it’s incomplete. The problem is that, because the Government doesn’t have the skills in-house, it outsources to people like this, because these are the only people large enough to be deemed “worthy” of building a website for a government department.

The devil might advocate a line of “they’re too busy sorting out big things to deal with these small things”

But I tend to think that the inability to sort out these small things points to a much greater inability to sort out big things. Because big improvements come when you make lots of mall improvements with no great fuss.

Underlying problems are:
- inflexibility of outsourced countracts
- still general inability to see things from customer’s point of view at any stage