According to whom?But they do work well don't they?
Tables were created as a result of research, sometimes years and years of research. The goal was to find the most reliable algorithm, something as close to the middle of the range of possibilities between absurdly conservative and probably going to get DCS. If you follow the results of that research carefully, the odds of your getting bent are extremely slim. The more you stray from them, the closer you get to the edge of safety, the greater your chance of getting bent.
My own personal research into NDL diving practices indicates that divers who begin their ascent prior to NDLs have a wide range of possible safe ascent strategies. For example, the PADI RDP was created using the 60 FPM ascent rate then in common use. I was told by someone at PADI headquarters that their research did indeed show that ascending at that rate from these limits was safe, but it was OK to ascend at slower rates, too. That is why the language they use says to ascend at a rate no faster than 60 FPM--meaning slower is OK. If you look at DAN language on ascents, you see that same vagueness--they tell you not to ascend too quickly, but they don't tell you what would be too slow.
People who use the methods some advocate for planning multi-level dives with tables are not using valid processes, but there is a good chance they are OK doing it because they have not moved close enough to the edge of safety. They might be darn close to that edge, but they have no way of knowing how close because the tables cannot be used to calculate that. In contrast, a computer can be used for that, because it does make those calculations. If your ascent is too slow, it will tell you. The tables will not tell you that, even if you think they do. You're really just guessing.
We will never have good data on those guess, because the percentage of divers using tables is tiny to begin with, and the percentage using them incorrectly to plan multi-level dives is undoubtedly only a tiny percentage of that tiny percentage.