Hacker News Reader

Comment by jiggy2011 | Hacker News Reader

jiggy2011

The technologies that could be called "the internet" solve more well defined problems than the technologies that could be called "the web".

TCP/IP for example is a description of an electronic locomotive network. It ferries packets around. What is in the packets? Who knows? Who cares? What about privacy, authentication, authenticity? Not our problem!

The web on the other hand is a mishmash of adhoc parts, HTTP,HTTPS,HTML,Websock,DOM etc. Are we a client/server document protocol? Are we a distributed runtime? Do we guarantee secure communication? Perhaps. Are we stateless? Technically, but state tracking is what people use us for! We're shipping around a turing complete language everywhere, is this an integral part of the web? Or just something we slapped on top for good measure?

View on HN

It's interesting to note that choosing not to worry about those things ("privacy, authentication, authenticity") were very much design choices - perhaps the most consequential design choices ever?

...and it's a shame that the WWW is good enough for people to attempt to kludge something on top of it, and continually fail, while not being bad enough for anyone to succeed in launching alternatives that fix those problems.

See also SMTP which has flooded the Internet with huge amounts of garbage traffic but which is simple and works well enough that any alternatives fail.

See also SMTP which has flooded the Internet with huge amounts of garbage traffic but which is simple and works well enough that any alternatives fail.

I agree with your main point, but I wanted to add that there are plenty of alternatives to sending e-mails using SMTP these days. They mostly go by names like "Facebook" or "Google+" and run over protocols like HTTPS talking to a centralised service provider, and they solve the fundamental problem -- letting people send text and picture messages to friends/colleagues for viewing at their own convenience -- in a different and mostly incompatible way.

Decentralisation is quite important for a basic business communication process.

If business switched to facebook as the primary method of communication then facebook becomes critical infrastructure for the economy.

If it suffers downtime then there becomes a massive economic cost, so "move fast and break things" is no longer a reasonable motto.

Of course anyone who needs reliability and privacy will question the use of the specific public social networks I mentioned as examples, but the same principle applies when businesses deploy centralised in-house systems. They still handle messaging (and often a lot more besides) for their key functions using dedicated tools, so general purpose e-mail is no longer necessary for those functions. Increasingly these organisations also use VPNs to allow employees and other associates who are not physically present to hook into the same centralised systems remotely.

Sure, but I think you would be hard pressed to find a company that didn't have some form of email system for dealing without people outside of the company etc even if internal communication is done by some other tool. It's a network effects problem, you could build a better email system but you can't get rid of email completely until it's as commonly used as email.

Such a common tool needs to be highly resilient to catastrophic failure in the same way email is. For example the DDOS on sendgrid affected sendgrid and their customers but the rest of us were unaffected.

The implications of a system that allowed groups like Anonymous to cripple worldwide trade at will via DDOS would be quite scary.

I agree with this too. I'm not suggesting that e-mail is dead or anything, just that a lot of use cases aren't so much being replaced by an alternative to SMTP as being replaced by another communication channel entirely.

"The technologies that could be called "the internet" solve more well defined problems than the technologies that could be called "the web"."

I'm pretty sure the original problem the web was meant to solve was the problem of enabling hyperlinks between documents that are distributed across many servers. The original concept solved this very well, scaled very well, and was then hijacked by people who thought that "document hyperlink system" means "application delivery platform."

(OK, there were a few things that happened that I glossed over, but the web's original goals were not as poorly defined as you make it seem.)

Reply on news.ycombinator.com