I remember sitting in the offices of The Observer on a gloomy Monday some time in 2003 thinking about the internet and how wonderful it was for newspapers.
The Guardian website had just logged its first 100-million-impression month and there we were, scrapping for a few thousand readers with The Sunday Times. It all seemed a bit pointless.
The Guardian Unlimited online audience was measured in the millions rather than the thousands and it just kept on going up and up, effortlessly it seemed.
We would routinely spend £250,000 on a TV ad and be delirious if it gave us an extra 50,000 sales. Meanwhile over the road, the web people could rack up 50,000 page impressions before their coffee machine had even warmed up. We were obviously doomed.
It wasn't really until last week, when I sat down with a calculator and the figures for the newspaper and website I trust the most, The Guardian, that I realised what a fool I'd been.
In fact, like many journalists, I've been totally conned by the myth of the print-conquering newspaper website.
As it turns out, even four years later, every single British national newspaper website still has a lower audience in the UK than the newspaper it is supposedly killing.
The best newspaper website has 30 per cent of its print audience. And with internet broadband penetration in the UK fairly high, there isn't any reason to believe that this web audience will explode any time soon.
Those who want the full maths can go to my blog, The Inksniffer, and look under the hood. But this is basically what I did. First, I chose a newspaper. I plumped for The Guardian, which also has a strong website product built on top of a weaker newspaper.
So if its website isn't outperforming the paper, I dread to think what that says about The Times and The Telegraph.
Then I tried to level the playing field. No foreigns or bulks for the print title, UK unique users only for the website. The print number is easy to derive from the ABC: 293,940 a day.
The online numbers are harder. Two metric companies say GU had 2.1 million unique users in April in the UK (ComScore, Nielsen/ Netratings); another calculates it at 5.3m (ABCe). GU, not surprisingly, uses the latter on its AdInfo website.
And that, of course, is problem number one. No one actually agrees on how many people are reading newspapers on the web.
Worse, the measuring companies regularly change their metrics quite dramatically when methodology flaws are exposed and advertisers don't bat an eyelid.
Users can be counted by server-based metrics twice if they read at home and at work, for example.
For print circulation people, used to arguing the toss with the ABC over a deal for a couple of thousand bulks, the double standards must be particularly irksome.
Back to the maths. To avoid claims of rigging, I used ABCe's larger figures. But instead of the monthly figure above, I started with the lesser-used daily unique users.
I did this because my old colleague Simon Waldman, now digital supremo of GMG, says on his blog that this is the figure that most easily compares to a newspaper ABC.
By this measure, there were an average of 795,812 unique GU users per day. I reduced that to the 34 per cent who are UK users and you get 270,576 UK unique GU users each day – a good 20,000 fewer than the purchasers of the newspaper.
It gets worse. We want to compare users to readers, not users to purchasers. For this I went to the National Readership Survey, which suggested that each copy of The Guardian is read by 3.36 people.
That would make 986,586 UK readers in print and 270,576 UK readers online.
Many people challenge the idea of a newspaper being read by three people, but having used the more generous ABCe numbers for web metrics, I wasn't going to be unreasonably cautious on print readership.
So, even after 10 years of spectacular growth, with broadband penetration now high and newspapers said to be in danger of total irrelevance, The Guardian is still read by vastly more people in the UK than Guardian Unlimited.
And that's true for the best newspaper website in the world, whose online operation substantially outperforms its print product.
Why is this important? Because the people running newspapers believe that the future is online and it has driven their thinking for the past five years at least.
This is bad news for newspaper people, because not only are the reader numbers not very exciting online, neither are the prospects for selling advertising, which has indeed grown dramatically and is about level with print advertising in the UK.
But 50 per cent of internet advertising is search, which newspaper websites won't attract in any quantity.
And for the other 50 per cent, newspapers are competing online with, well, everyone in the world, from specialised bloggers and ultra-low-cost suppliers such as Craigslist, to highly-funded behemoths such as YouTube, Yahoo, MySpace and Facebook.
Not to mention everything in between, including every other old media content owner. The effort expended on making newspaper websites passably useful online would be far better spent on making newspapers more competitive and useful in print.
More flexible distribution would be a start, so we could offer a different product to different people and get it to them when they want to read it. Better products help too.The Observer sells more copies now than it did in 1995 thanks to the past seven years of product innovation.
Sadly though, conned by the numbers from their web departments and aided and abetted by laughably inconsistent web metrics, it is more likely that newspaper owners will strip newspapers of the resources they need to reinvent themselves in order to nurture an internet beast that they believe is a rottweiler puppy but is, in fact, a fully grown poodle. They are barking mad.
John Duncan was managing editor of The Observer from 1999 to 2005. He now works as an international newspaper consultant and writes The Inksniffer blog, blog.inksniffer.com