Comment by abritinthebay | Hacker News Reader

Comment by abritinthebay | Hacker News Reader

Parent Full thread

> but have we ever verified our dating techniques over 5k years? 10k years?

Yes.

> What’s to say that the hot desert sun and strange weather patterns, radiation or whatever couldn’t skew the dating

Lots. For a start - only one of those would possibly affect the outcome of testing and would also be detectable.

> It seems further unlikely that such a bone from, 85k y/ago as they say, would be so close to the surface

Your incredulity doesn't affect the outcome one way or another and unless you want to dispute the fact that artifacts that are much older are often found on or close to the surface it's not based in any way to the real world.

> In mechanical engineering we don’t ever “trust” the calculations until they are thoroughly validated

And neither does the science around these dating techniques. In fact they're usually better supported than any engineering is.

> It seems backwards to hinge all our science on it when (AFAICT) it hasn’t even been validated itself!

And here's the thing: You're incredibly misinformed and that's why it makes no sense to you.

> And neither does the science around these dating techniques. In fact they're usually better supported than any engineering is.

Given that the results of most if not all engineering can be directly tested, this seems unlikely.

I believe his point is that these methods are contentious for various reasons (religious objection being an example) so they have been heavily scrutinized and where exceptions have been found so have explanations. While most engineering doesn't receive the same level of attention, even if it could.

> In fact they're usually better supported than any engineering is.

I don’t care who supports what in engineering. If you put a dozen [gauges, potentiometers, accelerometers] on a truck and drive it for 1,000,000km varying conditions and the data matches up with the cross derivation for each instruments relation to the other, it is no longer a case of “who supports” -> it is either _fact_ or not and completely indisputable.

But it doesn’t become “supported” data for dropping said truck out of a plane. That’s lunacy - a dozen things not accounted-for in the normal model will occur.

Our models might hold for a few thousand years, maybe even 10k .. but how many times have we dated an >20k year artifact under varying conditions and been able to validate it without relying on another model?

Or as they would say about FEA, crap in, crap out (even though multiple models may agree, none are validated in such a method).

> it is either _fact_ or not and completely indisputable.

Not really. You're all over the place. I don't know why you think engineering is some kind of special branch that is not susceptible to the model breaking down even after repeated testing? Products of engineering fail all the time. Was it "bad engineering" or is "engineering bad"?

All of engineering lives up the abstract tree where you can "take things for granted" about materials, their properties and the repeatability of operation and use of things like transistors, resistors, tensile strength of beams, elasticity properties, etc. "Good engineering" takes into account tolerances and failure points. So then why do products built on these assumptions fail in the field all the time? Because engineering models are built upon assumptions that fail. You then incorporate these failure modes and get better at it. Just like everyone else.

>Our models might hold for a few thousand years, maybe even 10k .. but how many times have we dated an >20k year artifact under varying conditions and been able to validate it without relying on another model?

Because once something is validated and established, people don't waste their time re-testing it unless there is a specific reason to. Can you give a specific reason why it should be retested? What data have you produced that doesn't fit in the model? Or what changes in the model do you propose so that the new model best explains the current data? Or do you want to scrap everything? What is your reasoning?

For someone who has no expertise in the field, its rather odd you seek to cast doubt on something you don't understand, rather than asking a question that will inform your understanding.

Reply on news.ycombinator.com