4 ms·
> adopt and supporting open interoperability standards Can you imagine not being able to send a UPS packet to a DHL address or not being able to call an AT&T
by Hokusai 6y ago
> adopt and supporting open interoperability standards
Can you imagine not being able to send a UPS packet to a DHL address or not being able to call an AT&T phone from a Verizone one? And how long until I cannot send mails from my Gmail address to you Outlook one?
We live in the Wild West of the digital age.
I still remember the times that Microsoft was trying to own the WWW thru non-compatible Internet Explorer extensions. And, also, the time that Microsoft tried to own Java by adding Microsoft extensions to it. (https://www.cnet.com/news/sun-microsoft-settle-java-suit/ https://www.cnet.com/news/sun-microsoft-settle-java-suit/)
Companies have many incentives to create lock-in software and proprietary protocols that go against the interest of their own users.
- sascha_sl 6y agoOne of them being complexity and and how fast you can move on new features. Adding features to a spec'd standard is a horribly slow process, this is the reason the developer of Signal decided against decentralization. You end up building XMPP, and that doesn't go so well if your specs are not very explicit. You can see this in how unreliable OMEMO is with multiple resources connected. The Web is already a mature platform and has not actually made things more interoperable. It replaced a lot of the previous internet infrastructure that was (IRC, Newsgroups, SMTP which still somewhat clings to life)
- Arathorn 6y ago> Adding features to a spec'd standard is a horribly slow process It doesn't have to be. The Web evolves relatively rapidly these days, and Matrix does too (although it could be faster, for sure). It's important to support freedom as well as privacy - as per https://matrix.org/blog/2020/01/02/on-privacy-versus-freedom https://matrix.org/blog/2020/01/02/on-privacy-versus-freedom.
- syshum 6y ago>> Adding features to a spec'd standard is a horribly slow process Only if your not one of the Big Tech companies.... If your Google you just say "We are doing this, you can either add it to the spec or not, but we are doing it" Then it magically gets added the spec very quickly
- sascha_sl 6y agoor it turns into signed http requests where 5+ different partically-incompatible internet drafts exist!
- pjc50 6y ago> Can you imagine not being able to send a UPS packet to a DHL address or not being able to call an AT&T phone from a Verizone one? For much of the 20th century there was exactly one phone company in the US, and it insisted on total control of all equipment connected to it. Even non-electrically connected mouthpieces. https://en.wikipedia.org/wiki/Hush-A-Phone_Corp._v._United_States https://en.wikipedia.org/wiki/Hush-A-Phone_Corp._v._United_S...
- mikro2nd 6y agoCan you imagine not being able to call a German phone number from the US? AT&T's telephones still had to interoperate with (e.g.) the British Post Office's phones (and everybody else's) to remain relevant/useful. AT&T was very far from controlling all telephony equipment in the whole world.
- O_H_E 6y ago> And how long until I cannot send mails from my Gmail address to you Outlook one? FWIW, a few of my friends at an Elite engineering undergrad program didn't think that was possible. I was baffled.
- addicted 6y agoOne of the biggest reasons people have stopped hosting their own mail servers is precisely because there is genuine worry that GMail may simply reject mails from your server.
- fakedang 6y agoMy thesis advisor falls into this group of people. He typically uses a custom server, instead of the Institute provided email and half his emails go to my spam when we occasionally communicate. Thanks to my thesis days' experience, I've begun checking my GMail once a month, and thoroughly, including spam.