[pseudonym]

While that's true, that's only the case for sites you regularly visit. If you get linked to a random site or blog post from HN, Reddit, Digg, Facebook, or what have you, your first thought is not going to be "gosh, I really want to pay for a subscription to this site". In fact, if you run into a paywall, you'll probably just bounce back to the site that linked you to the article and ask for the article text or a non-paywalled version of the same.

[oneeyedpigeon]

Maybe people would be happy (well, happier) with extreme micropayments. The use case you describe should be a rounding error as far as my bank account is concerned, but enough when multiplied by reasonable traffic to sustain online publishing.

Does anyone know the actual value of a single ad impression?

[joshvm]

Depends on a large number of things. As an advertiser you can pay based on location on the site (e.g. the advertiser will pay more for a banner than a little side unit below the fold).

Some adverts are pay per view, pay per click, some are commission based (referrer gets a cut). Sometimes it'll be a few dollars for a few thousand views, other times it'll be a few hundred dollars for a week's visibility.

For click throughs, fractions of a cent to a few cents I guess. Commission is obviously worth a lot more (a few percent of the purchase price typically), but harder to convert users.

Have a look at buysellads to get an idea of how this works. Unsurprisingly Adblock has a field day if you try and view the website with it turned on ;)

[_delirium]

> Does anyone know the actual value of a single ad impression?

This is often summarized as RPM (revenue per mille) or eRPM (effective RPM), i.e. how much income a site's ads bring in per 1000 page impressions, regardless of how it's paid out (per-click, per-view, per affiliate purchase, etc.). And that varies hugely between sites, anywhere from below $1 to above $100. A typical case is around $5-10 for many publishers, I believe. So that's about 0.5-1.0¢ per page view.

[ansgri]

so why don't they charge me 10..20¢ per article, with an option for $x (or $xx if they publish a lot) unlimited monthly subscription, should I like them. Seems like 10-20 times more profitable, yet a rounding error.

[_delirium]

Micropayments seem to have had trouble getting off the ground. It's not a lot of money to pay a few pennies per article, but if you put up a paywall and make people pay it, they seem to go elsewhere. Either people are too cheap to pay it, or the interfaces are too clunky, or some mixture of both.

Subscriptions do work for some sites, but from what I can tell most have trouble selling enough subscriptions to cover what advertising would pay for.

[ansgri]

I suspect the clunky interfaces are to blame: if the paywall was only 1 click ('confirm $0.1 purchase') instead of a long long form to fill most people would just click.

[oneeyedpigeon]

Even that is too clunky, IMO. I really liked the model that Readability tried to pioneer: pay a monthly subscription, view websites, at the end of the month, all money paid is distributed amongst publishers in proportion to views. Readability's purpose meant that 'view websites' was actually a more complicated step involving explicit user choice, but I don't see why this same model couldn't be used with a set of 'certified' publishers and a central authority. The common link could be 'publishers who reject advertising'.