When Anki geeks see efficiency and spaced repetition in the same sentence, only one thing comes to their mind:

Settings.

Yeah, those endlessly convoluted settings that seemingly require a computer science degree to comprehend.

But in this lesson, we’ll make it simple.

There may be a LOT of intimidating Anki deck settings, but not all of them are worth tweaking.

We’ll talk about the MOST IMPORTANT ones, and you’re also going to learn them in a very simplified way — in a way that integrates with the strategies we’ve discussed so far.

And when you learn them, you can start customizing your settings so that your reviews fit better in your lifestyle.

These are:

1. Learning steps
2. Graduating Interval; and
3. New Interval

In my experience, these are the most high impact settings that are also the simplest.

Let’s get to it.

Learning Steps — Baby Steps for New Cards

Learning steps are the intervals that a New Card has to go through before they become “affected” by the Anki algorithm.

The only way up these learning steps is by pressing “Good” during your reviews; pressing “Again” on a card puts you right back at the beginning of the steps.

Here’s what that looks like using a 10m 1d 3d learning steps setting:

So the learning steps are literally like a set of stairs — when you get a New Card wrong while you’re still in the learning steps, you fall back to the bottom.

This is a good thing for a few reasons, but mainly avoiding "Ease Hell" further downstream:

• Filters out the cards with poorer foundation (because these cards won’t reach the end of the learning steps)
• Learning steps give you enough repetition of a new card before the more aggressive review spacings kick in
• You can customize how often you see a card during the early stages

You can say that it sets the initial foundation of retention after you’ve done the encoding principle.

Now you may ask:

What happens when a new card is at the END of the learning step?

When you press “good” after that, then the new card graduates and becomes a Review Card.

Thus, it uses the next setting, the Graduating Interval.

Graduating Interval — Where the Algorithm Begins

The Graduating Interval is simply the initial interval of a review card once it has gone through all the learning steps.

This is crucial, because all the intervals beyond this point are calculated from this specific setting. Which means modifying the graduating interval will have an ENORMOUS effect on the spacing of your reviews. More on this in a bit.

Now let’s say you have 10m 1d 3d as your learning steps and 7d as your graduating interval.

If you got everything correct, here’s what your reviews will look like:

Beyond that, the next review schedules will be calculated based on this interval — and by default, the next review schedule will always be 2.5x of the current interval if we press only “Good” for correct answers.

At this point, review calculation becomes a bit more tricky because it now uses the SM-2 algorithm. Pressing “Easy” or “Hard” will create a lasting effect to the calculations.

To put it simply, though:

Just like an adult, all actions at this point will have a huge consequence:

• When YOU press “easy,” the spacing grows a bit faster than a 2.5x rate from now on. The reviews appear less often until you press Hard.
• When YOU press “hard,” the spacing grows a bit slower than a 2.5x rate from now on. The reviews appear more often until you press Easy.

The specifics are NOT important, because quite frankly, the 2.5x rate of spacing growth is extremely more efficient than what you were previously doing before Anki.

On top of that, we have established a robust foundation using the encoding principle, because real talk…being “time-efficient” but not learning anything for real is the TRUE waste of time.

Plus, we’re going to be using “easy” and “hard” conservatively, anyway.

Finally, note that each interval for your cards grow until it reaches a “Maximum Interval”.

The Max Interval for cards is 36500 by default. That does NOT make sense. There’s a lot that could happen in a year, and there’s most likely interference happening in practice if you’re learning new things often.

So if you want to be able to review a card AT LEAST once per year, then you can set this to 365.

Or you can use this to compute:

Max Interval = 365 / (Min. times you wanna see a well-reviewed card in 1 year)

Okay, that’s all for getting review cards right.

Now what happens when you got a review card wrong, i.e. you pressed “Again” for a review card? What setting comes into play?

That’s where the New Interval comes in.

New Interval

By default, getting a review card wrong will reset a card’s intervals to the ground — zero. Suddenly you’re reviewing a well-learned, well-repeated card as if it were a new card.

Again, that does NOT make sense at all.

Enter the New Interval setting — when you set it to, say, 50%, then pressing “Again” on a review card will return that value only to 50% of your current interval.

Here’s an example for a 50% (or 0.50) New Interval setting:

You can change this depending on how conservative you want to be with your cards, but I found that 60% works just fine with my 4000+ card collection.

Putting it all together: “So what are the best settings for X subject?”

The “best” settings are impossible to find, but a “good” set of settings is much easier.

What you need to know is that you don’t need to “review a card when you’re about to forget” — that’s theoretical, unknowable BS.

You only need to have a guess of a long enough interval to balance the spacing (for efficiency) and retention (for confidence).

So it is based on your OWN needs.

If your goal is to lower the sea level of your reviews, this simply means adjusting the Learning Steps and theGraduating Interval based on the density of the things you learn.

The better learned/interconnected the ideas are in a single deck, the longer intervals you can use.

For example, Anatomy subjects don’t seem to have a ton of encoding opportunities aside from when you actually see and find the damn body parts yourself.

Sure, you can use encoding techniques for that. But this this doesn’t change the fact that the ideas in Anatomy aren’t very connected. So you could use shorter intervals, like:

• Steps: 10m 1d 3d
• Graduating Interval: 5d

Cookie-cutter settings

I’m not a cookie-cutter guy for the most part, but I thought this would make things easier.

Again, all we are changing here are the Learning Steps and the Graduating Interval, so it’s not so much of a big deal.

So here are the settings for the intervals:

• For “memorization-heavy” subjects: short intervals + more steps
  • Steps: 10m 1d 3d 7d
  • Graduating Interval: 10d
• For conceptual subjects: longer intervals + few steps
  • Steps: 10m 1d 6d
  • Graduating Interval: 7d
• For general learning: make your own
  • Steps: 10m 1d 3d
  • Graduating Interval: 7d

For the other settings:

I just set the “Easy” interval to something greater than the graduating interval. Useful for when you want to skip the learning phase. (I don’t recommend this unless you’re putting in things you’ve known for a long while…)

Will these make you significantly more efficient?

Just to be clear, You don’t become super efficient nor “hack your memory” (ugh), but it will make reviews more spread out, and the intervals now start to make sense IN PRACTICE.

And speaking of spreading out reviews, there is one more thing you can do to make your reviews even better:

Making them leaner.

More in the next lesson…