I did 98,000 Anki reviews. Anki is already dead

I think people underrate the utility of a premade deck when constructed well and used appropriately.

A use case I've found is if you can find a deck that corresponds to a book you're reading.

I found a deck for the Rust book and it's structured such that you can see cards about things in the order you read about them. You simply read the book as usual, learning from your reading and entering code into a terminal as instructed, and then test you understanding with the cards.

When you end up reviewing older cards, you end up getting the benefits of putting them in long term memory but you also get the opportunity to make more connections as you revisit concepts which has its own benefits for deepening understanding.

I've found this makes reading the book 10x more effective. I get so much more out of it.

This all depends on having a source from which you're learning and the deck is just for testing understanding.

But yes anytime you're using Anki to learn/understand instead of to remember, you're likely misusing it. Anki is a tool for memory.

I would go further to say that anki just gets an update where you can now have more dynamic context surrounding the words on cards. Just bold the target word. The same font is also something that can be easily fixed.

Correct

I strongly agree with the author's claim that Anki incentivizes you toward recognizing the target sentence, but not with his main claim that it's dead. For context, I've been using Anki about 2 years and ~180k reviews.

Grinding the same thing over and over works well for some things but overtrains for others. I've noticed that I'm overtrained on the default sans-serif Japanese fonts, for example, which causes slight but noticeable pain when I switch to a different font - especially a more traditional serif one which tends to have a lot more visual information. Even if you know all the stroke order stuff, suddenly switching to a new font to read hundreds or thousands of characters is a lot of extra input for your visual cortex to process. It bothers me enough that I'm switching things around to match the JLPT exam typefaces as closely as possible so I don't feel it dragging me down under the pressure of a ticking clock and high stakes.

On the other hand I think the Anki engine is more than flexible enough to accommodate dynamic cards, and it still works great for multimedia and the like. I'm leery of handing too much of this over to LLMs both for the hallucination risk the author mentions, and the reliability of the audio generation - pitch and rhythm are bad on LLM-generated audio in English and I suspect they'd be even less reliable in Japanese. A few sentences sound fine but when you start doing it on longer texts or across thousands of sentences it's gonna creep in, and for language learning toxic input is deadly.

IME the best thing once you get past the beginner stage is to read as much as you can. The biggest day-to-day gripe I have with Anki (but which is not actually the fault of the software) is you're learning words and/or sentence patterns in isolation, so the work you do in understanding a sentence is thrown away immediately. You might spend 3 minutes pondering the intricacies of 'if I don't leave the house by 7am, I'll be late for my train', but then the next sentence is 'I savored the sound of my enemies' screams while watching their ship sink.' In contrast to the i+1 competency issue, you're also missing a +1 contextual issue - the little exercise you did to understand the first sentence doesn't carry you forward into the next one, so you're starting from zero every time. Even in the exam situation, you're tested on 4 or 5 questions related to each written or audio subject matter which can help you maintain focus even where there are gaps in your understanding. I think the constant fragmentation of attention by individual sentence cards is bad for concentration and retention. Although Anki allows for sequences of cards to be presented together by the deck editor, this context gets broken up both by your daily card limit and by the fact that review order has little or nothing to do with the initial order of presentation.

By contrast, when you're reading (even if it's just paragraphs or pages) new vocabulary items or grammatical structures tend to repeat and reinforce each other, you can draw contextual inferences about the meaning of unfamiliar words rather than just being given a meaning-token in isolation, and the new material is embedded in some sort of narrative or argumentative context. I've noticed that new vocabulary I pick up from reading beds in far faster than that from isolated word or sentence cards, where everything is isolated and the only context is 'that day's Anki session'. It's like the difference between enjoying a movie and and watching an endless succession of closeups from different scenes. You could work out the story from the jumbled sequence of closeups, but it would take a lot longer and not really be enjoyable or coherent-feeling.

In my view the best changes Anki could make in this regard is to get away from the deck as the main data structure and have an intermediate structure of sets, essentially sub-decks that have something in common and can reinforce each other. It would help a lot if there were better tagging options than just ctrl+1/2/3/4//5/6/7/8 too. I use a few for noting cards I am having particular difficulty with (yellow or red) and a few others for categories, but there just aren't enough of them. I would love to able to pull tag sets of just school stuff one day, just work stuff another, just transit cards another day and so on.

Somewhat related: you might dig Andy Matuschak's Quantum Country project, which aims to incorporate a text's flash cards inside the text itself.

https://quantum.country/

Later edit: I should have said "Andy Matuschak and Michael Nielson's".