Notices by Raven667 (raven667@hachyderm.io)

@inthehands im assuming some of the negative comments are inauthentic, or from the dangerously stupid, but the only reason this incident happened at all is because the volunteer resistance effort, the only reason there are multiple videos are from other volunteers, this whole incident is a sign of serious resistance, not just showy protests that the authorities ignore

@thomasfuchs i appreciate that they still maintain the update infrastucture, it might just be some static files at a hardcoded url, but many orgs cant manage that

@skinnylatte I feel like a lot of those things are distributed nationwide, or at least the Costco in my college-town in Wisconsin also carries Chinese New Year stuff. It could be a per-store thing because we do have thousands of Chinese nationals as students and professors (less now due to the whole *gestures widely* situation) which might be enough to trigger locally-relevant stocking patterns, I'm not sure how fancy Costco is with their supply chain management, but they are known to be a competant retailer.

@skinnylatte That is the most American business-brained nonsense I've ever heard of, 15-minute handjobs are probably a normal part of sex work anywhere in the world, but _incorporating_ and offering "training" and "retreats" like any other corporate consultancy is a unique take on the concept for sure. Way to SF business culture to take something normal (if a bit controversial/risque) and make it truely weird.

@dalias maybe you are defining a browser differently as well, for a new from-scratch engine to replace Firefox and Chromium its going to have to do substantially everything that a modern browser does, you dont get to redefine the problem as just document management, all the webapps have to work too, including the features you personally dont agree with or think are stupid, or people arent going to switch. You can differentiate on UI, privacy/security, professional features for researchers, but you cant drop widevine or break webrtc or whatever because potential users arent going to put up with Netflix or Teams not working, even if those companies are hostile to your effort and actively trying to undermine you.

Its not hopeless or anything, Servo exists and is making slow steady progress, but its unlikely to be as influential as FF 10+ years ago

@JessTheUnstill @yannsionneau @doctormo @daniel_bohrer

@dalias maybe but you seem to be describing a simplified and idealized version of reality than what exists. I agree that UI redesigns are more often about some anonymous dev working for a promotion rather than the needs of users, and why the minimal resource needs of Firefox or Chromium are less than their current budget, but i dont think it is as much less as you seem to.

@JessTheUnstill @yannsionneau @doctormo @daniel_bohrer

@dalias @JessTheUnstill @yannsionneau @doctormo @daniel_bohrer maybe thats the confusion, im not talking about having a single undifferentiated team of 50-500 people, the communication overhead to coordinate the work would be impossible, but i fundamentally dont believe that a single team of devs could spin all the plates which keep a browser working and add new UI features, implement new web standards, refactor code for changes in the OS platfoms, improve perrformance and sandboxing and be competant to fix bugs across the entirety of tech which makes up a browser. That team size is enough to maintain a branch of a fork and keep it building but not much more. Imho.

I mean, we are just chatting about our opinions here and what we think is more likely, there are no stakes here, i dont need to convince you, you dont need to convince me, we are just sharing and making sure our opinions are understood, even if they differ. I think we are probably close to having that understanding.

I do also think that the major browser devs being large companies (since Opera/Presto is no longer with us) that the standards committies have a bias toward complex problems with complex solutions that only the members have the resources to implement, a side effect that is not in their interest to alleviate.

@dalias ive read Brooks, a long time ago for sure, but while the ideal team size might be around 10, software projects are routinely larger than that and were even in Brooks day, which is why they are broken into team-sized components which integrate according to Conways law.

I feel like i must be missing something because this communication doesnt quite make sense to me, like we are talking past each other in some way

@JessTheUnstill @yannsionneau @doctormo @daniel_bohrer

@dalias that seems a wildly low estimate of the resources required given the breadth of the web ecosystem in active use. I must be misunderstanding but i would expect to maintain a browser needs closer to 100-500FTE, not 10-50, that lower number would be enough to keep a small fork compiling but thats about it. The browser must be the most complex software project ever attempted by humankind, it is an entire operating system environment in its own right. Each component can use 10-50 people, JavaScript language, JIT and library/API development, Vulkan/OpenGL graphics engine, DOM managment and CSS selectors with quirks along with the CSS rendering pipeline in the graphics subsystem, all the different local APIs and underlying tooling, like SQLite for local data storage, and im only scratching the surface of things that come to mind. Each of those could use a team of experts to maintain it, its not physically possible for one human to be an expert in all the underlying tech that builds a browser and even 10 world class experts probably couldnt understand it all.

Itd be interesting to hear from the lead devs pf Opera/Presto and MS/Edge (before MS/Chromium) to get a level set on how much effort they perceived it would take. Maybe throw in Apple/WebKit (remember them) although i understand they deliberately under-fund the web because it competes with their App Store, theyd much rather take 30% cut from a mobile app which wraps a website than have mobile websites work to their full potential, removing the justification for a dedicated app.

@JessTheUnstill @yannsionneau @doctormo @daniel_bohrer

@skinnylatte for those that dont have pets, dental work is basically a surgical procedure and requires the pet be under anasthesia, neither of which is cheap or easy

@skinnylatte I had that happen once and it was the weirdest thing. I interviewed at a tech startup in '08 which I didn't know much about, but turns out they scraped facebook and did profiling/data analytics for hire, which is kind of skeevy, but that's not the weird part; after I interviewed there someone sent me an email to my personal account from a yahoo.com address using the same name as the company which read

> You may want to think twice before joining [redacted]. The financial situation of the company and outlook isn't as optimistic as it appears.

I was offered but declined and ended up finding a much better job with a local University a few years later. I just checked and the company seems to have existed until at least 2019 so their warning may not have been factually accurate but it's a super weird red flag if someone at the company had access to my application info and used it to privately warn me off, whatever was going on, I don't want to be a part of. yikes.

@dalias @becomethewaifu @zzt i am not a hardware designer but i think that is one reason, on x86, the provided OS often has motherboard chipset drivers provided by the hardware vendor, which can tweak them for how they _actually_ wired up the components. Even when the chip has a perfectly functional driver for another OS, the driver isnt going to have knowledge of implementation specific quirks unless someone has that specific hardware and the time/tools to make it work. Thats why your business-class or very common, long-lived models work, and more aggressive vendors with a lot of skus and model turnover, only *mostly* work.

And i also think it is accurate to say that the lowest level hardware state is often managed by firmware and not the os kernel software, so is dependant on whether the firmware apis have quirks, as this is also model specific

@dalias @futurebird @iwein it does feel like the advertising is pushing a breakdown of social cohesion that says you are only safe in a Mad Max post apocalyptic world if you have your own terrifying blacked-out death-machine. The isolation fits in with suburban aesthetic where everyone is in cosplay colonial homestead in a hostile wilderness.

@inthehands Christ, that's dark, but fair. The culture that allows people to keep trying to "make number go up" without any accountability for what that would even mean in the context of human society. They are all paperclip maximizers unteathered from any philosophy of humanity, humility or utility.

@thomasfuchs I'm not sure there isn't fundamental design flaws where the Starship just doesn't have enough margin of robustness so that it'll be catastrophic when they put any amount of weight on it or anything in the flight is even a little rougher than expected. If they had a safety margin they wouldn't have so many blow up

@thomasfuchs its extra frustrating because the Falcon9 is the best rocket that has ever been built, they clearly had the engineering talent at one time to leapfrog the entire industry while Boeing went down the tubes, where is the evidence of that now? Starship seems like a clown show, it seems Musk scares away all the talent now that he's gone fully off the deep end, plus an inability to be flexible and recalculate when making a bad decision

@inthehands I don't have enough mastery of estimation to make a fair comparison of the cost of build vs. buy. Especially now, there are enough high-quality libraries and toolkits that are easy to license (FOSS) that the cost to build what you want really isn't that high, and you can simplify a design greatly if it doesn't have to be highly abstract to cater to a wide audience but instead only has to do one thing in one way.

I think what's really important is to always have a cadre of developers on hand who are capable of building out custom services as needed, even if you do just buy something, they will be needed to do integration and reporting work, but having the capability means you are able to take advantage when there is an opportunity to do something cheaply in-house instead of letting a vendor gouge you. Esp when _you_ know how much work it takes to build the thing they are selling you, and are not impressed.

@inthehands One thing about Universities is that they often share information and resources with each other in ways that private businesses do not, or at least they do in my limited experience, so one school building a tool they can license to other schools for a nominal fee, or each pony up an FTE for development/maintenance, is a reasonable approach to compare with buying something from a proprietary commercial vendor. Internet2 hosts a bunch of infrastructure projects that have this model, but not many of them are direct end-user edtech services.

@dalias @JessTheUnstill I think some politicians want to *appear* to be stronger than they are, and they are afraid to confront a problem like insubordination to democratic rule from the forces head on, even to name it what it is, because they bring attention to their weakness, they prefer to work on problems where the political upside is more clear and there are fewer risks, although plenty are happy to work on nothing at all.

@dalias @JessTheUnstill they pretty much openly threatened DeBlasio's daughter when there were mass protests in 2020 against police violence. The NYPD Union official Twitter account sent out a message to the effect of "we know where your daughter lives" that was a pretty open threat.