>Data is often full of errors and the bigger the set the more errors there are likely to be
Strange thing to say, what they say next is better
>no prior hypothesis
If you are trying to measure some distribution more samples gives you a more precise result. You have to be trying to fit the right model though, so you are accurate [0].
>The idea that big data will enable more control of behaviour may be a lot of hype
Overall I think its true. Just because there are some correlations in data, doesn't mean there is actually any causation [1].
Strange thing to say, what they say next is better
>no prior hypothesis
If you are trying to measure some distribution more samples gives you a more precise result. You have to be trying to fit the right model though, so you are accurate [0].
>The idea that big data will enable more control of behaviour may be a lot of hype
Overall I think its true. Just because there are some correlations in data, doesn't mean there is actually any causation [1].
[0] https://en.wikipedia.org/wiki/Accuracy_and_precision [1] http://www.tylervigen.com/spurious-correlations