I've read (I think in a Kodak white paper) that without direct comparison to an unfaded reference, it takes something like a 20% - 30% change before a person who is familiar with the original can detect a change. Assuming the ability of a print owner to detect a change is what matters, and given that no print owner will have an unfaded reference or be taking spectrophotometer measurements over the years for comparison, many high quality OBA containing papers can hang framed on an interior wall for a very long time without any detectable change.
I don't want to drag this off topic either, but the well known Kodak studies on consumer tolerances for objectionable fade don't really give us a good reference for image reproduction quality in more discerning applications like fine art prints. The kodak study was about consumer tolerances for easily noticeable and objectionable fading of photos in the consumer photo finishing market. The study thus used chromogenic color print technology which can fade quite differently in appearance compared to many inkjet systems, and it sought to determine levels of fade that would likely cause the typical consumer to actually object to overall print quality. This is a far different market than the market where my fellow printmakers are discussing the nuances of EEF paper texture and color compared to other expensive papers! The Kodak study derived the 30% density loss figure largely from mid point density areas in the prints (e.g., 1.0 density patches fading to about 0.7, etc) that could trigger typical consumer objections in overall image quality. This 30% density loss figure of merit is about balanced fade. Imbalanced fading in the color channels is less forgiving, and was found to be about 10-15% to most consumers. I don't disagree with those findings at all, but for fine art applications the tolerance values will drop much lower than that. We need to put it in proper context when dealing with high quality image reproduction versus "good enough to serve some purpose" print reproduction quality.
If you want to visualize a 30-35% "faded" but still balanced color appearance, take a look at newspaper print reproduction quality and compare to high quality glossy magazine image reproduction. Just look at representative output which strives for decent color balance. Newsprint white-to-black dynamic range is a very good example of that 30-35% Kodak cited density loss compared to reproduction densities in high-end book and magazine reproductions. While newsprint does indeed serve it's intended purpose and the newsprint buyer isn't complaining about image quality, does it really take a side by side comparison to recognize the significant image quality improvements afforded by 30% more printable density range? If you now remove richer more colorful prints from view, you may even start to become more appreciative of low cost plain paper print quality. However, to say that a 30% density change in image reproduction aim points goes essentially unnoticed without a dual-stimulus viewing condition (i.e., two prints side-by-side) is a real stretch. Hence, I don't believe the Kodak research means people can't notice density losses less than 30% without an unfaded comparison print handy, just that typical consumers won't complain about it when the print in question is an inexpensive photo finished print. And of course, some images have a wide tolerance for reproduction aimpoints, for example, a deep sunset being reproduced as a much lighter "sunrise: picture without anyone really being able to figure out what time of day it was really taken. Other images are far less forgiving. Ask any B&W printmaker how important the absolute image hue throughout the tonal range is to his/her artwork.
Which brings us now back on topic to EEF. Much of the appeal of EEF is indeed it's initial media white point. In LAB numbers, it has a b* value of about -5 (measured with UV-included spectral measurements on a Gretag Spectrolino). Increasingly negative b* values produce increasingly "bluish white" media white points and increasingly positive b* values appear increasingly "yellowish white" under D50 lighting. The typical media white a* values also come into play, but let's save that discussion for another day. Yellowishness versus buishness resides mainly with the b* value. Photographers are responding favorably to EEF's OBA-induced "bluish white" paper color which incidentally depends on UV content in the illumination, so as others have noted, if you want the viewer to see the appeal of the EEF bright white color, you can't use conservation framing that blocks the UV.
Papers without any OBA can achieve b*=0, but +2 to + 4 is very common. The super bright white OBA loaded papers will get to about -9 or -10 b* (again, UV included data). A change of 1.0 unit on the b* scale is considered my most color geeks to be a "just noticeable" difference in color. Looking at the measured values of many inkjet papers, the full b* range for commonly encountered media white points covers a range of approximately 10 b* increments. To put this into more perspective, the I* metric (also based on CIELAB colorimetry) that I use in my tests for color and tonal accuracy evaluations would rate an error in desired media white point of 1 b* unit (instrumental error removed) at 90% color accuracy, a 5b* change at 50% color accuracy, and a 10b* change at 0% color accuracy. In other words, if you took a brand new warm white paper and compared it to a brand new very cool white paper, the discrepancy could be about 10b* units, and the I* metric would say there was little or no retained color accuracy if you desired the cool white paper but ended up with the warm paper or vice versa. Many people are happy to use all of these media whitepoints interchangeably for their images, so one could successfully argue that for some end-users none of this media white point discussion matters one wit. Yet the profiling process usually adjusts the image colors relative to the chosen whitepoint, and many people like Dean's wife do easily perceive the impact that media white point can have not only on paper appearance but well into the overall image appearance as well. Simply put, media whitepoint color constancy (appearance under different illunminant conditions) and retention of color accuracy over time are both important factors to consider for discerning buyers.
In my own work, I try to balance all of these factors including color constancy, long term stability, and appropriate illumination and framing choices. I do use both OBA and OBA- free papers, but I hold my selections to a tighter initial whitepoint range from about -3 to + 2 b* values for color printing, and up to +4 b* value for certain warm toned B&W images. I consider the UV included/UV excluded color difference for each paper (which indicates potential problems with UV blocking conservation framing choices), and I evaluate how the initial whitepoint holds up in lightfade tests. If the OBA performance doesn't do well in light fade testing, it has a high probability of being gas fade sensitive as well, Gas-induced OBA burnout can occur far more rapidly and non uniformly than light induced OBA burnout. There are no clearcut recommendations when others ask about paper preferences or alternative, but some papers do maintain their initial qualities much better than others, and I would add scuff resistance to the pile of important factors to consider. Discriminating printmakers simply need to be aware of the subtle and not so subtle aspects of OBA use in fine art inkjet papers and then make an informed choice. The best way to evaluate all of these OBA dependent issues is with instrumentation and knowledge of what the CIELAB values mean to visual appearance. That's why I include the media white and max black values measured both UV included and UV excluded on the description page of each AaI&A test report. Its a real pain to gather that data, but it's important information for discerning printmakers.