What additional information--since you mentioned anonymity, I mean in an entropy/information gain sense--is learned from income taxes? I should say from the get

What additional information--since you mentioned anonymity, I mean in an entropy/information gain sense--is learned from income taxes? I should say from the get-go that I assume these public income tax disclosures would map names to income, but not names to addresses. I would think we already have extremely predictive, though broad (e.g. interval censored), indicators of income and wealth that are sufficient to cause the threats you mention. Currently, we can map ZIP code to income. And we can map targeted names to ZIP code, or use consumer files to map large numbers of names to ZIP code. There may theoretically exist a class of recluse who lives in a shack but makes a gajillion dollars. That person would be partially de-anonymized by this process. But under the current threat model, there are paper-poor people living in expensive or apparently expensive places who face additional risk as well. Net-net, hard to say which group has more people. If the costs are the same, then the argument should focus primarily around the comparative benefits. That's a case I expect the article and other comments can articulate better than I can.