I am reminded of the whole “low-key anki” series, which you can read through the beginning here if you’re interested.
There are two key insights I would extract from it.
We can’t really judge how well we know something
Not to mention the ever-present possibility that the only reason you found a card easy to answer in a specific moment was due to the same sort of memory fluke that I talked about in the previous section. In fact, ja-dark actually cited studies that found that people’s subjective judgments of how well they know something are highly inaccurate. So, just because a card felt easy to answer at the specific moment you reviewed it, that doesn’t necessarily mean that it is less intrinsically difficult than Anki had previously thought, and thus can safely have its interval grow at an expedited pace. If you mistakenly grade a card “easy,” it’s extremely likely that its interval will grow too quickly, leading you to lapse it in the future.
decision fatigue
From Wikipedia:
“In decision making and psychology, decision fatigue refers to the deteriorating quality of decisions made by an individual after a long session of decision making. It is now understood as one of the causes of irrational trade-offs in decision making. For instance, judges in court have been shown to make poorer quality decisions late in the day than they do early in the day. Decision fatigue may also lead to consumers making poor choices with their purchases. There is a paradox in that “people who lack choices seem to want them and often will fight for them”, yet at the same time, “people find that making many choices can be [psychologically] aversive.” Notably, major politicians and businessmen such as former United States President Barack Obama, Steve Jobs, and Mark Zuckerberg have been known to reduce their everyday clothing down to one or two outfits in order to limit the number of decisions they make in a day.”
You are already straining your mind trying to correctly answer your cards; you don’t want to be wasting additional mental energy deciding how well you knew something. Ja-dark also cited a study that found that self-scoring how successful you were at recalling something taxes working memory. Eshapard summed it up perfectly when he said, “From my experience, it really slows you down if you’re always trying to figure out whether a card you remembered was Good, Hard, or Easy. These are really subjective categories and it’s a lot of work to try to use these options for all but the most obvious cases. […] Trying to remember the facts is mental effort enough for me. I don’t need to split hairs over how difficult the card was.”
Either you already know the card as well as you would like to, and press “good,” or you would like to know it better than you currently do, and press “again.” It’s that simple. And because the answer to this binary choice should be obvious in nearly all cases, not only is decision fatigue minimized, conserving your mental energy for actually answering your cards, but also time spent making decisions about how to grade cards is reduced; shaving just a few milliseconds off your average answer time can lead to saving large amounts of time in the long term. In this way, low stakes (no ease penalties + high “New Interval” after lapse) makes grading less stressful, and binary choices make grading more clear-cut, in turn preserving mental energy and allowing users to grade cards more quickly.
The above two insights drove some reasoning around the Undo button, because it can add cognitive load–you have to decide if it’s worth undoing or not every time you answer a card.
In the case of the Undo button, I think the cognitive load is low enough and adds value 
Both of the above key insights could be challenged, e.g. we can justify how well we know something by how often we interact with it in immersion as @lorentz points out and that sort of reasoning sounds great.
The above two insights are interesting to consider with the central problem: time to complete reviews.
We’re still not sure what we’ll do next about it, all of the suggestions thus far in this thread have some weight to them regardless of the above two insights.
Thinking out loud, there are some ways to make reviews even quicker that aren’t necessarily about “skipping” but instead changing the interaction.
One idea would be Autocomplete – There’s still the active recall part, but your keystrokes are reduced since you can just tab to the answer you think is best.
Thanks for the feedback!