Aha i saw this video in a comment on another HN thread this week: https://news.ycombinator.com/item?id=49791117 Avoiding the babbling-idiot failure in a time-triggered communication system
Did he though? TMK Shannon did specifically not give a way to deal with noisy channels, no. He invented a way to quantify what could be sent on a noisy channel if you figure out the optimal way to do it. This is very similar to "no matter what you figure out about faster travel, you can't go faster than light in a vacuum".
Calculating the limit is easy with Shannon's theorem. Approaching it in practice is hard.
> Calculating the limit is easy with Shannon's theorem. Approaching it in practice is hard.
Many modern modulations already operate basically on Shannon's limit for a given band/SNR. Well, on raw data, the encoding almost always use some kind of error correction so the decoded bitrate is few % lower than the wire one
Alvin Lucier wrote other highly interesting, if unconventional works, although similar to John Cage's 4'33", this is the one that achieved meme status as conceptual art.
The way I've understood it is Shannon showed that it's possible to reliably send a signal even if you know the signal will get corrupted before arriving at the other end.
Shannon's theory is foundational to modern communication systems. You can expect to have WiFi remain functional even as other devices use the same medium. Up to a point, the Shannon limit in other words.
The full talk seems to be at https://au.cloud.panopto.eu/Panopto/Pages/Viewer.aspx?id=fe0..., but with a confusing interface (text chat flying by at speed, and no actual video of Alan, though you do get a prolonged view of a whiteboard). Maybe there's another copy that exists or will show up somewhere?
Yes it's a pretty klunky interface. Easy to miss that there are multiple camera icons to the right of the time slider, and you can switch to see different feeds like the speaker or their slides. It's still kind of hit or miss with some of the speakers, but at least Alan got a chance to test things out and debug all the glitches before his talk, so he didn't end up repeatedly interrupting himself!
This place being well-known now for the prohibition on "shallow dismissals", I think it'd be great if we brought back the rule against submissions that are merely "shallowly interesting" (which is frankly something that this, er... "performance"* doesn't even qualify as):
* hard to read this characterization and see it as anything other than the product of unvarnished, I-want-to-believe (that emperor is wearing magnificent robes) hero worship
It would have been kind of great to see an OOP language with Alan's design, but with a better syntax. I dislike smalltalk syntax. Barely anyone uses smalltalk today.
Even then, we'd also have an OOP language to be really really fast. Otherwise people will just use C.
Java itself is too verbose and has a rather boring OOP model.
Aha i saw this video in a comment on another HN thread this week: https://news.ycombinator.com/item?id=49791117 Avoiding the babbling-idiot failure in a time-triggered communication system
Did he though? TMK Shannon did specifically not give a way to deal with noisy channels, no. He invented a way to quantify what could be sent on a noisy channel if you figure out the optimal way to do it. This is very similar to "no matter what you figure out about faster travel, you can't go faster than light in a vacuum".
Calculating the limit is easy with Shannon's theorem. Approaching it in practice is hard.
Fair distinction
> Calculating the limit is easy with Shannon's theorem. Approaching it in practice is hard.
Many modern modulations already operate basically on Shannon's limit for a given band/SNR. Well, on raw data, the encoding almost always use some kind of error correction so the decoded bitrate is few % lower than the wire one
Alvin Lucier wrote other highly interesting, if unconventional works, although similar to John Cage's 4'33", this is the one that achieved meme status as conceptual art.
Can someone please explain to me what this is all about?
The way I've understood it is Shannon showed that it's possible to reliably send a signal even if you know the signal will get corrupted before arriving at the other end.
Shannon's theory is foundational to modern communication systems. You can expect to have WiFi remain functional even as other devices use the same medium. Up to a point, the Shannon limit in other words.
I've added an explanation to the toptext which is taken from a comment by DonHopkins, who recorded (but did not submit) the video: https://news.ycombinator.com/item?id=49622607.
Hopefully Don won't mind!
The full talk seems to be at https://au.cloud.panopto.eu/Panopto/Pages/Viewer.aspx?id=fe0..., but with a confusing interface (text chat flying by at speed, and no actual video of Alan, though you do get a prolonged view of a whiteboard). Maybe there's another copy that exists or will show up somewhere?
Yes it's a pretty klunky interface. Easy to miss that there are multiple camera icons to the right of the time slider, and you can switch to see different feeds like the speaker or their slides. It's still kind of hit or miss with some of the speakers, but at least Alan got a chance to test things out and debug all the glitches before his talk, so he didn't end up repeatedly interrupting himself!
I can't wait for the autotune remix of this.
This place being well-known now for the prohibition on "shallow dismissals", I think it'd be great if we brought back the rule against submissions that are merely "shallowly interesting" (which is frankly something that this, er... "performance"* doesn't even qualify as):
<https://hn.algolia.com/?q=%22shallowly+interesting%22>
* hard to read this characterization and see it as anything other than the product of unvarnished, I-want-to-believe (that emperor is wearing magnificent robes) hero worship
It would have been kind of great to see an OOP language with Alan's design, but with a better syntax. I dislike smalltalk syntax. Barely anyone uses smalltalk today.
Even then, we'd also have an OOP language to be really really fast. Otherwise people will just use C.
Java itself is too verbose and has a rather boring OOP model.
How about Elixir and Erlang? They do pretty much exactly what Kay preaches.
Interestingly, the OO model that Wirth and Gutknecht implemented in the Oberon system corresponds quite well to the message-based approach.
Java implemented the Simula 67 object model according to a 2017 Gosling lecture (as did early C++ and Smalltalk-80 to a significant degree).
Crystal is a fast, compiled OOP language. It has static type inference with Ruby-style syntax. Ruby's OOP model took inspiration from Smalltalk.
https://crystal-lang.org/