I think the "build a product to scratch your own itch" is one of the worst pieces of advice for entrepreneurs, especially first time entrepreneurs. The reason is that most entrepreneurs end up building businesses to scratch an itch, not solve a real serious problem.
From where I'm standing, both of your statements are true. It's important to build for a need (itch), but you should be sure it's something important that customers are willing to pay for.
I think that it's good advice for people with interests outside of tech. If you are a developer, then it's easy to think of things that you could make for other developers, but for various reasons, developers aren't a great commercial market.
There are probably still a lot of non-developers out there using bad software, because developers with a genuine interest in that field and the skills to make great products for them haven't showed up yet. I work for a SaaS company that found a poorly served market, and it's been like pushing on open doors: there were a whole bunch of people with the same problem, but it's not a problem that developers care about, so the most popular solution in the market before us was Excel spreadsheets.
I disagree. We know that most first time entrepreneurs will fail. Not b/c of a terrible idea but because of lack of experience. Someone's first venture should be relatively easy, not necessarily a great idea or even an idea that can raise funding.
You mentioned in another comment a bill splitting app, definetly an itch, nothing unique, and in itself will never become something special. BUT the lessons you learn from building and shipping such a simple product will help you immensely in your next startup.
Why would you even waste time building a bill splitting app if it never had potential in the first place? Yes, you would have learned something, but that in and of itself isn't a justification for making a poor decision. You could have both worked on something with true substantial potential and learned something at the same time, as opposed to work on something with no possibility of success and learn something. In fact, you'll probably learn much more doing the first thing than the second.
I've built a bill splitting app. It was stupid and I learned something, but I wish I had been smarter and built something that [might have] solved a real problem with that time instead.
100% disagree. Building something to scratch an itch is by far the easiest way to proceed initially. Many, many tools and services were built to scratch an itch.
The point is that "build a product to scratch your own itch" is a dangerous statement - you should be finding an itch that (a) a lot of people have and (b) that doesn't have an overabundance of scratching solutions.
The itches of typical developers fail in both of these criteria; typical developers are a small niche of consumers and they can scratch their own itches, so there are many solutions and if it's still itching, then probably simple scratching won't help.
You need to find itches in other domains. It might be your own itch if you have extensive background in some specific business domain and just happen to have development skills, but for most people you'll have to go out of your way to find and understand the itches that you couldn't possibly ever have - possibly because scratching them would be so easy for you that you wouldn't notice the problem if you had it, while they have been suffering for decades.
I can't tell you the number of times I've been pitched a bill splitting app. It's an itch, not a real problem.
For all of the positive examples of companies that were "built to scratch an itch", there are thousands or tens of thousands of failures where the itch turned out to be a tiny market worth nothing, or no market at all.
OK, maybe "itch" isn't the right word. The more painful the itch, the better. I would agree that bill splitting apps are completely stupid. But I've never felt bill splitting was very itchy.
Source: been there, done that.