Hacker News Reader

danabramovon Bluesky Trademarks ATProto

I don't mean it as a linguistic gotcha ("ha, you said instances! but we call them something else"). I'm saying this is a category error because the network topology is different. This is like saying that web is not free because people aren't running Google-scale indexes in their garages.

Tried to clarify this here: https://news.ycombinator.com/item?id=48932956

danabramovon Bluesky Trademarks ATProto

If you have concrete counterarguments to what I said in my articles, I would be happy to debate these counterarguments here.

danabramovon Bluesky Trademarks ATProto

I don't know what you mean by this. You absolutely can self-host all your data on your own. And if you also don't want to rely on PLC for identity, you make a `did:web` identity, and then you're completely decoupled from Bluesky-operated infra.

danabramovon Bluesky Trademarks ATProto

The reason is that "why aren't many people running Bluesky instances" is a loaded question — it presupposes that's something you should expect. But it's like asking "why aren't many people running Google". It's expensive to store billions of posts!

This is usually compared to Mastodon, but it's not apples to apples since "running a Mastodon instance" always mean a tiny partial view of the network. You can do a partial view of atproto too for the same cost, it's just not very interesting because what's the point? Every app can "see" the entire network (unlike in ActivityPub) so there's a lot less motivation to create these fragmented views into it.

I've already linked elsewhere, but I've written an article on precisely this topic, curious to see if this can clarify it for you: https://overreacted.io/there-are-no-instances-in-atproto/#so...

danabramovon Bluesky Trademarks ATProto

>First off, Bluesky doesn't offer any GUI tools for PDS migration. If you want to get off their PDS, you'll need to bust out command-line tools and possibly do some steps in advance of when you need to migrate in order for everything to work properly.

This is getting much better recently. I migrated to https://eurosky.tech/accounts/ with no command line — just needed to wait for a wizard in my browser to move my records. It could've been much faster and seamless (the app was confused after this and I had to re-login), but it's definitely accessible to non-technical people, and the share of independent hosting is growing.

danabramovon Bluesky Trademarks ATProto

It's a different network topology. ActivityPub couples three things into one box: community, hosting, and an app.

Atproto decouples them: there's app-agnostic hosting (which anyone can run), and there's hosting-agnostic apps (which anyone can run).

I wrote about it here: https://overreacted.io/there-are-no-instances-in-atproto/#so...

danabramovon Bluesky Trademarks ATProto

The solution to that is to create more apps, and for one of them to become more popular. This is entirely in our hands as developers.

danabramovon Bluesky Trademarks ATProto

>that also runs the only viable instance?

This is a category error. There are no "instances" in atproto.

https://overreacted.io/there-are-no-instances-in-atproto/

It segments the network differently (into "hosting" and "apps"), and on both axes anyone can run their independent app-agnostic hosting, as well as an independent hosting-agnostic app.

danabramovon Bluesky Trademarks ATProto

>Cool people use ActivityPub.

ActivityPub solves a completely different (arguably, smaller) set of problems. This is like saying "Cool people use email instead of web".

I have a few diagrams here that may help see just how different they are: https://overreacted.io/there-are-no-instances-in-atproto/#so...

>Single vendor owned & controlled standards always turn out badly, just a question when.

Yeah, which is why atproto is going to IETF: https://atproto.com/blog/kicking-off-the-atp-working-group. You can join the mailing list to participate in the discussions: https://mailarchive.ietf.org/arch/browse/atp/?gbt=1&index=.

danabramovon Ask HN: Why isn't Google indexing information about the AT Protocol?

Kagi's are good, yes. The issue is with Google's.

danabramovon Ask HN: Why isn't Google indexing information about the AT Protocol?

I can reproduce the same in incognito mode.

danabramovon Workers Cache

This is not true. I regularly get HN front page hits, and still use em dashes. Nobody accuses me of AI writing. Writing with em dashes is not a problem if you have something to say.

danabramovon Workers Cache

There is this one! Though I have not tested it. https://github.com/stephenturner/skill-deslop

I found it funny to just read through.

danabramovon Workers Cache

Please please please write your own blogposts. I already read “but here’s the biggest unlock, and it’s the part that’s hardest to see” all day.

I just don’t understand why undermine your own announcements by delegating comms to the machine. It’s disrespectful to the reader.

danabramovon Goodbye, Forever, Probably

In practice, "what's actually best for the community" (growing a community of engaged engineers who feel listened to and who understand what you're doing) is not necessarily easy to show by metrics. I think the author is making a subtler point that, even if you don't hate DevRel work, the most useful kind of work is often unappreciated or devalued by decision makers.

danabramovon ATProto Permissioned Data Proposal Draft

I think so, maybe.

danabramovon ATProto Permissioned Data Proposal Draft

Rendered Markdown: https://github.com/bluesky-social/proposals/blob/permissione...

Note this is an early draft and will likely change, as PR description says.

danabramovon Japanese verb conjugation the simple hard way

>You claim that only 2 groups are taught when 3 get taught.

I skipped the handful of exceptions because they have no rules associated with them, you just have to memorize them. Yes, it would be 3 if I counted exceptions. I am aware that they exist; you can check their list at the end of my post.

>You claim that you are given arbitrary tables to memorize, but it's usually explained to foreigners that you replace the last romaji to i when conjugating the verb into -masu form.

I must have been unlucky. I don't remember what resource I was using but it was primarily teaching with a separate table per suffix. Maybe the pattern was called out but I missed it due to feeling overwhelmed with the tables.

>And based off your comment here the reason behind doing this may be you extrapolating how you first learned it to how people usually learn it.

That's fair — if most people don't stumble here because this is clearly explained, that's good, and it means I've overstated the "usually taught". I still find that I prefer the specific style of explanation where we consider the corresponding vowel a part of the suffix, like -(i)masu or -(a)nai, which is how some linguists view it. That is another part of the motivation for writing the article.

danabramovon Japanese verb conjugation the simple hard way

Again, it's my mistake in terminology. I'll fix it. It has no bearing on the actual conjugation procedure that the article tries to teach.

danabramovon Japanese verb conjugation the simple hard way

I wasn't aware that this community gatekeeps the word "teach" to exclude personal blog posts. I have no desire to participate in this community anymore, but you're welcome to mentally substitute it with some other word.

danabramovon Japanese verb conjugation the simple hard way

Thank you! Appreciate the kindhearted comment.

Frankly, this thread was kind of wild, and it really got to me. It's not my first rodeo on HN, but the amount of wilful misunderstanding, weaponized debate devices, and manipulative grandstanding was off the charts compared to anything I've ever received in response to articles about programming. I'm still not sure what to attribute this to, and it will likely torture me for some more time.

I mention this at the end of the article, but the paper that inspired me is "No consonant-final stems in Japanese verb morphology" (https://doi.org/10.1016/j.lingua.2010.03.026). It uses a different notation but it also considers the suffixes to be -(i)masu, -(a)nai, and so on, and specifies rules for how empty-ish slots concatenate. Maybe you'll be able to understand it better than I did! I found it very interesting but dense.

danabramovon Japanese verb conjugation the simple hard way

Is it “learn kana” or something else? I know kana. I’ve read through both replies once more and I can’t find any other unifying thread.

Did you assume that I don’t know kana because I’m not using kana in the article?

danabramovon Japanese verb conjugation the simple hard way

What is the advice though?

Is it to not write at all about what works for me? Is it to change what I write into the “accepted” approach and away from what I actually wanted to write? Is it to present my writing differently?

What’s the actual thing that you think would make people in this thread happy (or maybe just you for a start). Like outcome-wise. Next time I write, I do… what?

If it’s to use kana in an article like this, it’s kind of like if I painted a picture of a cat and everyone said they would rather see a dog. OK but that’s not what I was painting? I set it as a constraint for myself to not use kana in the article. It would be a completely different article if it used kana, and that article wouldn’t be worth writing to me.

I shared what works for me, not what I’m recommending for everyone else. So what is the advice? Maybe it’s not to write at all? Some secret third thing?

danabramovon Japanese verb conjugation the simple hard way

That seems like the best diagnosis so far for what’s happening in the thread and gives me some peace. I guess it also means I’ll try to keep my future explanations to myself because intersecting with a community with these norms is highly unpleasant. I wouldn’t say I regret writing this post, but this whole discussion killed my joy about sharing further on this topic.

danabramovon Japanese verb conjugation the simple hard way

I don’t know what post you’re referring to. I understand romaji is the character set.

If you’re nitpicking on this sentence:

> I thought it’s acceptable to refer to Hepburn as just “romaji”

I’ll expand it:

> I thought it’s acceptable to refer to Hepburn [-flavored romaji] as just “romaji”

I think it’s clear by context what I was saying.

danabramovon Japanese verb conjugation the simple hard way

I feel like there’s some kind of miscommunication, and I can’t bridge it, which is driving me a little crazy. I genuinely don’t understand why it’s bad that I wrote something for people who don’t want to depend on fluently reading kana for learning conjugation rules. I genuinely do not see the connection between the two — I don’t think there’s anything “simpler” in seeing how む becomes んだ than in how m_ + ita collapses into nda. It’s literally the same thing. There is no added clarity by using kana. I understand you’re “supposed” to learn it first and so on, but it does not actually aid the process of learning the specific things covered in the post. You mention identification of stems from kanji but this is not the thing I am teaching in the post? It’s about the rewrite rules once you know the stem, not stem detection.

The main thing I don’t get though is… I never claimed my post will be useful to “most people”. Where did you take that claim? Why is everyone assigning that claim to me and then refuting it? I repeatedly said many times in the thread that I don’t expect this to be helpful to anyone except people like me — and everyone is arguing that no, you should do it the other way around. Where is this assumption coming from?

danabramovon Japanese verb conjugation the simple hard way

Ah, I see what you mean. For me personally, knowing the shape of the system underneath has a sort of calming effect because I have a sense of the upper bound of how annoying it will be to learn by osmosis. And it's also nice to have a fallback for when I have a cache miss and need to really think through saying something.

danabramovon Japanese verb conjugation the simple hard way

>the explanation here is misleading

Misleading means it leads to a wrong conclusion somewhere. Please demonstrate which wrong conclusion my article is leading to.

>and that's because of the dependence on romaji / transliteration

There is no "dependence" on transliteration per se. I use it as a visual shortcut for moving along the kana row, as I show in the middle of the post using the table. For all the examples we're looking at, the relationship between the mora and the corresponding romaji version is bijective — so there is literally no difference except the notation. I wanted the article to be accessible to someone who's not fluent in kana (and in fact to someone with zero knowledge of Japanese), hence the choice of notation.

>First, there is a clear stem (飲＊), but you just don't know that yet, because you haven't learned kanji

What could kanji possibly have to do with this? We're discussing a thing rooted in phonetics. The verb stem often includes kana in addition to kanji (e.g. 食べ). There's nothing special here about kanji at all.

>To wit: a godan verb [2] shifts from the "u" sound [...] to the "i" sound in the same column of the kana table [...] Then, for casual negative conjugations, you shift to the "a" sound in the same column

Sure, and that's exactly what I'm describing in the article. Why do you need to pretend I'm describing something different? There's even a place in the article where we use the kana table for that. However, since I assume the reader might not want to constantly look at the kana table, I focus on the phonetic intuition. And the phonetic intuition is trivially explained with romaji, which is why I use them.

danabramovon Japanese verb conjugation the simple hard way

Intuitively I suspect that you severely underestimate how many people give up on the language somewhere in the middle of the process of learning this "basic part" in the traditional way. My aim was to show to this group that you can actually understand the entire mechanic in one evening with zero prerequisites.

danabramovon Japanese verb conjugation the simple hard way

> have some ideas about what works for language learners that is not expressed by language learners experts.

This seems like a culture clash. In programming blogging, it is completely fine to write blog posts about the mental model that works for you (as long as it doesn't contradict evidence) and then maybe it works for somebody else. The idea that I need to first get an approval from a commission of Serious Teachers That Verified Which Approaches Work Well For Learners Statistically is laughable to me.

I write for myself and for people like me, period. I do not claim this is useful to anyone but a tiny group. This tiny group is who it's written for. I explicitly say in the article that I struggled and this is what got me unstuck. If it isn't helpful, just close the tab.