The internet is fine.
I would change the web.
I would remove JS and design browsers to natively run python instead.
Ask HN: You have one shot to redesign the Internet – what do you change?
301–310 of 322 posts
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#302DNS.
Others here seem to be redesigning the entire intarwebs. I'll just pick one thing I might have been able to digest.
DNS is brilliant - but insecure, and centralised. The dependence on registrars was a huge mistake. The competition for names is an unintended consequence; the DNS created artificial scarcity, which resulted in commercial businesses that produce nothing of value.
So something like GNS, I guess. https://tools.ietf.org/id/draft-schanzen-gns-01.html
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#303The most basic problems seem to be: 1. Everything being 'free' by default drives us to ad-supported centralized services. Economics aren't a separable concern. 2. Too few IP addresses. (At least one of the pioneers, I forget which, said he pushed for longer addresses but was overruled. So the technical constraints probably did not force this.) I'm not sure how to fix #1, but here's an approach from the 90s: https://c…
Blockchains right now mostly are trying to do exactly this - bring economics into computation and network space. This is of course a higher level solution, but the one that is quite achievable now. Protocols like NEAR and Internet Computer can be good examples of this. I know this is a holywar topic and many would disagree simply by seeing `blockchain` word used here. But to my mind this exactly the place where this…
Yes, I only left this out because it was over the horizon at the time. I did find these papers inspiring in the 90s, and as with the Digital Silk Road paper I linked above, the ideas could have been thought of earlier (e.g. capability security goes back to the 60s): http://www.erights.org/talks/agoric/index.html
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#304Changes 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…
To be able to make these suggestions means you must have significant experience and knowledge of the internet backbone. Super impressive.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#305Earlier quoted context omitted.
Why am I the only one who likes all my devices in my house having one IP address. It's no one's business if I have 10 computers in here or 1.
There's nothing stopping you from continuing to route everything through one computer, since it's your address space you could even do NAT.
NAT would never have been a thing if IP6 was the starting point
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#306Changes 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…
> Routers will accept a "shut up" request. If A wants to block B, it sends to a router on the path, the router pings A to validate the source, and then blocks traffic from B to A for a few minutes. This also contains DDOS attacks. Routers can forward "shut up" requests to the next router in the path, for further containment.
This is an invitation to silence unwanted users by certain governments.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#307Honestly, just one thing, and I only need to go back to the early 90s. Just before the Internet was opened to commercial use in the mid 90s, would've made a perpetual prohibition of advertising over the Internet. Ads are what have ruined everything. Take just about anything unpleasant about the Internet today and it is either directly a consequence of ads, or an indirect consequence of someone trying hard to make you…
Ad's got us here to begin with. Without ads the internet would have been only for the rich. There would also have been no google, no amazon, no android, no kindles. I'd still be ordering everything from the Sears Catalogue, waiting 6-8 weeks for it to arrive.
It'd be kind of cool to have a search engine only for pages without advertising. Kind of like Wiby, but specifically geared for that.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#308Changes 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…
> Routers will accept a "shut up" request. If A wants to block B, it sends to a router on the path, the router pings A to validate the source, and then blocks traffic from B to A for a few minutes. This also contains DDOS attacks. Routers can forward "shut up" requests to the next router in the path, for further containment. This is an invitation to silence unwanted users by certain governments.
You might have missed the "from B to A" in the parent comment which is quite important.
Moreover governments have other (legal) ways to silence users.
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#309Here are a few things off the top of my head (and trying not to duplicate too many things): - Get rid of ARP - just append the LAN address to the network address like other networks. By default LAN addresses are random. (Note IPv6 enables this basically.) - Support encrypted DNS and authenticated BGP. - Let DNS return other metadata including the port as well as the IP address. - Let DNS caching work. Don't misuse sh…
Biggest thing I'd change: Build in the ability to do long-duration keepalives from the git-go to avoid three-way handshakes when you want to send data later after establishing the connection.
A whole bunch of other changes to reduce latency in general - no one really quite realized how important latency was back when this stuff was designed, but in all fairness, it's almost, "how could they?"
Lastly, extend DNS to provide not only encryption and non-repudiation, bu most importantly, more info, or just absorb Project Athena's Kerberos and Hesiod into DNS at first opportunity. This also allows AFS or other global distributed filesystem support, which in turn could have changed database architecture and semantics for the better. Imagine how different the net would be with a scalable global federated name and directory service that could do everything that Yellow Pages did in the 90s.
(N.B.: When I was at Chevron, we built our own version of YP that was hierarchical and multi-domain. It worked really well ("really well", as in damn near flawlessly from the Monday morning we turned it on after a coffee-fuelled weekend hacking session by one of the three true geniuses I've ever met!), seamlessly syncing the info that was in Hesiod/Athena, YP, and DNS to make as much of it as made sense available to clients of each. I've never seen anyone else ever do anything like that, even in all my consulting exposure to other big companies' network services architectures.)
Re: Ask HN: You have one shot to redesign the Internet – what do you change?
#310Here are a few things off the top of my head (and trying not to duplicate too many things): - Get rid of ARP - just append the LAN address to the network address like other networks. By default LAN addresses are random. (Note IPv6 enables this basically.) - Support encrypted DNS and authenticated BGP. - Let DNS return other metadata including the port as well as the IP address. - Let DNS caching work. Don't misuse sh…
Biggest thing I'd change: Build in the ability to do long-duration keepalives from the git-go to avoid three-way handshakes when you want to send data later after establishing the connection. A whole bunch of other changes to reduce latency in general - no one really quite realized how important latency was back when this stuff was designed, but in all fairness, it's almost, "how could they?" Lastly, extend DNS to pr…
Although DEC's LAVC (Local Area VAX Cluster) protocol had shown how important low latency was to those that were paying attention...