I'll try to take a look, but the next 2 weeks are chaotic and hectic, so I may not get the chance.
The way you talk about this makes it sound like you are contemplating a total rewrite, a brand-new code base, developed from scratch. This is almost surely a strategic mistake. There have been any number of articles, analysis, blog posts and erudite studies on software project failures. Complete rewrites almost always fail. The school of hard knocks is a tough teacher. Much better to learn from someone-else's mistakes. When I skimmed Alexey's papers, it seemed that the ideas there could be easily accommodated in the existing AtomSpace (again, I may be mistaken) so I don't really understand the urge to rewrite from scratch. Perhaps Alexey (and others) don't see a path from the existing code base, to one with the desired properties. That doesn't mean there isn't one. Based on past experience, it seems opaque and hidden at first, but then reveals itself. I'm saying: don't go off half-cocked. You did this once, with the GroundedObjectNode debacle. Don't make that mistake again. There was a communications breakdown shortly before that episode started. I'm concerned we're still looking at flawed communications, and that this will lead to poor engineering decisions. (BTW, all the groundwork has been set up to redeem and reclaim the GroundedObjectNode concept, with all the various bugs and infelicities fixed. It can be brought back from the dead, but no one has done this. I assume everyone got bored, or decided they didn't need it any more?) -- Linas On Wed, Aug 4, 2021 at 7:32 PM Ben Goertzel <[email protected]> wrote: > Hi Linas, > > The documents on that "Atomspace" page that Mike linked are not really > the key ones for the Hyperon design, and nor are the 2017 documents > about the DAS so important right now (Senna is in the midst of putting > together something new regarding DAS actually, informed by more recent > investigations...) > > In the happy circumstance you decide to take the time to understand > the crux of what we're aiming for w/ Hyperon, Alexey's various > documents linked from > > https://wiki.opencog.org/w/Hyperon:Atomese > > are important; and then general formal/theoretical context is given by > my 4 papers in the first bullet list in > > https://wiki.opencog.org/w/Hyperon > > (which are significantly inspired by, and refer to, your work on > sheaves and germs in OpenCog .. you can see that instantly from the > diagrams in the "Folding and Unfolding on Metagraphs" paper). > > I understand this is a lot of dense material to go through tho, and > you may not have time... > > We are aiming to have concrete proposals for Atomese2 language > (probably with a different name) and {whatever form of Distributed > Atomspace or Atomspace-distributed-processing-framework we have > concluded is most sensible} by (hopefully) end of September.... Your > feedback on these proposals will be valued at that time, and possibly > instrumental... > > We had hoped for wrapping up this early first-stage-of-design phase in > July rather than September but with various peoples' summer vacations > and a lot of work for Awakening Health it's taking a little longer. > Anyway these are far from trivial matters as u know... > > ben > > On Wed, Aug 4, 2021 at 3:49 PM Linas Vepstas <[email protected]> > wrote: > > > > > > > > On Wed, Aug 4, 2021 at 10:20 AM Ben Goertzel <[email protected]> wrote: > >> > >> > >> So the motive for redesigning/rebuilding is not a feeling > >> that the current Atomspace/OpenCog is bad, but rather that through > >> this work (and prior work) we are seeing that some things that are > >> fairly awkward in the current system > > > > > > This is indeed the correct way to do things. The current system is the > result of using earlier systems, and discovering they were awkward and > ill-suited to the task. > > > >> could be made much > >> cleaner/easier/more-efficient in a variant system with some different > >> characteristics... > > > > > > I would really like to know these. > > > >> (key examples are real-time learning interactions > >> btw external deep learning software and OpenCog; > > > > > > Yikes! This is exactly what the Values subsystem was designed to do > well. I suspect that it is not being used as intended. Probably because it > is mis-understood. I'm very interested in getting this stuff to work right, > because this is important. > > > >> > >> and large-scale > >> logical-inference meta-learning...) > > > > > > No clue what the problem here is. This sounds like a Nil thing. > >> > >> > >> The Hyperon wiki page contains links to a quite a lot of our > >> early-stage thinking on all this as it's been evolvlng over the last > >> year... > > > > > > As mentioned, what is there now seems to be like minor tweaks to the > current system, some of which has already been implemented and is > operational. So I'm perplexed that it's being presented as all-new, > all-improved. > > > > --linas > >> > >> > >> > >> > >> > >> > >> -- > >> > >> -- > >> You received this message because you are subscribed to the Google > Groups "opencog" group. > >> To unsubscribe from this group and stop receiving emails from it, send > an email to [email protected]. > >> To view this discussion on the web visit > https://groups.google.com/d/msgid/opencog/CACYTDBduKz%2BcPHTpmaJvMRJS3Y4y56u6RGrHRzaDVq62Ysy3cA%40mail.gmail.com > . > > > > > > > > -- > > Patrick: Are they laughing at us? > > Sponge Bob: No, Patrick, they are laughing next to us. > > > > > > -- > > You received this message because you are subscribed to the Google > Groups "opencog" group. > > To unsubscribe from this group and stop receiving emails from it, send > an email to [email protected]. > > To view this discussion on the web visit > https://groups.google.com/d/msgid/opencog/CAHrUA36nXLwq6N-MGWd5nJM-UVk_RLjnNQvmejMM1hSACPN_GQ%40mail.gmail.com > . > > > > -- > Ben Goertzel, PhD > http://goertzel.org > > “He not busy being born is busy dying" -- Bob Dylan > > -- > You received this message because you are subscribed to the Google Groups > "opencog" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/opencog/CACYTDBea3_HRLMWsax%2BwSPh5pJ8ZtOsQES5j005KKjJnrKEwtA%40mail.gmail.com > . > -- Patrick: Are they laughing at us? Sponge Bob: No, Patrick, they are laughing next to us. -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/CAHrUA34m3nC1Wk_E1FYFzmx-qWsxbe1BqrE-9fh_brRWUis9jg%40mail.gmail.com.
