It's because they're procured based on verifiable requirements, and yet there's no way to specify a verifiable requirement for user interaction.

Security is almost the same. 'Must be secure!'. Unless they procure and schedule independent penetration testing and code audit (if they even get the code), the vendor is able to deliver insecure code with a horrible UI and still be 100% within the terms of the contract.

The reason government doesn't do more agile is because the politics of the stakeholders can get crazy, with the end result being constantly shifting goal posts and nothing delivered. So they require Commercial Off The Shelf (COTS - give us something that already works), and yet due to the requirements... there's always significant customization required. I wish someone in government understood this : P

The short answer is that it's not just a website. The website is just the front of a large, complex project that involves many other things.

I have noticed how shocked people are when they hear how much custom software really costs. I don't know what you do for a living, but here's an example:

A family member hears you make software, ask you if you can create a website for his business for a friendly fee. Sure, you think. After reviewing his requirements, and deducting 50% because he's your uncle and you like him and want his business to do well, you give him a quote: 3 months and $10,000. He nearly gets a heart attack. What had he expected then? Well, a copy of windows costs less than $100. How could it be more than that?

One of the few things I agree with our current government in the UK on is their move to bring government IT in house from external companies. That decision has led to Government Digital Services, who are building out http://gov.uk/, and a few other digital departments within government.

A friend of mine works at the Ministry of Justice, and told me recently he has the easiest job in the world, because even if he were to just sit at his desk and do nothing he'd be delivering a better service than the agencies that proceeded him, and saving the ministry huge amounts of money.

Part of this is confirmation bias. There's no shortage of private web sites which were costly, look like shit, work like shit and are easy to hack. We just don't take any notice. We see a strikingly bad government site, it makes us angry and we compare it to top 10% or top 1% websites in our heads.

No idea what it costs, but GDS[1] is doing a pretty good job on http://gov.uk. My guess is that only the savings on closing hundreds of sites [2] justifies it. Maybe it also prevents some of the uk citizens from not going nuts while trying to do gov business online ;)

[1] http://digital.cabinetoffice.gov.uk/ [2] https://www.gov.uk/government/news/hundreds-of-government-we...

In Portugal, I know of about 5 systems that were developed by either companies of family members of politicians or by companies were one of the shareholders/owners were the decider on which company to use. The INEM (emergency response unit) hardware and software was sold by to INEM by the company the person deciding on which company to use created and then left INEM. Sorta went like

Public Contest -> Manager creates company and submits project -> Manager chooses own company -> Manager quits INEM to 'work' at the new company. This is public knowledge but there isn't much we can do about it

One reason is ridiculous schedules. Last year, I worked on a government contract where a law was signed on September 1st that required a system be in place to make payments on January 1st. So four months for vendor procurement, development, testing, and roll-out. The development time ended up being two weeks. That's not going to be quality software. If it even works, everyone is happy enough.

No competition.