I don't disagree that you should copy any valuable data off of it, and potentially pull it if the machine it's in needs super high uptime. But after that, should you throw it in the trash immediately, or should you take it to the diagnostic bench and see if the errors are truly fatal or not?
The thing is, there's really no "errors of this type" -- all it means is that something went wrong when reading a sector, but that something could be many, many things, some of which are big problems and others which are not. If that's all the information you have, then treating it as having a 30% chance of failure makes sense. But there's some things you can do to get more information, if you have the right tools and skillset. If you don't have the tools or skillset to diagnose it further, and you can't hire someone who does for less than the cost of replacement, then replacing immediately makes sense.
It's a bit like the check engine light coming on in your car -- it could be a blown head gasket, or it could just be that your O2 sensor's wire came loose. It may well be true to say that 30% of cars with a check engine light on will be dead within a year, but it's also true that the majority of them have very little wrong with them. All I'm saying is it can be worth it to take it to a mechanic for a more thorough diagnosis before you take it to the scrapyard (unless it's so old it's not worth much and you were planning to replace it soon anyway).
