You expect a tool to take this into account when it was the first "i" sale which sold for a significant price? That's asking a lot. Most domainers wouldn't have appraised this for even close to this amount previous to the sale.
I don't disagree. You're right about human appraisals because there is tremendous variance in even human assignment of value. And in a way, that's actually the more important point of this thread: a bot, and humans as well, have difficulty achieving a reliable system of domain value assignment. There are all of the obvious metrics that we use to estimate value (on which estibot relies completely), but then there are
many variables, both subjective and complex (like apply to rare collectibles, works of art, or business assets), that can greatly affect the price of a single domain. Personally, I would never rely on any "automated" mechanism, and place little weight on a single human appraisal. This is especially true for higher qaulity, upper echelon domains the likes of hotels.xxx, business.xxx, etc.
Common appraisal metrics have more applicability for lower quality domains that have no intrinsic brand appeal. A common "trick" is when buyers attempt to assign low value to a generic domain based on traffic metrics. Traffic is one measure of possible value, but a poor measure in many situations. Take iReport.com. The brand appeal and other somewhat esoteric qualities far supercede type-in traffic. It takes a savvy, experienced individual to weigh the more subjective variables. Value is not a perfect science and depends in large part on market/buyer perception, need, suitability for a specific purpose (as in ireport), business niche, timing, etc.
Conversely, a poorer quality name that has amassed traffic through development, seo, and promotions is better suited to traffic stats analysis as a measure of value/price. The range of value/price can be much more easily confined to a narrow range when the name has no brandability or true generic power behind it. Thus, the appraisal industry.