3 ms·

> Much of technology is pure fact I think there are actually very few bedrock facts in computing - a program executing as intended on a CPU is fact. Nearly eve

> Much of technology is pure fact I think there are actually very few bedrock facts in computing - a program executing as intended on a CPU is fact. Nearly everything else is culture or opinion, including the programming language that produced the program. When you get to the level of arguments about programming languages, tools, web development, software architectures, etc this is all discussion in the realm of the mostly subjective, with perhaps a side reference to facts when something is actually not achievable. The most wonderful thing about computing is that it’s pure castles in the air, and that if you’re sufficiently motivated you can build the whole castle. The most dreadful thing about computing is that it’s pure castles in the air and you have few real facts to anchor your arguments to.

bawolff 5y ago
> "a program executing as intended on a CPU is fact" processor bugs, power glitching, cosmic rays,...
- II2II 5y ago
  Processor bugs, power glitching, and cosmic rays affecting how the CPU executes a program are still in the domain of facts. Nothing says that facts must lead to deterministic outcomes, though we should have a way to quantify the distribution of outcomes in order to help us distinguish facts from opinions.
  - SantalBlush 5y ago
    The quote includes the phrase "as intended," which is important here.
    - bawolff 5y ago
      Yeah, otherwise the statement in essence would just be, its a fact that whatever happens, happens. While i don't disagree, its not a particularly interesting claim without the "as intended"
retrac 5y ago
And one of the hardest things in software engineering at scale, is organizing and coralling humans and groups of humans, possibly over many years with everyone involved being replaced at some point throughout, to somehow still cooperate enough that the end-product is attainable. Someone like Linus is more of a politician than an engineer, in some ways.
readflaggedcomm 5y ago
The author means that technology is deterministic. Those castles aren't fantasy once the idea of them is implemented: they are behaviors of mechanisms which could be measured, poked, and prodded.
chasil 5y ago
Let me give you a set of basic, specific operational facts that I discerned a few years ago. NFS as a protocol engages in clear text transfers of filesystem data between the client and server. I had some experience with stunnel, and I found a way to push NFS traffic through it, forcing either aes-gcm or (with TLS 1.3), chacha20-poly1305. The method and complaint that I raised was heard by protocol maintainers, and resulted in a draft RFC to embed TLS at the RPC level, extending encryption to all RPC services beyond NFS. I also later found notes of the NFSv4.2 approval suggesting that TLS should be added, but it was not. That fell to me. I don't think that this was subjective. If I google "onc RPC encryption," I find: https://tools.ietf.org/id/draft-ietf-nfsv4-rpc-tls-01.html https://tools.ietf.org/id/draft-ietf-nfsv4-rpc-tls-01.html "Special mention goes to Charles Fisher, author of "Encrypting NFSv4 with Stunnel TLS" . His article inspired the mechanism described in this document." If I google "NFS encryption," I find me: https://www.linuxjournal.com/content/encrypting-nfsv4-stunnel-tls https://www.linuxjournal.com/content/encrypting-nfsv4-stunne... I think that it was worth the effort. It was in the back of my mind for some time, and I probably should have been more motivated to do it, for the impact it had.
eyelovewe 5y ago
Aesthetic revulsion at software bloat is the intuitively derived conclusion of logically derived facticity.
nofunsir 5y ago
> The most wonderful thing about computing is that it’s pure castles in the air, and that if you’re sufficiently motivated you can build the whole castle. The most dreadful thing about computing is that it’s pure castles in the air and you have few real facts to anchor your arguments to. The most harm being done to this industry — a world of a trillion floating castles — is by those pretending their castles are founded, never revealing the Big Secret, all while fervently demanding would-be competition fully justify their most minute castle-building minutia down to asking which grain of sand is the Seed of Truth, upon which anyone would DARE build a castle in the air. One day, Intelligence will build itself a castle, and it will not be in the air. I think I should like to go build a boat, and study something with a more solid foundation, like quantum physics or Jazz.

bawolff 5y ago

> "a program executing as intended on a CPU is fact" processor bugs, power glitching, cosmic rays,...

II2II 5y ago
Processor bugs, power glitching, and cosmic rays affecting how the CPU executes a program are still in the domain of facts. Nothing says that facts must lead to deterministic outcomes, though we should have a way to quantify the distribution of outcomes in order to help us distinguish facts from opinions.
- SantalBlush 5y ago
  The quote includes the phrase "as intended," which is important here.
  - bawolff 5y ago
    Yeah, otherwise the statement in essence would just be, its a fact that whatever happens, happens. While i don't disagree, its not a particularly interesting claim without the "as intended"

retrac 5y ago

And one of the hardest things in software engineering at scale, is organizing and coralling humans and groups of humans, possibly over many years with everyone involved being replaced at some point throughout, to somehow still cooperate enough that the end-product is attainable. Someone like Linus is more of a politician than an engineer, in some ways.

readflaggedcomm 5y ago

The author means that technology is deterministic. Those castles aren't fantasy once the idea of them is implemented: they are behaviors of mechanisms which could be measured, poked, and prodded.

chasil 5y ago

Let me give you a set of basic, specific operational facts that I discerned a few years ago. NFS as a protocol engages in clear text transfers of filesystem data between the client and server. I had some experience with stunnel, and I found a way to push NFS traffic through it, forcing either aes-gcm or (with TLS 1.3), chacha20-poly1305. The method and complaint that I raised was heard by protocol maintainers, and resulted in a draft RFC to embed TLS at the RPC level, extending encryption to all RPC services beyond NFS. I also later found notes of the NFSv4.2 approval suggesting that TLS should be added, but it was not. That fell to me. I don't think that this was subjective. If I google "onc RPC encryption," I find: https://tools.ietf.org/id/draft-ietf-nfsv4-rpc-tls-01.html https://tools.ietf.org/id/draft-ietf-nfsv4-rpc-tls-01.html "Special mention goes to Charles Fisher, author of "Encrypting NFSv4 with Stunnel TLS" . His article inspired the mechanism described in this document." If I google "NFS encryption," I find me: https://www.linuxjournal.com/content/encrypting-nfsv4-stunnel-tls https://www.linuxjournal.com/content/encrypting-nfsv4-stunne... I think that it was worth the effort. It was in the back of my mind for some time, and I probably should have been more motivated to do it, for the impact it had.

eyelovewe 5y ago

Aesthetic revulsion at software bloat is the intuitively derived conclusion of logically derived facticity.

nofunsir 5y ago

> The most wonderful thing about computing is that it’s pure castles in the air, and that if you’re sufficiently motivated you can build the whole castle. The most dreadful thing about computing is that it’s pure castles in the air and you have few real facts to anchor your arguments to. The most harm being done to this industry — a world of a trillion floating castles — is by those pretending their castles are founded, never revealing the Big Secret, all while fervently demanding would-be competition fully justify their most minute castle-building minutia down to asking which grain of sand is the Seed of Truth, upon which anyone would DARE build a castle in the air. One day, Intelligence will build itself a castle, and it will not be in the air. I think I should like to go build a boat, and study something with a more solid foundation, like quantum physics or Jazz.