Little has gone right for Yahoo, which is one reason the flailing tech giant decided this spring to sell off the bulk of its patent portfolio. Now it looks like the move will produce a further disappointment as a report suggests most of the patents are weak or outright worthless.
A summary of the report, published Monday by TurboPatent, claims that 44% of the thousands of patents have “high severity” issues and that nearly all of the patents, including pending ones, have deficiencies that could lead them to be invalidated. These findings throw further cold water on optimistic predictions that Yahoo YHOO could make up to $4 billion from the sale.
TurboPatent, which produced the 155-page report, is a Seattle-based analytics firm that analyzes patent quality and helps firms with the patent application process.
According to CEO James Billmaier, the company deployed its machine learning and artificial intelligence technology over several days to assess the Yahoo portfolio, which consists of 1,757 patents and 896 patent applications.
The process looked for similarities between Yahoo’s portfolio and other patents to determine how vulnerable the Yahoo ones would be to challenges that claim they are invalid. The conclusion was stark.
“[Our] analysis found Yahoo!’s Excalibur patent portfolio contains a much higher percentage of potentially worthless patents than the average patent portfolio,” the company said in a statement.
In addition to the 44% of patents and applications deemed to have “high severity” vulnerabilities, TurboReport found another 53% had low or moderate vulnerabilities.
According to Billmaier, “high severity” indicates an over 90% likelihood a patent would not survive a challenge, while”moderate” indicates around a 50% chance of such a likelihood and “low” equals around 10 or 20%. He added that the firm’s assessment was based on an automated review, but also involved human spot reviews as well.
Get Data Sheet, Fortune’s daily tech newsletter.
PatentTurbo’s findings, if accurate, make the portfolio an unattractive asset since the patents it contains would be difficult to enforce. They also add credence to an earlier news report that Yahoo’s gambit, which involved putting the 2,653 patent assets into a shell company called Excalibur, amounted to a “fire sale.” Yahoo has denied such assessments.
“As we’ve said from the beginning, we believe this portfolio represents significant value,” said a spokesperson, who declined to provide further specifics.
Alice’s Arrival and a Shifting Patent Landscape
TurboPatent identified four categories under which the Yahoo patents could be found vulnerable. One of these categories concerns Alice, a landmark Supreme Court decision from 2014 that found existing ideas can’t be patented simply because they are implemented with the help of a computer. This has cast a cloud on many software-related patents, including those in the Yahoo portfolio.
Another TurboPatent red flag that a patent is weak is a dearth of citations from other patents. The company, which considers 30-plus citations to be the sign of a strong patent, found that only 169 of the Yahoo patents met this mark, while also noting that 725 of them had been cited by exactly zero other patents.
The other two categories used by the company to flag suspect patents are title and assignment issues, and “technical quality,” which evaluates the language and process used to seek patents.
TurboPatent is not, of course, the one that gets to decide whether Yahoo’s patents should be valid. It can only predict, and its predictions might be wrong.
That said, the company and its CEO appear well-versed in the arcane aspects of patent law. More importantly, unlike other companies that offer intellectual property services, TurboPatent does not appear to have a strategic interest in the Yahoo sale going off well or poorly—Billmaier told me it is simply using the Yahoo portfolio as a way to show off its technology.
More broadly, the findings are not surprising given the nature of Yahoo’s business and changes to patent law that have taken place in recent years.
Many of the company’s patents fall under categories like “database and file management” and “financial…price determination,” which are likely to confront Alice challenges if anyone asserts them in a lawsuit. Meanwhile, changes to the patent system (recently upheld by the Supreme Court) have made it much easier and less expensive for defendants to challenge low-quality patents—held by Yahoo or anyone else.
The upshot is that predictions of Yahoo receiving $4 billion or even $1 billion for the patent portfolio appear way, way overstated. (Potential buyers would also be aware that some of the patents date from 1996 and will expire soon, and that maintaining the portfolio could cost more than $20 million in fees to the Patent Office).
Finally, there is the question of who would want to buy the Yahoo portfolio. As the veteran tech writer Steven Levy noted, it could go to another tech giant like Google GOOG , or it could be snapped up by investors who want to “unleash a flurry of patent trolling.”
Whatever the outcome, the Excalibur sale is likely to mark another milestone in Yahoo’s apparently permanent decline.