Hacker News Reader

Comment by er4hn | Hacker News Reader

It's incredible how Arthur C Clarke foresaw the downsides of full codebase rewrites and putting all organizational weight behind untested and cutting edge technologies all the way back in 1951. Truly prescient sci-fi writing.

View on HN

It could be argued that German military technology in WW2 illustrated some of the same issues, and perhaps gave ACC the idea. They had the V2, TV-guided missiles and operational jet fighters, some of which had massive development / opportunity costs, but none of which materially changed the outcome of the war. The ME262 jet in particular had engines that only survived a few flights because by then German industry couldn't access the specialist metals required.

"Professor-General Norden" might be a dig at the Norden Bomb Sight, an American high-tech wonder weapon that under-performed. https://www.youtube.com/watch?v=U6D5rXbMBKo tells the tale.

Not to mention their tanks, IIUC? Particularly the late-war models. Very advanced, very maintenance-intensive.

Yes - the German armaments industry was very quick to respond to emerging requirements, leading to a proliferation of platforms and versions that were logistically difficult to support, especially as the supply chains were disrupted. Some mass-produced weapons (e.g. the Panzerfaust) were widely distributed and well supported, but the higher-tech ones weren't.

I believe that they also had growing quality issues due to sabotage from slave labour and shortage of materials and alloys. I remember from Wages of Destruction and Albert Speer's memoirs that the loss of certain mines in the East were devastating to them because they could no longer produce certain alloys.

There are organisations that have been following the rewrite strategy for 2 decades that are running 5 or more parallel systems and teams all doing a lot of common things with each with unique properties often in entirely different languages and a lot of integration hell. They will quite happily be starting a 6th such project rewrite this year to solve the problem of the previous 5, this time surely it will work.

Another organisation I know of replaced its "old" .net desktop client with a nice shiny Microsoft derived web solution using some customised COTs and it would solve all their maintenance issues. Sigh its a disaster as expected made all the worse because they fired the developers before the switch was even really done.

Over and over this mistake is made to rewrite and replace instead of refactor on the assumption the old code is crufty rather than battle hardened and that writing new code is easier than editing old. Never do these organisations learn to develop more maintenance friendly systems along the way.

Reply on news.ycombinator.com