unsignedint

In general they also don't seem to collaborate much with vtubers outside of their space, so I'm sure there are some rules in play there.

I think it's more about market dynamics than strict rules. There are clear benefits to collaborating within the same agency because it helps introduce audiences to other talents under the same umbrella. Everyone involved has a chance to grow together.

Once you start collaborating outside the agency, you're inevitably splitting viewers' time, attention, and even spending across different groups. Since every fan only has so much time and money to go around, those collaborations don't always make business sense.

That's not to say they never happen. If a collaboration is expected to bring in new audiences, significantly increase exposure, or otherwise benefit both sides, then it's much easier to justify. But without that kind of upside, it can end up helping another agency just as much, or even more, than your own.

I'm a Vtuber myself. I'm an indie, self supported, just for fun type rather than someone trying to make a living from it.

If you're genuinely interested in becoming a Vtuber, the barrier to entry is surprisingly low. You don't necessarily need to commission a custom model or a unique character design. You can use one of the VR avatar generators, which is what my avatar is based on, or even start with a modified VR avatar if its license allows it. That's actually a pretty common approach in the indie scene.

As the article mentions, the market is definitely saturated, so I wouldn't go into it expecting to make a lot of money. For what it's worth, a large number of Vtubers don't even meet YouTube's monetization threshold. Personally, I think that's part of what makes the community interesting. I tend to enjoy following smaller creators more than the big agency names.

One thing I'd add is that I think the article overgeneralizes the AI topic a bit and leans more toward wishful thinking. A lot of the hesitation around AI, especially at major agencies, comes from the fact that their character designs are often commissioned from well known artists. That creates a relationship between the performer and the artist, often referred to as the "Mama" or "Papa" relationship, and there can be a lot of sensitivity around how those designs are used. To be fair, those dynamics can show up in plenty of situations besides AI.

If you don't have those constraints, people tend to be much more relaxed about AI. Some creators who rely heavily on fan art are naturally more opinionated about it than others, but it really varies from person to person.

The way I've come to think about Blender, and to some extent other tools in the same space, is that being a master of the tool as an operator and having the artistic ability to create meaningful work are two different skills. The best artists often have a good balance of both, but plenty of people are much stronger in one area than the other.

Since Blender is free, I think it's attracted a lot of people whose strengths are heavily weighted toward one side or the other.

Going back to the soccer analogy, you might be incredibly good at kicking the ball, but playing competitively and making meaningful plays on the field requires a completely different set of skills.

I'm feeling the same way. People naturally pick up writing habits from what they read, especially if they haven't really developed their own style yet. So it seems only natural that, after spending a lot of time reading AI generated text, some of those patterns start creeping into their own writing.

The less established someone's writing style or vocabulary is, the more susceptible they probably are to that influence. I don't necessarily see it as inherently good or bad. People have always picked up habits from the things they read. AI is just another source of influence, alongside books, blogs, journalists, or other writers. Some influences are great, and some are less so.

Of course, they can't realistically police every use of AI any more than they can police someone using Grammarly or similar tools. I think the core of the argument is really about drawing a hard line against low-quality "write a response for me" type usage.

In reality, people are going to keep doing that no matter what. But at the same time, it's still probably a worthwhile line to draw when it comes to discouraging more irresponsible or disengaged uses of AI.

While things like the expectation of lifetime employment (or at least very long tenure) may sound appealing, it also creates a job market with very low fluidity. In practice, if you miss that narrow “fresh out of school” hiring window, you can end up facing pretty unfavorable prospects later on.

People can still get hired mid-career, of course, but many companies traditionally hire based more on long-term potential than immediately usable skills, since they expect to train employees heavily through OJT. That also means the number of openings for experienced hires can be relatively limited. And because of the seniority-based structure, even experienced workers may end up starting near the bottom anyway.

There was an entire generation of people who missed that initial hiring window because of economic downturns and hiring freezes, and many of them still struggle to land stable permanent positions even today.

Things are gradually changing, but many structural assumptions are still there. For example, parts of the legal and employment system are historically built around the assumption of lifetime employment, which also makes it difficult for companies to dismiss permanent employees once they are hired.

I'd argue there's really no way to make OpenClaw truly safe, no matter what you do. The only place it really makes sense is within trusted environments, like B2B coordination or tightly controlled processes between systems that share the same assumptions.

The moment it steps outside that boundary, you're sending the bot into unpredictable territory. At that point, things can get ambiguous pretty quickly, and in some cases even adversarial.

The article does a good job calling out the more serious offenses, although I’d personally argue that nigiribashi is just as bad as the other two. Most Japanese people would probably react with a bit of shock to those.

That said, chopstick etiquette is definitely evolving. Something like chobujubashi isn’t enforced as strictly anymore, especially with more awareness around left-handed users. Kaeshibashi, on the other hand, is becoming more common, and in some social circles, not doing it can actually come across as rude.

Not only have there been those cancellations, but they’ve also been cutting back features in a lot of areas, especially in the Pro tier, and doing it pretty drastically without any notice. Honestly, I think that might be the bigger issue, particularly since many of the affected users are paying customers, and quite a few of them paid for a full year upfront.

I think the only practical litmus test here is whether you can stand by the text as your own words. It’s not like we have someone looking over commenters’ shoulders as they type.

Ultimately, this comes down to people making a good-faith judgment about how much AI was involved, whether it was just minor grammatical fixes or something more substantial. The reality is that there isn’t really a shared consensus on exactly where that line should be drawn.

I guess this kind of rule feels less pragmatic and more philosophical. For one thing, it’s nearly impossible to enforce in practice, and drawing a clear line between simple grammatical correction and AI-assisted editing is a pretty hard problem.

Just my take: I don’t think “AI” automatically equals “slop.” There’s plenty of human-made slop too, and some AI-assisted content is genuinely useful. I’d rather see this framed as “report low-value/spammy content” than “report AI slop,” since the AI label tends to turn into “this looks AI” witch-hunting. That said, our baseline assumptions seem pretty different here, so we probably won’t fully agree.

I think this really needs to be framed as a "report low-quality content" feature, not a "report AI slop" feature. Otherwise, it just incentivizes people to hide their process, and it risks turning into a witch hunt where everything gets judged on whether it "looks AI" rather than whether it’s actually bad content.

I think part of it is that, in the US (and probably in many other countries too), the roles of maker and consumer tend to be more clearly separated. In contrast, among Japanese users, that line feels much more blurred.

One thing I’ve noticed is that there are a lot of “avatar worlds” where people just go in and pick premade avatars, but these are almost nonexistent—or at least not widely used—by Japanese audiences. The main exception seems to be worlds specifically designed for trying on sample avatars, rather than adopting them as-is.

I provide full livecast coverage by visiting every booth at Vket, which takes place in VRChat twice a year, and over time I’ve noticed some clear patterns.

In Japan, the market for 3D models and other VR/metaverse assets has steadily flourished. Within VRChat, it’s fairly common for users to purchase avatars from platforms like booth.pm and then customize them to their liking—sometimes as simply as changing colors, and other times by adding clothing, accessories, or other elements. The market itself is quite approachable: some avatars are used by thousands, or even tens of thousands, of people, while others cater to much more niche tastes. Either way, there’s something for almost everyone.

Originally, the focus was largely on avatars themselves. Over the years, however, we’ve seen a noticeable shift toward clothing and accessories. Looking at booths in recent Vket events, roughly 40%—if not close to half—of the offerings now fall into those categories. Tools such as ModularAvatar and Mochifitter have made applying and adjusting these items easier than ever, lowering the barrier even further. More broadly, many Japanese users don’t seem to find working with Unity particularly daunting, and that comfort level has helped form the foundation of the ecosystem we see today.

While comedy and roleplay certainly appear from time to time, many people treat their avatars as genuine representations of their identity. This doesn’t mean that identity is fixed—some users switch between multiple avatars—but there is often a strong sense of attachment. The avatar functions not merely as a surrogate in a virtual space, but as something that defines how they present themselves within that world.

This emphasis on originality, combined with a general avoidance of ripped game assets or avatars based on existing IPs (at least compared to trends outside Japan), appears to have played a significant role in shaping this distinctive Japanese VR culture.

I think the real litmus test should be whether the comment adds anything substantive to the conversation. If someone is outsourcing their ideas to AI, that’s a different situation from simply using AI to rephrase or tidy up their own thoughts—so long as they fully understand what they’re posting and stand behind it.

Saying "I asked AI" usually falls into the former category, unless the discussion is specifically about analyzing AI-generated responses.

People already post plenty of non-substantive comments regardless of whether AI is involved, so the focus should be on whether the remark contributes any meaningful value to the discourse, not on the tools used to prepare it.

A typical font contains around 7,000 characters. In everyday use, you rarely touch all of them—most situations stay comfortably within the realm of jōyō kanji. However, there are many edge cases, especially with personal names, where the required characters fall outside the jōyō set. Fonts must be prepared to handle all of these possibilities, including the less common name kanji.

Having known him for decades—not in person, but through various email exchanges when I reached out to BF—I'd say it’s a bit more than “just recognizing the name.” I’ve followed his journey since well before the OSS crowdfunding days, and it’s honestly amazing to see everything he’s built. Thankfully, it sounds like he’s not stepping away completely, which is great news.

As for the new leadership, Francesco Siddi comes from an animation background and is already managing Blender Studio. I’m genuinely glad to see the organization will continue to be led by people who deeply understand the tool and its community.

Product placement ads can be the best kind when they’re done well. The catch is they take far more effort to weave naturally into content, and that limits the kinds of sponsorships you can accept.

The sweet spot is when it feels seamless, but too often creators overdo it and the result is hilariously awkward. Think of someone discussing, say, the dangers of mountain climbing, then suddenly blurting out: “And you know what else is dangerous? An unprotected connection. Which is why you need X VPN!”

“Polish” is subjective. If what a platform provides aligns with your needs, it feels polished. If it doesn’t, that same “polish” can actually work against you. In other words, polish depends on how much you agree with the platform’s way of doing things.

iOS (and Apple overall) tends to be more opinionated. It says, “Do things our way and you’ll have a smooth experience.” Android, by contrast, has historically been more of a flexible “toolkit.” That gave you room to shape the platform to your liking, though it often meant less guidance and structure.

In recent years, Android has shifted toward more out-of-the-box convenience, closing some of that gap. But ultimately, it comes down to what you value: iOS offers a cohesive, guided experience, while Android gives you more leeway to adapt things if you don’t agree with the defaults. Neither approach is inherently better—it’s about what fits you.

The whole “the staff ate it later” routine is really just a symptom of a broader intolerance in Japanese media. After years of getting complaints over the most innocuous things, Japanese TV shows have started slapping disclaimers on everything, even the most trivial situations.

You see it everywhere: statements like “this is just one of many possible hypotheses” to appease people who might disagree, though to be fair, Western media sometimes include similar disclaimers, or “this was filmed with the owner’s permission” even when it is not really necessary. Then there is the excessive blurring—if someone with even a minor scandal appears, they are edited out or blurred, and a message like “this was recorded on MM-DD” pops up, all to avoid viewers asking, “Why is this person on TV?”

Of course, I understand the need for disclaimers in situations that really warrant them, such as scientific experiments that require proper oversight. But the disclaimers added just to dodge silly complaints do nothing but infantilize viewers, and honestly, they are kind of insulting.

Ultimately, this is part of a bigger problem with Japanese TV. It has dumbed itself down to the lowest common denominator, pandering to the most vocal complainers who often lack basic critical thinking skills. This is not unique to TV, either; Japanese businesses in general have long been hypersensitive to the “customer is always right” mindset. Thankfully, there is some pushback against that now. Still, TV is especially vulnerable since broadcasters get access to public airwaves at relatively low cost and are expected to act like a public utility, making them an easy target for complaints.

Ironically, all of this is helping drive younger generations away from TV, not just as a medium, but because the shows themselves feel less and less relevant.

I've been consistent in framing this as a structural critique. The question isn't whether most users inspect metadata. It's whether the system makes that inspection possible. That's the architectural affordance I've emphasized throughout. Email exposes trust signals. PSTN does not. Its opacity isn't a side effect of user behavior. It is a design feature. That's the distinction I've been drawing, and it remains central to the evaluative lens I've laid out.

I’m not claiming universal truth. I’m stating that under the evaluative lens I’ve made explicit, PSTN doesn’t meet the threshold for viability. That’s the scope of my argument, and it stands on its own terms.

You’re welcome to use a different standard, but dismissing mine as "not compelling" without engaging the actual framing isn’t critique. It’s rhetorical displacement. If you’re not addressing the criteria I laid out, you’re not engaging the argument.

Speculation cuts both ways. If you want to challenge the standard itself, do so directly. Otherwise, implying imbalance or extremity without entering the terrain is performative, not substantive.

Even if I don't run my own mail server, I can still open an email header and see context. Last time I checked, I can't walk into a telco and ask for the telephone equivalent of that. On PSTN, there's nothing beyond a bare caller ID string, and that lack of context is exactly why I see it as a problem.

Email gives end users multiple signals and filters to work with. PSTN doesn't, and that's why I disagree with your equivalence.

I get the distinction you’re drawing, but that just brings us back to the same fork: if decades of FCC involvement haven’t produced a trustworthy caller identity system, then the reliance on regulation isn’t solving the structural weakness, it’s just propping it up.

At that point, we’re repeating the same values clash — you see regulation as a workable fix, I see it as evidence of fragility. I don’t think continuing this line is going to get us any further.

I’m not “eager to give up” interoperability — I’m saying interoperability without trust isn’t worth much. Calling it orthogonal doesn’t change the lived reality that the abuse rides on that interoperability at scale.

If the only way to preserve interoperability is to accept decades of unresolved abuse and perpetual patchwork fixes, then that’s not a trade-off I find compelling. At that point we’re not debating facts, we’re debating tolerance levels — and mine is lower. I think that’s a good place to leave it.

Sure, but my point still stands — you have more tools to work with in email. For what it’s worth, I can usually contextualize a message from the subject line and the sender’s address without needing to dive deep into headers. (Phishing is definitely a real problem, but it’s not unique to email in this discussion.)

Now compare that to the PSTN: what does 555-123-4567 really tell you? Not much. It’s just a string of digits with no inherent context. And unlike email, I can’t even choose to outright refuse delivery of a call at the network level.

I only said “good chunk” because there are already more than a few reports pointing to a shift in behavioral patterns around calls. (Feel free to Google, Bing, or DuckDuckGo it when you have a chance.) The point is less about an exact percentage and more about the fact that user behavior has already adapted in ways that erode PSTN’s practical value.

And if the FCC were fully capable of solving this problem, I don’t think we’d still be here after decades of the same issues. That longevity itself is part of my argument: it’s not that people haven’t tried, it’s that the structural limitations resist a clean fix.

Also, IP addresses aren’t a great analogy. They’re not a sole indicator of origination — we have layers of metadata, routing, and reputation systems around them. I’d accept that comparison more readily if phone numbers were spoof-proof. But they aren’t, and that’s yet another area where the FCC hasn’t managed to close the gap.

Yes, absolutely — I’m not suggesting email is anywhere near perfect or free of abuse. But at least collectively, the layers of filtering, reputation systems, and standards make it usable for most people most of the time.

The key difference for me is that the PSTN moves at a snail’s pace, maybe because of regulatory entanglements, maybe because of interoperability constraints. The result is that problems which have been rampant for decades — spoofing, spam, robocalls — remain trivial to exploit. Email has plenty of its own problems here, but at least you get more signals to work with (headers, DKIM/SPF/DMARC, filtering, etc.) than just a string of 10–12 digits with no real context.

That’s why I’m less inclined to “cherish” a system whose shortcomings shift so much burden onto the end user’s well-being, all in the name of interoperability. If interoperability means putting up with abuse at this scale, then that interoperability isn’t worth much — and that’s where my frustration comes from.

I think Chesterton’s fence is a fair analogy—but it only works if the “fence” still serves a protective function. With the PSTN, the fence is riddled with holes, and the people meant to maintain it can’t keep up with the erosion. Interoperability was indeed its greatest strength, but today that same universality is what lets malicious actors reach everyone at trivial cost.

Comparing PSTN to instant messaging walled gardens is interesting, but I’d argue the real parallel is email: a federated, open standard that also suffers from spam and abuse, yet still manages to limp along thanks to heavy filtering and layered trust systems. The PSTN never evolved those trust layers; instead, it relied on scarcity (call cost, geographic constraints) to keep abuse in check. Once those costs collapsed, the trust model collapsed with them.

As for the FCC, sure, they can try to incentivize carriers. But the fact that we need constant regulatory intervention just to keep basic trust afloat suggests the system is no longer structurally sound. Band-aids like STIR/SHAKEN prove the point: we’re bolting authentication onto a protocol that never envisioned it. That might extend its life a little, but it doesn’t make the foundation any less fragile.

So the question isn’t whether the PSTN once had value (it did, massively), but whether preserving it now delivers more value than the cost of propping it up. If a good chunk of people already treat unknown calls as spam until proven otherwise, then the social contract around the PSTN has already been broken.