Ted Nelson on Zigzag data structures

[@aoeu256]

Yes graphs are a generalization of a trees like files, linked regular tables. However, we need a good query language to arrange all of the views.

[@benjaminscherrey2479]

The earlier database models before Codd's relational model, CODASYL and network model databases, were all navigational models implemented using pointers. This talk doesn't describe how a "zigzag" structure is modeled any different than that. Normalized data structures standardized by the relational model allow us to build tools that allow us to more generally reason about data without having to know all the details about how it's implemented behind the scenes - in theory. Rarely do most people model their data in a manner that allows this to happen practically but it does work. The biggest difference, from a practical term, is that adding new relationships in a navigational model db is a very expensive operation whereas a relational model db allows one to construct views on the fly of one's data without creating any new data to support this relationship. I'd be curious if zigzag structures are supposed to allow for this as that would be the one that that differentiates them from what the original databases from the 1970's were doing all along (and today's "No-SQL" databases). I kinda doubt it.

[@Russtopia]

One would need to design a whole new OS, not just a filesystem. The idea of 'one doc == one file' and 'folders contain folders' hierarchy is baked into the entire UNIX/POSIX/WIN32 API family, which then is brought out into the C standard library, then used to implement C#, Java, Python, ... class libraries. Just like zero-terminated strings, which have been a HUGE security nightmare for decades, but are baked right into base-level OS kernel APIs of all current system. It's a base assumption.

[@GGShinobi77]

He surely has a point - why restrict ourselves to not being able to store information as it really is? Too often you have to use tricks / workarounds if some data doesn't fit into the data structure that we use. I like his idea. But hey, perhaps that's because he is a jedi and used a mindtrick on me!!! :)

[@otakucode]

The really vicious part is when people are made to make their life regular to fit the software or other type of system. People just take it for granted. When someone says 'that's our policy' they imagine that somehow that is a reason. If someone creates a system or policy and someone else falls outside it, the system is always at fault, never the person. But we get it the wrong way so very often, and people get sick joy from imposing arbitrary rules on people. A dangerous flaw in human nature.

[@otakucode]

I imagine representing it as a tagged graph would be the most straightforward way to define it. I'm curious if we've got the algorithms for indexing and making such things fast at scale...

[@notnotjake]

Fascinating but seems hard to use

[@heikohaller]

“We should not impose regularity where it does not exist.” Very nice quote (at 3:08).

[@aaronsloman8406]

Thanks for this. I wonder if you are aware of Common Lisp and its related, but different, cousin language Pop-11, both of which allow not only regular matrices (of any dimensionality) but also linked list structures where a link contains some information and a pointer to another link, a combination that allows enormous variety of structures, especially if some lists are constructed with empty fields that can later be filled with pointers to lists constructed later (which may point back to earlier lists). This allows information structures to be arbitrarily tangled (like the world we live in). Pop-11 also allows some lists (or parts of lists) to be 'dynamically' created: i.e. some of the contents don't exist in the data-structure until something inserts the contents later, for instance allowing newly created or discovered items to be inserted in the appropriate locations in an existing structure, a process triggered by the event of being interrogated. However none of this seems to capture accurately the information about continuously (i.e. not discretely) changing structures and processes that humans (and presumably some other intelligent species) encounter or create in their environments. I am not claiming that only CL and Pop-11 provide such facilities. Of course, the flexibility of linked list-structures introduces both time and space costs -- but those matter less as implementation machinery becomes cheaper, more compact physically, faster and more power-efficient.

[@otakucode]

I wonder what he thinks about NOSQL databases, key-value stores, etc?

[@zarjesve2]

is it possible to obtain this software somewhere?

[@LiweiChou]

Just like graph database?

[@TheThorns]

Why are we not using this!

[@RogerBarraud]

Because you're a n00b?

[@cyborgtroy]

hey, it's a graph.

[@photonhunter]

@zarjesve2 I've put a link to the page with more details in the video description. You'll find a download link there.

[@joyview1]

is it copyrighted? can i use it for my project idea?

[@meltcollective5249]

copyright is open now!

[@meltcollective5249]

can I collab wiht you on a project?

[@GhostGuy764]

if i'm understanding you and mr. nelson it seems that the problem is with our field of view on the universe. As 3rd dimensional beings we see things in a 2nd dimensional state. We can move and see a different version of the 2D state we saw them a second ago but it still is just 2D to us. The data structure being discussed here is ∞ dimensional which is much harder for us to understand. Though I'm new so the point is likely going over my head sorry.

[@MikySan]

When he speak, ther is nothing to do that try to understand the so simple concept and ask how many other years we have to wait for the full adoption of the Nelson theory.

[@frankgreco]

How long did it take for humans to grasp the earth revolves around the sun?

[@zarjesve2]

@photonhunter thank you very much!!!

[@GGShinobi77]

Agreed!

[@joyview1]

Amazing

[@ikethekike]

Surname: von Nassau-Dietz-Orange Given Names: Jan Willem Prefix: Prince of Orange

[@boltar2003]

Why do the lights get brighter when he takes his glasses off at 1.53. Is he really Jesus revealing himself and who has come back to earth to save us via new data structures and better office applications? :)

[@Magnetohydrodynamics]

What. This is some pointer spaghetti.

[@zarjesve2]

show this to really young kids!!! :D

[@bloodaxe5028]

Wait why is he mentioning 4chan at the start of the video

[@Rebassed]

He is saying fortran you dumb ass

[@player1111ful]

why can't people correct people without insults, numbnut.

[@danielestepar.5235]

it's great work o.o

[@zarjesve2]

but they (kids) does not have presumptions ! maybe this zigzag would be natural to them.

[@Exiro]

He keeps on talking about the curse of hierarchies. So why isn't he working on a replacement for the hierarchical file system that's used in every OS?

[@frankgreco]

He has been. For decades.

[@andybaldman]

lol

[@RonJohn63]

There's a *reason* why "techies" impose this order, and it's not because we're some sort matrix-loving fascists: 1) the physical reality of hardware, 2) performance. Traversing through a genealogical tree with a few thousand entries is one thing, but performance tanks when scaled up. This was especially true back when h/w was so limited in speed and capacity. Even now, many of the problems are still the same, since while h/w is much faster, the data set sizes are so much bigger.

[@frankgreco]

Then maybe the hardware is wrong.

[@cebruthius]

@@frankgreco Pray tell, what alternative do you propose?

[@frankgreco]

@@cebruthius Watch some Alan Kay vids along with Ted Nelson vids and you'll get an idea. Our solutions are gated by our experience with conventional hardware. We thought separated data and code was the right thing to do... and we did that for decades with assembly, FORTRAN, COBOL, etc. Simula was a different way to look at software (management) but it was ignored for 30 years because it was "conventional". Multi-cores were thought to be THE answer for at least 15 years until dark cores made us think differently. Instead of using just optimizing conventional data structures, ML researchers are eliminating all data structures and using pattern matching as a foundation. I'm not smart enough to propose a holistic solution to next-gen hardware alternatives, but I have studied technology forecasting and innovation continuities. They have occurred since the dawn of science. My point (from 2 yrs ago) was that RonJohn63's comments were based on the existing computing world and did not consider other views of 'reality'.

[@GhostGuy764]

they'd get it too easy lol.

[@mrbinky7]

Nerds