I'm not entirely sure what the reference class is either, but my brain matched this to my professors' lectures about paying attention to users/customers. (Most of these in undergraduate university classes in things like human computer interaction, web design, and data modeling.) I haven't had a real job yet, but I'm told that designers failing to focus on what users actually want is a major problem in software and web design. (Though it's getting better.)

These are the examples I could think of off the top of my head:

• Browsers with big toolbars that leave you with not enough space for the web page
• Facebook's web design (I don't use Facebook much, but I have difficulty finding things when I do. It also seems to be routine by now for Facebook's design changes to completely change the layout, and for users to complain very loudly when they do.)
• Possibly Windows 8's metro mode - I have not used it myself, but I'm told that it's the default for non-mobile computers and works terribly for them. Whether this matches depends on what the exact reference class is - Microsoft was trying something new, but it seems like the sort of thing they might have noticed in testing.
• I see Tumblr users doing a lot of odd things to get around the built-in constraints. Things like posting screenshots of "asks" as photos and writing the reply below them (instead of replying normally), because captioned photos can be reblogged but asks cannot. Whether or not this is easily fixable depends on how well written Tumblr's source code is, but it doesn't seem like it should be a difficult problem to solve.
• The default toolbars of Microsoft Office products include rarely-used functions (which makes it harder for typical users to find the things they actually want to use)
• University or corporate homepages that don't emphasize the things that people actually come to the website to find - I suspect this happens when a committee picks what goes on the homepage, which makes it become a matter of politics instead of design

Doors are the classic non-software example - designing doors that people don't try to push when they need to pull (or vice versa) is supposedly a solved problem (give a handle that doesn't look pushable when you need to pull, and metal plate instead of a handle when you need to push), but designers don't always do that, probably because they focus on making something nice looking rather than practical. I suspect that sinks with knobs instead of lever-type faucet handles are still made for the same reason.