The article is fighting strawmen. I've never heard anyone say that starting off with a good idea isn't important.
A good idea is a necessary but not a sufficient condition.
What the author doesn't argue against is that potentially good ideas are fairly abundant and easily obtained. These 'potentially good ideas' (that is, ideas which sound like they would work, but for which there is not yet any good evidence either way) are not worth much not because it isn't important to start from one of them, but instead, because they are not a scarce resource. Give me a 'potentially good idea' along with some quantitative data that suggests there is a huge market, and that the likely cost to acquire customers is far less than the likely lifetime present value of each customer, and suddenly the former 'potentially good idea' is much more valuable - because ideas that come with the data to back them up are much more scarce.
Btw, evidence for good ideas can only be obtained by actually implementing the thing and throw it in the hands of customers. No amount of market research can substitute that (notice the hundreds of failed projects coming out of big corporations that have the resources to do that, but still fail).
If you're having a good idea for which evidence exists means you're not the first to try it out, and this diminishes the value of the idea since you have to differentiate yourself with a better implementation and / or marketing campaign.
People need to make a difference between having a (somewhat) original idea versus copying successful recipes with a few improvements here and there.
For example ... I'm building a service similar to Yelp (simpler, with a few twists); but I'm doing it localized for a non-English speaking country. I do have evidence that it works; but the idea is basically worthless as there are already both international and local competitors; and I'm only doing it because I know I can provide a better implementation to all of the alternatives for the people targeted (only question is the actual time to build the thing, because I may be too late at some point).
"evidence for good ideas can only be obtained by actually implementing the thing and throw it in the hands of customers."
Only if your measure of "good" is what customers accept or want (at the moment you present the idea to them).
This is a common measure for business people, who value profit over all other concerns. But it's certainly not the only way of measuring value.
It's also important to note that what customers accept may have little to do with the idea itself but rather be a matter of sales and marketing.
Take a look at Java, who many on HN loathe with a passion, and yet it's achieved great market share, and many people accept and use it daily. Does that mean that Java is good?
I think a lot of confusion can be avoided by talking about "market success" (as measured by profitability, market penetration, etc). Even people who hate Java can probably agree that Java has been a market success. But that doesn't mean it's better by every measure.
Going back to ideas, it's clear that just because an idea may be "good" by some measure, doesn't mean it will inevitably become a market success. To do that requires many other factors such as implementation, sales, and marketing.
I think this is a case of having a very vocal minority[1]. I do see this a lot on various online sites. And when people profess something loudly and repeatedly, others have a tendency to associate that opinion with a majority opinion.
Actually, if we treat "ideas" as more than just the starting notion, but the myriad small decisions and vision of user behavior, and we look at the ecosystems of
Startups, then "potentially good ideas" look a lot more scarce. The crowdfunding landscape is a great example. The "good idea" isn't about needing crowdfunding for x, it includes all of the decisions about incentives that make them work (or not). When the world is full of startups bouncing over really working through a problem and their vision of a solution - which is happening more and more in my experienc - it's not a straw man to make an argument that dismissing the value of ideas weakens the startup ecosystem
A good idea is a necessary but not a sufficient condition.
What the author doesn't argue against is that potentially good ideas are fairly abundant and easily obtained. These 'potentially good ideas' (that is, ideas which sound like they would work, but for which there is not yet any good evidence either way) are not worth much not because it isn't important to start from one of them, but instead, because they are not a scarce resource. Give me a 'potentially good idea' along with some quantitative data that suggests there is a huge market, and that the likely cost to acquire customers is far less than the likely lifetime present value of each customer, and suddenly the former 'potentially good idea' is much more valuable - because ideas that come with the data to back them up are much more scarce.