Develop stronger cultural and technological safeguards to let people remain anonymous and unmasked, preventing the eventual erosion of privacy.
Ask HN: You have one shot to redesign the Internet – what do you change?
271–280 of 322 posts
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#272Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#273Use IPv6 from the beginning and banish NAT.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#274Encrypted SNI and No Paywalls
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#275Remove the cookie consent popups.
Or, if we're talking about the ground work, specs the cookies such that browsers must implement the cookie consent and therefore sites can't build it in js.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#276One general change and one specific:
- A general idea: Arrange so that sending packets costs way more than receiving, just like snail-mail. (At the moment, a large website or spammer pays very close to $0 per message, individuals pay much more per byte to receive junk.) This would encourage decentralisation, nicely small web pages, and discourage spam.
- And "disappear" XML. It might be "ok" (just) as document markup, but it's a terrible for structured data and config, and for transfers.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#277Changes I would have made in the early days: - 48-bit static IP addresses. 70 trillion should be enough. 128 bits was overkill. - Nodes, not interfaces, have IP addresses, so you can use multiple paths. - IPSEC available but initially optional. - Explicit congestion notification, so packet loss and congestion loss can be distinguished. - Everything on the wire is little-endian, byte oriented, and twos complement. - Y…
> 48-bit static IP addresses. 70 trillion should be enough. 128 bits was overkill The least significant 64 bits in IPv6 are used for link-layer addressing, which is how IPv6 supplants older L2 protocols (ARP etc) with NDP etc, thereby fixing (or at least improving) some significant issues with larger IPv4 subnet scalability/reliability
Right. The reference IPv6 packet header is only twice as long as that for IPv4 (320 bits vs 160).
What is crucial, combined with the fact darkr points out that since they include the link layer the IPv6 protocol deals with more of the network architecture, is that IPv6 headers are much simpler, with only 8 fields, the 6 non-address fields all in the first 64 bits, while IPv4 has 14 fields, with the 12 non-address fields taking 96 bits, and they are better designed with attention to what routers find most important. This simplicity offers routing subsystems increased flexibility in the way they can be put together and handle their workloads.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#278Ban all commercial activity. Make that law loosely defined so it can strike down any company trying to work around it.
A centralized internet will inevitably do more harm than good.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#279Changes I would have made in the early days: - 48-bit static IP addresses. 70 trillion should be enough. 128 bits was overkill. - Nodes, not interfaces, have IP addresses, so you can use multiple paths. - IPSEC available but initially optional. - Explicit congestion notification, so packet loss and congestion loss can be distinguished. - Everything on the wire is little-endian, byte oriented, and twos complement. - Y…
"- Mail is not forwarded."
I experimented with this one over 10 years ago, pre-Wireguard. The idea was to first establish a peer-to-peer connection, e.g., via L2 overlay, then have each peer run their own smptd. A supernode runs on a publicly reachable server at a hosting company and is only necessary for extablishing a peer-to-peer connection, not for routing traffic. Each peer has an Ethernet interface, e.g., a tap device with a private address for the L2 overlay, and each peer runs their own smptd. As for spam, the trick is to limit the size of the overlay network. Users might belong to several L2 networks, e.g., work, home, school, etc. If it networks are kept small and the peers do not give out their email addresses to people not on the network, then it's possible to have a small, spam-free email network among the peers. If end-to-end email on small disparate networks became the norm, spammers would have to find all these small overlay networks and infiltrate every one.
As I understand it, this was the original intended design of email: smtpd's directly communicating with each other. This still happens but users do not run smtpd's. Instead they were asked to run pop clients.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#280Make everything small endian, its a small ask and would make so many lives easier.