The WaPo has a longish article on the failure of the FBI's major IT product: The FBI's Upgrade That Wasn't. Oddly enough, I have an informed opinion for once -- I do this type of work.
It is not all that alarming that a new custom solution has hundreds of bug reports. I don't know what universe Mr. Azmi comes from, but it's not custom software development. I assure you that Microsoft Word has hundreds of significant bugs and some hideous design flaws -- but of course it's crap. Even Excel, however. has bugs - and it's been in production for about 15 years.
It is alarming that they intended to do a big-bang deployment of what was beta software. Big-bang deployment has advantages, but mostly with relatively proven solutions. (That said, some large health care systems are doing big-bang deployments of Epic's fairly new acute care software and I haven't heard of big flame-outs yet.)
I'm not as alarmed abut the idea to "write from scratch" rather than modify a commercial system. What would they modify? I presume they were using commercial database systems (Oracle probably) and middleware, etc. "Scratch" doesn't mean what it once meant. I fear they might have tried Java on the client -- but they weren't the only company to follow Sun's path to disaster. That flaw can be remedied.
The biggest problem seems to have been that the FBI is said to be a devastated agency, with most of their senior management gone and a very limited IT staff. It sounds, though the article is skimpy here, that the FBI didn't support the analysis process; didn't have the resources or the leadership needed, and the vendor didn't bail from the project. That would create an IT disaster.
So this story is interesting, but primarily as a "dead canary" warning of poisonous gases in the mine. The FBI is either in intensive care or it's now dead. That's a much bigger problem than a failed IT project ...