Hacker News Reader

Comment by skissane | Hacker News Reader

> First non-U.S. access of any kind to the "internet" was in 1973, though physical access continued to remain heavily restricted for years.

One could argue that the current "Internet" based on TCP/IP has only existed since 1983, and the 1969-1982 ARPANET was a different internet given it was based on a different and incompatible protocol stack (NCP).

"An internet" is just any network which connects networks. In that sense, there have been (and may yet be) other internets running other protocol stacks. "The Internet" is a particular internet running a particular protocol suite (TCP/IP), and defined as such the Internet began operation on 1983-01-01.

View on HN

But in 1983, the entirety of the internet switched from NCP to TCP/IP. Furthermore, the TCP/IP as implemented in 1983 was very different from how it is implemented today. At some point you devolve into a "ship of Theseus" argument, so in some ways the internet is 50 years old, in others it is 40ish years old in others it is 30ish years old (for many "internet" and "world wide web" are synonymous).

If walled-garden Apps ever, become synonymous with the internet, then the internet was born in 2007 (or '08 depending on where you count from).

> TCP/IP as implemented in 1983 was very different from how it is implemented today

The gap between 1983 TCP/IPv4 and 2019 TCP/IPv4 is arguably smaller than the gap between 1982 NCP and 1983 TCP/IPv4.

For example, data flow in an NCP connection is one-directional, meaning that applications wanting bidirectional communication have to use a pair of connections, whereas TCP allows communication in either direction over the same connection.

A 1983 TCP/IPv4 stack probably could successfully communicate with a 2019 TCP/IPv4 stack, although changes in TCP congestion algorithms, MTU discovery, etc would likely cause significant performance issues over long-distance links. TCP/IP and NCP are completely incompatible protocols so TCP/IP-only host could not communicate with an NCP-only host at all.

Good points. A better analogy might be switching an entire LAN from token-ring to ethernet. The LAN is identical, despite the fact that the new network is incompatible with the old.

The nodes in 1982 and 1983 were largely the same, so if you look at the internet as the nodes, then there is continuity, but if you look at the internet as compatible intercommunicating protocols, then there is not.

The human side might help. The point where the actual network of connected hardware actually started to affect human society. All of the key ingredients for what is around today, but not necessarily the same hardware and software - because that has changed dramatically but not the scope of the problem space.

Correct. It's a matter of semantics. Though "50-year-old internet" could only refer to ARPANET.

As an aside, English style guides formerly required "an internet"/"the Internet" distinction, however that has been changing in recent years to lowercase "internet" in both cases.

In school I was marked down for writing "the internet", and now English professors say that is in fact the right way to do it. Not even English is immune to versioning.

That’s because it was seen as a name of a product. Now it is a common idea/service.

Think “couch” vs Ikea.

I thought it was about proper noun vs common noun.

"an internet" is common noun – an abstract concept of which many particular implementations could exist.

"the Internet" is a proper noun – one specific implementation of that concept. Right now, the only one of any significance, but that may not last forever.

If humans someday colonise Mars – TCP/IP is not suitable for interplanetary use. We'll likely end up with two separate TCP/IP internets, one on Earth and one on Mars, with non-IP protocols used to communicate between them. The Earth-based one would be descendant of "the Internet" we are using right now, the Mars-based one would be a new internet.

(By the time we colonise Mars, I wouldn't be surprised if TCP has been wholly replaced by QUIC, so instead of TCP/IP we'd all be using QUIC/UDP/IP.)

>If humans someday colonise Mars – TCP/IP is not suitable for interplanetary use.

https://datatracker.ietf.org/doc/rfc5325/

> TCP/IP is not suitable for interplanetary use.

I'm not sure that's true.

The congestion algorithms/parameters may have to be updated, just like they were when WiFi and cellular came along, but the protocol itself is quite robust.

TCP is unsuitable for interplanetary links, due to the massive round trip times (e.g. Earth-Mars RTT is between 3 minutes and 22 minutes.) Technically TCP could work, but the performance in practice would be atrocious. Wifi and cellular aren't comparable because (when working) they don't have RTTs measured in minutes.

That's why for interplanetary use IETF has been developing an alternative protocol suite called DTN. Unlike TCP/IP, DTN is not real-time, it is based on store-and-forward of messages (conceptually similar to UUCP.)

https://datatracker.ietf.org/wg/dtn/documents/

English teachers are generally petulant pedantic crocks anyway. We'd probably have more/better writers without them ruining the subject for people as they do.

It's interesting that you say this. I find that I had about as many good/great English teachers as I did in other subjects, but there weren't many mediocre English teachers.

I had a math teacher who literally never got up from his desk, visibly counting down the years until he could collect his pension. I had similarly checked out history and science teachers.

The not-good English teachers I had, however, seemed mad at the world and out for blood. The name written on an assignment had a stronger correlation with the grade than the content, so the only path to an A was to stroke the teacher's ego

I even had a pair of friends do a blinded experiment to confirm this. They each would type out their assignment, then would call each other and flip a coin; Heads they write their own name, Tails they write the other person's name.

In another occurrence, I had gotten on a teacher's bad-side and so the grades on my papers had a hard ceiling of a C. I asked my dad for advice, and it was roughly: Talk to the teacher, apologize for "not applying myself" offer to redo back assignments. She didn't even have me redo any back assignments, but suddenly I was getting As and Bs again.

Another occurance: I got my first paper of the class back with the only markings on it "58, almost an A paper." Here was the dialog that ensued:

----

Me: How many points is this out of?

Teacher: 100

Me: Is the class graded on a curve?

Teacher: No, straight 90/80/70

Me: So how is 58 "almost an A paper?"

Teacher: Well, just a few tweaks and it would be great,[encouraging tone] I'm sure you'll do well in my course!

Me: ...

----

To this day I still have not figured out what was going on in that teacher's mind. It was a rare real-world moment where I honestly couldn't tell if I was being trolled or not.

You probably needed to stroke her, his, or it’s ego again.

I had a math professor do that to me in undergrad. Fear of his type led me to leave the field entirely after that degree. Just way too much power over me with little to nothing I could do.

it's = it is. You were looking for "its"

gratuitous correction because of the topic. hth hand.

autocomplete made me do it.

>"The Internet" is a particular internet running a particular protocol suite (TCP/IP)

This is nitpicky, but I think a lot of traffic considered part of 'The Internet' UDP as well these days.

To nitpick your nitpick, one could argue that "TCP/IP" is the name of a protocol suite, which contains more than just the two protocols that make up its name. UDP and ICMP are obviously also included. Its precise boundaries are unclear, and some may argue that other protocols such IGMP, DCCP, SCTP, DNS, DHCP, ARP, etc, form part of that suite as well.

Reply on news.ycombinator.com