It seems the lessons of the first dot.com bubble have already been forgotten. A business needs to have a clear path to profitability or else it's going to be left without a seat when the music finally stops; and the music always stops.
Google didn't have their business model figured out when they started, and they seem to have done ok.
It is now a dot-com boom endnote
But something one of the digg founders wrote on this site also comes to mind: http://news.ycombinator.com/item?id=143269
I would venture to say though that the classic rules can be broken for other considerations because the Internet is a new medium.
I will get to other considerations in a minute but the main thing to understand is that the Internet breaks the old rules by lowering distribution costs.
It makes sense to delay monetization if you have a compelling user proposition that can be scaled quickly through word of mouth(ebay,youtube,flickr) or by partnering with an 18 wheeler (google,paypal,skype).
The assumption here is that if you build a large user base it takes a very small number of paying users to make money.
Its easy to see that in each case there is a real value proposition. In fact none of these applications would exist except on the Internet.
Although some rules have changed, the old rules still apply i.e. building something with a real value proposition and then distributing it without breaking the bank.
Those that failed did not create real value in the marketplace not because of monetization.
where many people stumble is thinking in one dimension terms, like assuming Adsense is a business model (see recent posts by Josh Koppelman and Jeremy Lui) on just how big you need to be for this to be truly viable.
What I liked in the original article is that it offered a few, less obvious options for monetization.
Consider the following possibilities: your user base turns out to be very broad-based and large, your user base turns out to be smaller, but a significant percentage of them is from a particular demographic or vertical, or your product turns out to be most interesting to other businesses or other websites that want to integrate what you built into their own businesses. Each of these scenarios implies a different approach to monetization.
Given the way your understanding of a market can change as you actually meet the real people who use your site, a change in targeted user base can happen to a lot of different businesses. And if you spend too much time up front working on revenue, you may end up throwing all that work away when you confront the actual opportunities that present themselves.
For some businesses, you may have a precise understanding of who your customers are going to be in advance, and then that turns out to be exactly who they are. If that works out for you, cool. But my guess is that that's not the most common scenario, especially for a website.
It is likely common in failed web startups, also.
Google became successful because they increased per-user revenue, not because they increased the number of users.
As another user said, the companies listed so far failed for auxiliary reasons. One of them to be wary of is overexpanding, which is basically what Webvan did. The trick is to not do a massive land-grab unless you absolutely have to; Read this- http://www.joelonsoftware.com/articles/fog0000000056.html
Bandwidth and server costs are still costs. Whether you split those costs among four customers or four million, you still need to recoup them and show a profit to create a sustainable business.
A lot of web companies still need to "pay the rent" or other variable costs. Even without that, don't forget that revenue gives you a huge leg up when searching for funding.
Additionally, a revenue scheme isn't just something that you tack on later. Your options are narrowed further if your technology plan is implemented before you think about sustaining it.
And if you count revenue-less firms that got acquired (and were saved from having to figure it out later), the list is even longer.
Many of the replacements also started as free services and did just fine. Google went for 4 years before it discovered a revenue model. FaceBook and MySpace are doing just fine (profitable, even) as free services. Kazaa managed to turn a profit until it too was shut down by the RIAA.
The only companies I can think of that managed to get traction and still failed due to lack of revenue were those that deliberately priced their product below cost and had no other differentiators, eg. AllAdvantage.com, WebVan, Kozmo, Pets.com, Value America. This is economically very different from the current crop of web services, though: most Web2.0 sites get revenue over their variable costs (even a $1 CPM still leads to 80% margins over the cost of hosting & bandwidth for any non-video site), and so they can make up their fixed costs in volume. The Web1.0 E-commerce sites were pricing their product below variable costs, so they ended up losing money when they tried to make it up in volume.
OTOH, you can produce a popular, well-executed, low cost per user, etc. free service and still fail.
OTOH, you can produce a company with revenues and still fail. (Webvan, Kozmo, Value America, etc.)
Think about it this way: you can have the smartest business model in the world that maximizes the value of your users, but if no one uses your service then you'll make zero money and be totally irrelevant. By contrast, if you have a a service that people love and becomes a part of their daily usage, but you don't have a business model yet, this is a much easier problem to solve.
Attention is most important currency.