> the reason the magnitude doesn't matter is that those counts will be much higher in longer documents ...
To be a bit more explicit (of my intuition). The vector is encoding a ratio, isn't it? You want to treat 3:2, 6:4, 12:8, ... as equivalent in this case; normalization does exactly that.
Probably not a good approach to take here because there are plenty of examples of everything he said. However there are simultaneously plenty of counterexamples. For example certain west coast cities where shop lifting has been de facto legalized due to non-enforcement rooted in the name of the social cause of the day.
Actually what he wrote appears to match a certain political stereotype in the US while being the opposite of the other one. So I guess it says more about his view of the US than anything else.
> > at least when they want any at all. It's not consistent.
This is because you are treating Americans as a monolithic group, failing to differentiate between various major clusters.
However I will say that at least
> > A lot of Americans are way too into punishment. It's seriously like a fetish for them.
It actively harmed the environment. Once the thing has been manufactured doing anything less than driving it into the ground is wasting the upfront environmental cost.
I think it depends on the nature of the task for which the skill is used. But I agree. There are certain things that are just not realistic to master without rote drilling of some sort of supporting material.
At the same time I think rote drilling gets a bad reputation because poor teachers will tend to fall back on it in instances where it isn't particularly useful.
I personally had a low opinion of it until university, where I encountered two independent topics where it was clearly necessary.
You might have something unusual going on there. At the beginning I, and to the best of my knowledge everyone I grew up around, was also taught to sound words out. At some point I stopped needing to do that. That happened quickly for the common words and more slowly for the uncommon ones. At the extreme end, for example extremely specific medical or chemical terminology, I still have to do that on account of never having encountered the word before.
At no point did I intentionally memorize word shapes. At no point in my life have I ever been conscious of recognizing words by shape. That might well be what's happening under the hood, but if so it's not something I have any awareness of. I just see a word and (in most cases) instantly know what it is without thinking about it.
Presumably your job is to get results. If catering to someone in some way is able to achieve that then it's a perfectly legitimate approach to consider.
Formatted like a formal academic publication. No way (that I can tell) to grab a pdf. Comes across as a blog masquerading as academic literature to me. Am I wrong? Did I miss something and there's an offline version available?
Pages served up over http are ephemeral. An absolutely essential part of formal academic literature is the archival aspect - self contained, immutable, and referenceable in an unambiguous manner.
There's also an immediate practical aspect for me. I will likely never get around to reading this because I will forget it exists because my "reading list" consists of a pile of pdf files.
I have a very similar issue with subtitles in MPV sometimes. Some unknown (at least to me) combination of codec, encoding settings, subtitle format, switching subtitle tracks, and seeking will utterly break it to the point that I have to exit and reopen the file. I can't be bothered to debug the issue (so far).
In general I don't understand the negative comments I see VLC get. I've never encountered major problems with either piece of software. I've encountered minor bugs and annoyances with both.
> It's not at all clear that "restarting with a small subset of cards" is better.
Sure it is. It could grow as quickly as allowed for given the time invested by the user. That could mean a return to the full subset within the span of a single day, or it could mean many months. Perhaps even never. It all depends on time invested by the user going forward. Starving regular review for the sake of verification is an example of an algorithm failing when faced with the real world.
At minimum it is clear from what was said that (better) prioritization between conflicting goals is needed. That somewhat matches my own experience with it from years ago. The algorithm was simply not flexible enough to fit my own usage patterns. In other words I was not part of the target audience, which I found frustrating because I very easily could have been.
Agreed. Explicit is only better than implicit when you have a mechanism that can reliably identify instances where you forgot to do something.
Actually for resource management I'd go so far as to say that implicit is better than explicit in the vast majority of cases. Now if only the rules that determine what happens during initialization in C++ weren't so horribly convoluted.
The comment you replied to described an edge case and explained why it's broken in that particular case. You haven't actually responded to the example provided.
> it's just not clear if you're going to recall them or not. Figuring that out becomes the priority
Presumably the priority ought to be (re)starting with a small subset of cards and gradually trickling the others back in. The algorithm needs to account for time spent by the given individual and adapt to changes in that over time.
I haven't used Anki for about a decade so I'm not familiar with the current state of things. At the time a major factor in my dropping it was that I found the algorithm to be more of a hindrance than a help.
Similarly, if you visit certain instances you will be greeted with the polar opposite of that. The inhabitants of some instances are more political than others. If you actually create an account you will be able to pick and choose whose posts you want to see.
The strictly chronological approach seems to work fairly well as long as you only follow a few people but it doesn't scale very well.
You're swinging right back to the context and meaning of the rules that were presented during the assigned task. What I wrote isn't really about that. It's about the assigned task itself and the self assessment of whether or not it was completed faithfully. That's where the cognitive bias becomes plainly observable.
There are the rules presented during the task. Separately there are the instructions given for the task itself. To me it feels a bit like a failure to reason with layers of abstraction. Almost an inability of most people to reason about and interpret the rules differently in different contexts. They're stuck in the "real world" context and can't seem to switch to the "hypothetical framework" context laid out in the instructions.
> From the other perspective, hyper-rationality is a dysfunction ...
When obstinately adhered to in a general context, certainly. This was not a general context. It was an exercise with specific and reasonably unambiguous instructions. Openly deviating from them would be quite different than what can be observed in this comment section - deviating while claiming to have followed them.
On any other website I would be inclined to assume a certain lack of literacy or comprehension. Not so with this audience.
I think it's a fascinating practical example of how "baked in" cognitive bias is. The sort of people that use HN tend to be highly analytical. Yet nonetheless we see a massive public display of people rationalizing their failure to directly answer the question that was very clearly and unambiguously asked while pretending that they did.
The logical exercise is extremely close (by design) to one that commonly occurs in everyday life. In real life people want to bend the rules to achieve a certain outcome when applying them. They don't want to say "well a rule was violated but I'm exercising discretion". That's on full display here even though no meaningful outcome is actually being determined in this case.
Edit: I have no idea why anyone would downvote a link to this article. It directly answers the question with a decent level of technical detail. We are nowhere near single atom sized features yet, despite what node names might lead you to believe. There's still quite a ways to go.
To be a bit more explicit (of my intuition). The vector is encoding a ratio, isn't it? You want to treat 3:2, 6:4, 12:8, ... as equivalent in this case; normalization does exactly that.