Google was worth 1,838,389 workers in 1998, maybe
herdrick.tumblr.com
herdrick.tumblr.com
To give some examples of different kinds of value you could assess for a company like Google - Its value for its users (for early Google this would be basically how much value is gained by using Google as opposed to an alternate search engine) is for instance very different from Google's 'use-value' (how much money Google make from it), its market value (driven by a lot of speculation as to what Google's future might be, as well as its 'use value') and its 'liquidation value' (of assets - assumedly very low when Google began).
The link comes closest to looking at the value for users, however this would be better ascertained from a closer understanding of what Google was supplanting at the time (other search engines such as Altavista who had results which were poorer, but not hugely so) and then trying to derive some basis for how this might affect the 'user', rather than what the cost for Google might be to implement their search mechanism in a technology-backward way.
[edit] To clarify that I'm talking about valuing Google in its infancy (as per the OP) rather than Google at present (which is obviously more than just search) - thanks to martinced's comment below.
and:
"(other search engines such as Altavista who had results which were poorer, but not hugely so)"
I upvoted you but for many Google is, and since quite a long time, just so much more than just the search engine.
Actually I don't even search that often anymore. My girlfriend hardly ever searches for stuff on the Internet: online life for her is GMail, YouTube (I take it you could hard there's search inside YouTube), eBay, Amazon, FB and a few other sites but that's basically it. What Google offers goes way beyond search: to me the biggest benefits are, by far, a spamless webmail, online backups and online editable/shareable Google Docs. Search is a very very very distant fourth or fifth service compared to the rest that Google is offering.
Perhaps you can quantify how much gasoline and time jets save as compared to other methods of travel, such as by train or boat. But going on some hypothetical equivalent of Google search with hundreds of thousands of librarians made me stop reading midway through.
Just because I paid $4000 for a hand carved peruvian hardwood iphone case does not mean that its actually worth that.
That also accords better with general English usage, where what something is worth and what a particular person paid are not considered generally equivalent. There exist phenomena such as "overpaying" and "getting a great deal", where payment in a particular instance diverges (in either direction) from this more robust notion of "worth". Indeed, if that were not the case, arbitrage would be impossible.
Edit: It just hit me, a spacesuit sitting on a shelf has absolutely no value. Only when used for something does it have worth. So, since the biggest practical purpose of spacesuits is keeping people safe while they fix things in space, the labor worth of a spacesuit is the cost of building a robot to replace it and the human inside it, minus the cost of the human.
If the spacesuit is being used to so something a robot could do better, (like, say, hold a camera steady in orbit) then the labor value of the spacesuit is negative.
But spacesuits... hmm. That is really tough.
Yeah sure, with enough dedication you can measure anything and everything. But if it makes sense is another question.
I think even pg put in one of his more recent comments the question forward how innovation should get measured. And the answers were (at least for my taste) very, very limited.
This is called the "labor theory of value" [1], and dates back to Aristotle. It was useful for agrarian economies but misses some crucial points, and so modern economists generally don't consider it all that useful. Adam Smith's work (which later became classical economics) was largely in reaction to shortcomings in the labor theory of value.
It's an eminently fair metric in the context the GP was trying to use it for, which was to estimate the relative benefit to consumers (something like the "consumer surplus") compared to other innovations in other times and places in order to gauge relative progress esp. in saving labor.
Work force is grossly overestimated. 99% queries could be filtered with simple encyclopedia lookup. For rest some sort of caching would be used. I think you would need about 10 000 people to constantly handle this type of load without any computers.
I have a few comments about second article "The boat engine is worth 33500 Egyptian slaves"
First ancient Egyptians did not use slaves. Workforce in Old Kingdom which build pyramids were free men. Later it was more similar to feudalism. (BTW: women had more rights in Ancient Egypt than most women in 20th century)
Also towing is not really necessary on river Nile. There are northerly winds for most of year, so one could just use sail to travel upstream.
And third; 'Pharaoh Nile Express ' actually exist, but it had only 30 rowers :-)
All Google did was better results over Altavista
2. This is more subtle: the value of an innovation doesn't vary necessarily linearly with some attribute of it. This is because an innovation has no intrinsic value - it only has value when in the hands of a person who needs it. That is, value is an attribute of the need, not the innovation. Might sound academic, but it matters when an innovation is more powerful than needed. It gets better and better, but nobody care (see The Innovator's Dilemma). In your example, oddly enough, latency really does matter (though you can imagine a certain speed where it's indistinguishable from instant for practical purposes, probably around 100ms). Also in your example, the size of the internet also matters, because the more there is, the more likely the desired answer is present, and the more likely it is to be relevant. So, this subtle point isn't relevant o your specific example, but I think it is relevant to your concept of measurement with a "scale of innovation", in general.
But, if you figure that google's profit could instantly be about 10 times higher than it is now by firing the 9 of 10 employees that are not in the profit-making search group (I admit to gross simplifications) the JobsDisplaced=Profit/YearlyProfit-$60,000 value would come close to your ~2million number.
This would mean a much less microscopic view of its place in history but without at least basic knowledge of /how/ Google's implementation was better how could you know any of your calculations are accurate?