Computer Generated Value vs Human Brain
Around 1992, representatives of LSI (long before they became completely useless as an AMC) visited us in California to discuss some way to involve automation (as in AVMs) to eliminate appraisals. When I pointed out that they had JUST THANKED me for saving them from a quarter of a million dollars over valuation as a result of pressure from one of their very influential customers, and that the very same issue had arisen from that customers misapplication of data AND their potential misinterpretation of that data, they understood the connection. A KNOWLEDGEABLE human appraiser was required to resolve the difference. It was my opinion then that no technology could ever adequately replace the ARTFUL application of SOME scientific techniques as well as experienced use and application of the best computer ever built…the human brain.
Sometimes my brain literally leaps to the “right” conclusions without me even being aware of the process. It takes short cuts. Now, in fairness, we all agree that we cannot rely on our personal experiences alone to achieve proper appraisal results. We temper and quite frankly usually suppress our “personal PCs” results in favor of more traditional techniques that are becoming more and more cumbersome as time goes on. The more that academia gets involved in appraisal analyses and critiques, the further away we seem to get from reliable products (UAD & 1004MC comes to mind).
We listen to self serving morons promoting the ‘era of big data’ and how wonderful computer science & technology have become. “Why there is virtually NOTHING they cannot do!” Or, so go the claims. Yet with all this high powered computer ability and giant databases economists still cannot accurately predict stock markets, or even what the GDP, or inflation rate or unemployment rates will be more than a quarter in advance.
Why would we believe these same folks can accurately analyze the often subjective motives of real buyers reactions to specific pieces of real estate in terms of price and marketability?
‘They’ have attempted to replace the ‘art’ and experience based part of our profession with the arbitrarily purely analytical side. Analyses that fall abysmally short of even being real science since they modify or make so many assumptions to support their conclusions that the results are meaningless.
Zestimate used to publish disclaimers about how limited their product was. Most similar services don’t even do that. Most users wouldn’t understand the explanations anyway…they are not trained appraisers.
Despite most states prohibiting the use of the terms “appraisal” or “valuation” by anyone other than appraisers, it seems that these internet wizards keep getting a pass for their products. It does serious harm to Americans and the general public by offering services that are so easily confused with meaningful USPAP compliant appraisals which have been determined to be the “minimum acceptable standards” for federally regulated transactions.
Why would anyone be so willing to accept something that does not meet MINIMUM standards?
- The New & Improved Fannie Mae “FRAUDULATOR 2.0” - May 15, 2023
- The Scam of Racial Discrimination by Appraisers - May 10, 2023
- What Is My Incentive? - September 20, 2022
If there is any relation to what is coming in real estate related big data, to the problems already observed in big wall street data, we’ll have a serious problem on our hands. What is programmed in a computer, is easily manipulated by a person with biased interests. There will never be an avm or data reporting standardization approach which will successfully mimic, predict, or be able to actually understand, what a human is doing. One good illustration of this is road frontage, and the UAD notations for road adversity or average or by park, etc, etc. Although UCDP may track and recognize individual points of entry from other appraisers reports, the UCDP does not consider over all road frontage, and cannot make critical logical decisions on where that consideration regarding location begins, or ends. All the UCDP does is match data points, but it is not sophisticated enough to draw logical conclusions based on complex factors, like how an appraiser may review the landscape, and recognize where obsolescence begins and ends. And if the jr greenthumb appraiser who came before you noted that as no traffic, oh boy now there is a data emergency that requires additional explanation and report revision. Of course it must be the second appraiser who did wrong, because it was his data which did not match the existing data which came first. By the time UCDP ever maps all the properties which come through the pipeline in a 20 to 30 year period, and could successfully make sense of some of it, the programming and software will be long obsolete anyways. Big data will never replace a human. The arrogance of the people promoting big data is entertaining. They’re the only ones who have faith in their technogods. We have faith in something higher.