Lately, the word “technology” has been making the rounds.  Partly this was triggered by the GSEs (Fannie/Freddie) “modernized” report requirements.

The Appraisal Foundation also recently released its new 14 page Advisory Opinion 41 (AO41):  Use of Technology in an Appraisal or Appraisal Review Assignment.  The object is to “address the appraiser’s responsibilities” – under USPAP – “when technology is used.”

The Introduction page to this Advisory Opinion starts the discussion of “technology,” “tools,” and USPAP principles and standards compliance.  It concludes on reliance on the tools, documentation, workfile retention, and (possibly) necessary disclosures.

Since our hope is that this continuing series will connect appraisal issues to public policy, it is important to know what we are talking about.  My experience is that a lot of words around appraisal are used (or misused) colloquially, loosely, or even carelessly – rather than critically.

The Words

Technology:  “the practical application of scientific knowledge.”  I was surprised by the fact that every given dictionary definition emphasized the words “practical,” “scientific,” and “knowledge.”  My personal frame is that it is important for analysts separate the two general technology types:  product tech, and process tech.  Product technologies are generally things analyst/appraisers buy.  Process technologies reflect or help build things appraiser/analysts sell – such as summaries, opinions and analytic results.

The AO-41 introduction page is both general and specific.  It says, “Appraisers use a range of tools, including spreadsheets, regression and statistical software, automated valuation models, machine learning systems, and generative artificial intelligence.”

It appears the Foundation sets up use of the word “technology” in loose manner.  (Except for specifying “generative AI” amongst the several and broad types of AI relevant to analysts.)

Each of the above AO-41 “tools” has different context, impact, and nuances.

Tools:  There are several dictionary non-analyst definitions, but the most relevant seems to be: “Software designed to evaluate metrics, trends, or project performance.”  Nicely, this definition seems quite appropriate to what appraisers do.  (But not product stuff:  physical hardware and implements.)

The main point of the AO-41, is that the tool does not comply with credibility … it is the appraiser who must comply.  The appraiser must be believable/credible with the use/output of any tool.

It is worthwhile for appraisers to have clear definitions of technologies and tools, and also to understand what is not a technology or a tool!

Examples:  A spreadsheet is a software tool.  Regression is a math formula, a tool.  “Statistical software” is a very broad class of such tools.  “Automated valuation models” are of a competitive industry, not a specific model or tool for appraisers.  (Being black boxes, they are not usable by appraisers as such.)   Machine learning systems describe an extremely broad set of paradigms, architecture, operational roles, and deployment patterns.  Generally, machine learning identifies patterns and makes predictions, without explicit programming.  This is difficult to reconcile with appraiser responsibilities under USPAP.  Finally “generative artificial intelligence” creates new, original content by learning patterns.

If we are to follow advice opinion, perhaps we should know what it is we are relying about!

Any coming public policy must strive for clarity, else we shall remain in the muddy “undrained” swamp.

What is a tool?  What is “technology.”  Next up:  An internal conflict in USPAP.  What do I do?!