You can explicitly move old content to an archive and put big banners on it that it doesn't reflect current information. MS does an OK job of making it obvious that you're looking at documentation for an older product, for example.
As I wrote, product documentation is versioned anyway so that's fairly straightforward. But a ton of other stuff is variably still relevant and can get in the way of finding what's most relevant and accurate for today.
Whenever I would go to update something older, it would invariably take me longer than I planned because even things that were not flat-out wrong often didn't reflect current thinking, terminology, usage, etc.
In addition to what's most relevant and accurate for today, past context is also important; knowledge accrues. In computing we are used to fast technological redundancy, surfing the cutting edge of technology, where the greatest commercial relevance is.
But one of my hobbies is vintage or retro computing, and something everybody in this hobby realises sooner or later is how much of an ordeal it is to collate even 20-year-old internet information. And that's in a digital interest area! Now do the same for news or the detail of recent history. Pick an issue and follow it's thread through internet history, year-to-year. It should be easy, shouldn't it? The reality is, half the sources aren't even dated. If we were designing an internet future, the ability to step back through time, in terms of digital information, would be a huge boon for humanity. Because, as critically important as it may be!, today is still only today.
The way search evolved, it never had much of an explicit temporal component--probably because it didn't matter much early on. But now we have a situation where companies legitimately don't want old results polluting newer (usually more relevant to most people) results.
