Validation for Software Products IS hard

One might question whether startup validation as a whole is BS (a great blog article!). I would say, it's certainly hard. For two reasons:

Reason 1: Measuring the right thing is hard. 
People saying they like an idea is not a validation that people will use a product with that idea. In general, people saying X means not much. However, there are all these statistical sophisticated studies on hot topics such as "How often does the average German buys a new toothbrush per year". Where do these numbers come from? Certainly not from asking 1000 people "How often do you buy a toothbrush per year?". It would sound to them more like "Right here in public, what is a good-sounding number that will make you look like a responsible person?". No instead, what an interviewer asks is "What do you think, how often does the average German buy a new toothbrush?" As the poor asked person has no other knowledge, it can only infer from itself and will in fact tell a relevant number.

Another example. A famous local brewery in Karlsruhe wanted to introduce alcohol-free beer. So they sent out marketing teams to interview people: "Would you buy alcohol-free beer?" Almost nobody would. Somehow, still, the brewery went ahead and introduced the beer. It sold fantastically. How could that be? Again, they asked the wrong question. They should have let people taste the beer, no just imagine it.

Reason 2: Cost vs. Benefit
How expensive is the experiment vs. the cost of simply doing the product? With some software products, it's not easy to build very cheap MVPs.

So what remains is trying to sell a product, before it exists. Or just do the product and take all the risk. But always try hard to avoid the mistake outline as reason 1 and 2.