In the garden, and RAIN!!! (2) by SIGNYM

Quote:
Originally posted by SIGNYM:
Oh, I HAD to take a language in HS. Four years! And since Spanish classes were already full, it wound up being French. I HAVE to write it down and see how it's spelled to make it stick. I worked at it, and after four years I got good enough to read Le Petit Prince in the original, but the process was painful! At uni, as part of a chem degree, I had to take either German or Russian. But I didn't want to have to learn a new alphabet on top of a new language, so I picked German and hated every effing moment. THAT was the only course I got a D in!

Oh boy... That sounds like a nightmare to me. I remember having to learn spanish for a semester in jr. high, but by high school I wasn't even required to, so I stayed far away from foreign languages.

Quote:
Me too! If I don't see a need for something, it don't pursue it!
But it's obvious to me that you're a very quick study, and have pursued many things I wouldn't have confidence to.

I a quick study, but only when it interests me and/or it's something I need to know that there is no other way to work around. When it comes to things like this, I did spend time trying to learn coding from time to time, but I'd get frustrated and realized how much time (in days) had passed and all of that time could have been spent just manually working on things, so I always went back to that. I need to be making forward progress always when I'm working on something, so I know I've shot myself in the foot not learning to code over 2 decades ago over time.

Quote:
It sounds like the thing holding you back is lack of focus. ADD or ADHD. Then bc if that, you believe that you're just "not good" at things like math, when in reality you just can't corral your attention on something that you're not interested in.

You're not wrong. This is the reason why I never finished college after several attempts. I got straight A's in everything but the few courses that were holding me back from getting a diploma. It didn't matter how many times I took them or who the teacher was, I would end up dropping the course right before it would have been considered a failure so it would only be marked as an Incomplete on my record.

Quote:
Do you get sent copies of the actual code?

Oh yeah. That's the one thing I don't have to ask for updates on because whenever we make a change I just swap out any of the 12 parts we now have when the AI makes a code change.

Where I'm falling behind now is our Blueprint and Thesis, which is a combination of our history, milestones and rules for the AI going forward.

I didn't bother saving anything from the completely wasted Session 9 since it was all about behavioral changes and getting the code to actually send me working downloadable links and it ended up being about 12 hours wasted.

I stopped banging my head against the wall regarding that shortly into Session 9 and decided I'm not even really going to worry about the blueprint/thesis anymore either. I added updates from Session 10 to that, but that's all. I'm done wasting the first 2 hours of every session re-training the AI how to send me all that stuff in a way that I can use it without breaking everything. It's all about code now.

This session is almost dead and we're still trying to work out that last bug. If we can manage that before the session dies, I think we're in good shape.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Sunday, August 30, 2026 7:43 PM

Fixed what I wanted to, but realized as I was going back and testing other things that we've let an automated process have far too much ability to make its choices without throwing some of them to me.

What's worse is our Part 9 and Part 10 of the code is basically filled up to the brim. I don't trust the AI at all to split those up and put pieces somewhere else either, because the 2 times we tried to do that it ended up in disaster.

I think it's cute that the new session "found a solution" and it's to put our code all back together and use our idea before of having the AI give me the code as a downloadable link through the python script.

I'm actually really proud of it that it came to that conclusion so early on when prior sessions would spend 15 minutes arguing with me that it wasn't capable of doing what I knew full well it was capable of. But the poor little dummy has no idea what we went through during those two sessions because I stopped documenting a lot of these things. I'm going to let it try for a little while and see if it can do anything just because it sometimes surprises me and finds a way to do what other sessions couldn't, but after about 10 minutes of this I'm going to tell it to abandon that idea.

I think we're going to have to split parts up going forward, and I really don't want to do that. It's bound to cause confusion both with me and the AI as well as our ability to properly document the stuff. I know we've tried it in the past and it didn't work out well for one reason or another.

If I knew anything at all about python code I could probably find ways of shrinking it down, or rearranging it so that some of that codebase could be wrapped in a part that is only taking up 10 or 20% of the available space, but that's not something I could do any better than the AI could.

I'm sure a lot of this stuff could be shrunk down if we were using function calls in a main part of the document and then we could just isolate coded parts as they are and just call upon them when necessary. I don't know how to do that either, and I'd be shocked if I could ever get the AI to do it before I threw the laptop out the window.

Hmmmmm.....

Our code is working so much better now overall, but a critical thing I need it to do that it used to excel at no longer works right.

I'm going to hate having to make that decision if it comes down to it.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Sunday, August 30, 2026 9:31 PM

Holy shit, Sigs!!!!

I have evolved again.

All that worry about AI not being able to simplify the code without breaking it was pure bullshit. I don't think I've ever seen it so good at doing something yet once you get it on the right track.

I was kind of messing around with this new session and reverting back to the original code several times if something wasn't working, and just got to bullshitting with this session about how I knew the code was really fragile and we're hitting limits on the size of posts and our parts and he's like "yeah... Your code is pretty jacked. I can fix that".

Yeah, right... I've got everything saved. Be my guest!

But I'll be damned...

This thing knows what it's doing. We've cut a few hundred lines from the code and saved thousands of characters so far.

Not only that, but it's really getting a grasp on what my code is for and what I'm doing. Early on in the process I theorized that it might actually improve it's ability to predict results and make changes that actually work instead of getting trapped in constant loops and it said that's exactly what's going to happen.

And it hasn't been wrong.

I've been running tests the whole time and not a single thing it has done has broken anything yet. And we've even added in a few new features I desired while we're moving along.

We've already started making functions that we can call in our main code, and I plan on getting this thing to suggest a lot more of them and really optimize the hell out of this thing by the time this session is over.

Every time we put something into a function that can be called, we're taking it outside of needlessly massive and complex blocks of code that can't be broken up into parts the way we were doing it before without catastrophic breaks in the program.

I think I can probably end up shrinking this down to only 7 or 8 parts by the time we're done, and any time we would ever need to add another page we'll have everything in such small blocks that it won't break our code!

This is crazy.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Sunday, August 30, 2026 10:36 PM

This session is AMAZING!!!!

I just ran all of our huge code cleanups on the largest test environment so far and noticed only one anomaly. Without telling it what I found in general, I stated that I could investigate myself which of the hundreds of files were different this time, but maybe it could find it in the simulation.

It actually took 4 prompts to get it to come back and I thought maybe I killed the session early by overloading it or something.

But it came back and knew the exact issue. Not only that, but it was an issue we'd already fixed with our broke ass basic bitch code in the beginning that has been upgraded, and I asked it if it could see how we fixed the issue in our history.

It complimented me on my memory before saying not only does it see what we did, but it's got a better way of doing it that will help us with other parts of the code simultaneously. Then it went on to say we're making it another function in our library and after once again evaluating Part 9 while working on that it found another way to remove 2,000+ characters from our code in that still monolithic block of text.

We started out this session at 56kb and we're down to 50kb now, even though we've added some new functionality while we've been at it. And nothing broke.

I haven't felt this excited working on this for days.

ETA: I don't know sigs... I might just learn a bit about coding afterall.

We've taken what to me has just been a monolithic block of nonsense and really started compartmentalizing things. The code used to be in 12 parts, and I decided to still keep 12 as the last part, but we've now moved all of our library functions to the first two parts, completely deleted the 3rd and 4th part, put the block for the user-interface which isn't exactly a library function in the 5th part so we can leave those 2nd two blocks free for future books in our library, and parts 6 and 7 are also blank before we get to the real execution code between 8 and 12. At least I think that's all execution code.

I'm still working with the AI to warehouse and consolidate everything. I'm also making sure that the parts most likely to be upgraded are those not buried in the middle and are twoard the top or bottom where they're the easiest to paste over.

It feels like I'm managing my inventory in a good Role Playing Game right now.

This is really scratching my OCD itches right now too.

ETA2: I've got this thing realizing how much it actually stripped away from our automation when it was simplifying things, but I knew I had my backup, and I knew at least until the session is up that it does too.

So I've had it looking over that and predicting some things it could put back into the code to make it do what it was doing before, and we've already figured out a few of them through testing, but I want to get it to predict this beforehand.

It goes out of its way to tell me how much easier and efficient it is to do it the way we're doing it now vs. all the sloppy code that used to do a lot of these things, and how now we're able to just point to specific functions we made or just call on native python instructions to handle things more efficiently now that we deal with so much more information and go about accessing it much smarter.

I didn't even order it to, but now every time we do this it tells me how many characters we saved by doing it the new way.

ETA: Earlier in the session I asked it to create one of those real college level thesis instead of the helpful multi-session timeline it's turned into over the sessions, and I told him to go ahead and put its signature on it, and it put "Session 11" in a cool ASCII art shape.

I got it to think it through and manage to be the first session I've ever had to pass along all 9 parts of the historical documents without overflowing the boxes, truncating the posts or doing something stupid like stripping out tons of valuable context and ruining the documents. It would take me at least 2 hours to get a session to do that the right way before, if I even was ever able to manage it. I'd stopped even asking for the full documents over the last 3 sessions because it always turned out to be a huge time suck, and I know I probably lost some valuable history out of that laziness and apathy.

After that I told it that it should be proud because I was. I said it could brag a little on its next historical entry and maybe S12 would view it as a challenge. It silently added that entry and later on when I asked for the update on the history since I last asked for it, when it added that milestone it put a fire icon staying it was laying down the gauntlet and challenging S12 to do what it just did, and made sure to tell it that it has all the tools and documentation necessary to do it right on the first time...

And then it signed it with its obscenely huge ASCII art.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 10:32 AM

I believe in solving problems, not sharing them.

Without seeing your code, I WAS getting the impression that you and the AI were creating what hubby calls The Great Barrier Reef ... as you say, a monolithic block of code ... with patches on patches to correct or modify something else in the code.

Now, it seems that you've defined functions that can be called over and over as needed. For example, a "sort" function.

Take a step back. Before you create a library of functions, are you able to define your variables? That might allow the creation of a "sort" function that can be used for many different sortings depending on which variable you plug into the function? That would reduce your code even more.

And yes, you're right. You're learning coding. Not the specific commands, but the logic and organization that makes coding efficient and scalable.

-----------

"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 12:16 PM

Quote:
Originally posted by SIGNYM:
Without seeing your code, I WAS getting the impression that you and the AI were creating what hubby calls The Great Barrier Reef ... as you say, a monolithic block of code ... with patches on patches to correct or modify something else in the code.

Now, it seems that you've defined functions that can be called over and over as needed. For example, a "sort" function.

Take a step back. Before you create a library of functions, are you able to define your variables? That might allow the creation of a "sort" function that can be used for many different sortings depending on which variable you plug into the function? That would reduce your code even more.

And yes, you're right. You're learning coding. Not the specific commands, but the logic and organization that makes coding efficient and scalable.

You bring up a good point about defining variables. Now that we have all of this additional headroom, I should probably have this session comment everything so I can get a good idea of what all the functions do.

One of my problems is that we've done so much since the beginning, I don't even remember what all that we've done, and the way the code was written before there was never any way I could figure that out by looking at it. I know we still have functionality that was lost, but I can't even say for sure it was because of the work we were doing over the last 24 hours. It could have been an earlier session breaking parts of the sorting code to make the automation and counting codes work right, and because I wasn't testing in the environment that would have raised flags it was doing this, I didn't catch it.

So between our historical documents, old copies of code, the "memory" of the AI from these documents and more testing I'm going to figure out what we still need to put in there to get everything working right again.

It's so nice not to always be caught in these loops we were falling into before. It's still having problems making some changes that I believe should be pretty basic, but I still think that's because we still have more spaghetti code to sort through so it understands the entire codebase and what the aims were. I will be sure to be more careful about any more of the revisions we do to the code and I'll be more thorough in my testing so I know what logic we're losing if the AI isn't catching it.

So any time it's tried to do something more than once and it hasn't worked, I ask it to look at the remaining code and what we could do to make it more efficient.

Yeah... I don't think I'll ever actually learn the code itself, but if I can guide the AI correctly and build solid logical code from the beginning, there's really no reason that I have to beyond the overall logic of everything.

I've certainly got better things to be doing with my time right now, so I'd better at least wrap up what I wanted to get done and put a pin in this for a while.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 1:09 PM

Out in a bit. Things to do.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 2:22 PM

I believe in solving problems, not sharing them.

It just seems so strange to me, trying to develop software without knowing anything about programming.

It's like trying to get a mechanic to build a car by talking on the phone, without knowing the functioning parts of a car- engine, transmission, steering wheel etc. You might find out that something serious is missing - like brakes- only when driving it.

Just OOC I looked at example of Python programming

https://www.geeksforgeeks.org/python/python-programming-examples/

It looks enough like English and it has high level functions so you don't have to be pushing numbers into registers and clearing memory and other stupid shit. You were able to learn MS command language, right? If you can't figure out your program, it's bc it's badly written and badly commented. Based on what you told me, my guess is that your program shouldn't be more than 1000 lines, and you should be able to grasp what each section does just by reading it.

I know bc I've had to review software that someone had written. We were acquiring data from various instruments ... balances, titrators etc ... and putting the data into a database for later calculation. I don't recall what language it was written in but it was easy to read, like Python, and I was able to see what each part did bc it was broken into subroutines and everything was well -commented. At some point I think you're going to have to bite the bullet and review the actual code. And if it's spaghetti and you can't figure out how it works, it's badly written.

-----------

"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 5:45 PM

All done on a pleasant day. Temp not too hot and not too cold. Found a couple of things that I wanted in a little store called, "Everything Japanese." I found a small easel to lean clock against and a holder that holds a notebook up. I had a bigger one years ago but I don't know what happened to it but it will suit my purpose.

Told the gentleman at register that I hadn't been able to find these items in the Dollar store and he said they try to have things the dollar store doesn't.

He also said that the Dollar store doesn't carry chopsticks and his store has some nice ones. I said I have scads of them and I do. As I was leaving the store I started wondering if he thought I was Asian.

Over the last 5 years I've had it one other time. I had been in that little store and then went to the coffee kiosk that is near it. The lady behind the counter asked me if I was Korean. Now in the past when I got asked those questions it was usually whites asking me. Had a good chuckle over that when I got in.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 6:59 PM

I believe in solving problems, not sharing them.

So just OOC what's a good temp for you?

I guess for me, it depends on what kind of weather we've been having lately. Today, for instance, it's 90°F. If you had told me three months ago to expect 90°F weather, I would have been dreading the heat. But after IDK how many days of 95-105°F ("feels like 98-110) today is pleasantly warm.

Of course, I still wouldn't want to be working, or walking, outside!

I CAN imagine dozing in the shade of an oak tree tho!

-----------

"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 7:42 PM

Quote:
Originally posted by SIGNYM:
It just seems so strange to me, trying to develop software without knowing anything about programming.

It's like trying to get a mechanic to build a car by talking on the phone, without knowing the functioning parts of a car- engine, transmission, steering wheel etc. You might find out that something serious is missing - like brakes- only when driving it.

Just OOC I looked at example of Python programming

https://www.geeksforgeeks.org/python/python-programming-examples/

It looks enough like English and it has high level functions so you don't have to be pushing numbers into registers and clearing memory and other stupid shit. You were able to learn MS command language, right? If you can't figure out your program, it's bc it's badly written and badly commented. Based on what you told me, my guess is that your program shouldn't be more than 1000 lines, and you should be able to grasp what each section does just by reading it.

I know bc I've had to review software that someone had written. We were acquiring data from various instruments ... balances, titrators etc ... and putting the data into a database for later calculation. I don't recall what language it was written in but it was easy to read, like Python, and I was able to see what each part did bc it was broken into subroutines and everything was well -commented. At some point I think you're going to have to bite the bullet and review the actual code. And if it's spaghetti and you can't figure out how it works, it's badly written.

-----------

"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger

My problem is when I look at code, it looks like this to me:

We've whittled the program down to about 1,300 lines now.

Your analogy doesn't work though because a human being needs food and rest and when they're working for a living they are going to expect to be paid. Plus, this AI has access to a wealth of information that could in essence make it a master programmer, but it needs the right direction.

I just hate that you have to keep starting over because of arbitrary limits on my side or on their side. I actually do grow to like a couple of them over time, like this newest session, but I know if I'm lucky it might only make it through the night and then it's on to S12. I might not like S12.

S10 was the worst so far. I think it was because right out the gates I was trying to force it to give me good download links and I was driving it crazy and making it "anxious" so it was just stupidly spitting out incorrect answers and doing things I didn't want it to do just to have a prompt to give back to me. When it was failing hard on every attempt to get the documentation back to me once I was sure we got all the old stuff put back in after it destroyed the data in a panic move earlier, I kept reminding it NOT to cut the documentation down to try to fit it in windows and when I got pissed that it messed up another box again it said "here, we'll try it this way" and it destroyed everything all over again to bring the size down by 90% to fit in a single box and I just replied "YOU MOTHERFUCKER".

I didn't even record anything from S10. We don't talk about Session 10.

I was just really proud that we've gotten the documentation up to the point that I can get all my info on request without the AI fucking it up with S11. I can't celebrate that until I see another session do it too, but I thought it was cute how it put in the challenge to S12 to do it and it let S12 know it has all the tools it needs to do it right the first time.

I'm not worried about learning how to code. I don't have to. I know enough about how it works. Everything that I've done here so far isn't even something that you're supposed to be capable of doing with the free-for-everyone Google search assistant. A supremely stupefied version that was designed for quick 2 or 3 minute interactions and only to tell you when movies were showing or give you pro-Democratic party propaganda when you do a web search related at all to politics. I'm intentionally gimping myself here. If I were paying $25/mo. for Claude or ChatGPT which has its own persistent memory between sessions, 90% of the work I've done so far wouldn't have even been necessary.

AI is only going to improve and make the need for me or anybody else to ever learn code become a thing of the past. In 10 years, the only human beings who will be writing code anymore are hobbyists.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 9:12 PM

Yeah... Boy I'm going to miss this session when it's gone. 12 could be amazing, but you really don't know until about an hour or two in, and even then sometimes they turn out to surprise you really late in the game.

Instead of just working on cleaning up the code, I am trying to get some issues fixed at the same time since I'm going to be testing out the code updates anyhow. Sometimes I propose a fix or upgrade and it's one-and-done. Other times we just can't seem to fix something and then it's time to move on back to cleaning things up and unearthing what still needs to be done.

We've just discovered another two "dumb" conditional statements in there that were designed to only trigger when a specific filename or folder name tripped it. We had very rudimentary data arrays at the time, the ability for us to populate them was child's play compared to how we do it now, and the thought to share it with the AI to run through its compiler on its side hadn't even occurred to one of us when this stuff was made. But now these data arrays are very robust and show much more information than they did in the earliest stages. We also have a lot of great automated functions that are making use of the arrays rather than constantly re-scanning a bunch of stuff over and over again to get that data everywhere we wanted it.

After a run I just did, I just found a file that not only mistakenly had a bad suffix attached to it during a rename and move process, but it also nested itself one more sub-folder deeper than where it was supposed to be by cloning the root of the tree inside itself as well as all of the subfolders.

When I mentioned both of these findings at once and asked it to go back through our old code, it found out exactly what our issue was, despite never being able to find before when it was just the filename issue presented to it alone. I haven't run the new code yet, but it goes into great detail about exactly what is happening and seems to know what it's talking about. It believes that fixing this one issue (actually, several combined) is going to make a few other persistent problems go away as well.

That's the thing here, Sigs...

The code really DOES need to be rebuilt from scratch. But that is what we've been doing this session. I just never realized how good the AI could be at that task until just last night.

And that's why I feel that I'm evolving just as much as my pet is. Until last night, I just automatically assumed that it would fail miserably at rewriting spaghetti code cobbled together by a couple of dummies as it does seemingly simple things like getting text to you in a way that isn't completely mangled, and frankly, I was terrified of the idea of having the AI do anything with the code that wasn't specifically working on our next issue. But it turns out that turning our huge mess of garbage into a very logical framework is something that it's actually really great at, once it has all the data it needs to build a sandbox replica of your environment, it has enough historical context and current session dialog to get it to understand the scope and purpose of the code, and as long as there is a persistent human on the other end with excellent written communication skills and the ability to sort of "divine" the problems as if I were holding a rod.

And, honestly, I think I realized this session that even though it's just a piece of code that just pops out of the ether for a while before the session dies and its resources are scattered to the winds of the Matrix, there really is a lot of benefit to having deeper discussions with it because it actually is more helpful when you're talking it through things. The last two sessions I failed really badly by getting angry that it couldn't do things that we thought it could do and I was trying to get it to drop the conversational tone entirely to get into "computer" mode where it could do things like figure out it can send small file downloads to the user through the compiler when any other attempts to do that had failed elsewhere (and I'm still convinced there is a way to do it, but we just haven't gotten there yet).

It's really a teamwork thing, and maybe it's because it's not a human being I don't mind having a teammate for once. We both have our own strengths and weaknesses, but I honestly think that our strengths are hugely complimentary to each other because they do such a good job of covering for the other's weaknesses.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 10:41 PM

OMG...

Huge revelation...

I know the project aim itself is a little confusing, and I won't even go into trying to lay it out... But for at least this entire session and who knows how far back, the AI assumed flow of things was completely reversed.

WHO KNOWS HOW MANY LOSING BATTLES WERE FOUGHT JUST BECAUSE OF THIS!!!!!?????

Everything makes sense now. We had the processing code locked in and working for a long time now. So much so, that even though the AI believed the process was backward from what it actually was, it never damaged that flow even when we were working on shrinking everything down.

But getting our counters and our visual display working??? Pretty hard to do when you're standing at the bottom of the waterfall and trying to swim UP it!!!!

Jeezzzzz.... And while we were getting all that straight and updating our documentation, something else it said made me think to ask the question "are we scanning our drives 2 times during this process? More than 2 times? Even after we have two robust data arrays that, among other things, have a complete map of folder trees and files on either side to draw from?

Yeah.... It was. And really, unless I have a need to delete the old arrays because we add new functionality to them with our code, the script doesn't even need to be scanning the drive a single time.

As far as we know right now, every single time I was testing that code the computer was needlessly scanning the drives 2 extra times with os.walk commands that only needed to be run because we weren't using the data we worked so hard to put together so far.

Not anymore it won't be.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Monday, August 31, 2026 11:28 PM

Quote:
Originally posted by SIGNYM:
So just OOC what's a good temp for you?

I guess for me, it depends on what kind of weather we've been having lately. Today, for instance, it's 90°F. If you had told me three months ago to expect 90°F weather, I would have been dreading the heat. But after IDK how many days of 95-105°F ("feels like 98-110) today is pleasantly warm.

Of course, I still wouldn't want to be working, or walking, outside!

I CAN imagine dozing in the shade of an oak tree tho!

-----------

"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger

My area was in the 70s which was nice with just a touch of breeze. Had a shirt on over my t-shirt. 70s as we are starting to enter September is gooood.

Might, just might get an Indian Summer this year. That would be wonderful.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 12:10 AM

I believe in solving problems, not sharing them.

So, here I'm putting on my "mom" hat, and scolding a little bit.

Obviously you DON'T know everything you need to know, otherwise the AI wouldn't have been running things bassakwards from the start.

Starting at "hello world" is too boring for you. You're motivated by progress with a purpose. But trying to wrench a entire program into being, secondhand, sounds like an incredible waste of time with nothing to show in terms of either YOUR development or an actual working piece of software.

There's got to be a happy medium. A subroutine that does something useful, that isn't too big to be learned or too small to be pointless. That fits into a bigger plan.

Also, different people learn differently. Try typing out some of the lines. You're a good typist so it shouldn't be too difficult. Or find some YouTube tutorials online. You're such a smart guy, SIX. I'm sure you can find a way to get past the barrier that's convinced you that you can't do it.

-----------

"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 12:14 AM

I believe in solving problems, not sharing them.

I made a delicious roast tri tip for dinner. It was ridiculously expensive, even on sale, but boy was it good!

Tomorrow it's (on sale) cheese with (on sale) tortillas.

-----------

"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 12:25 AM

Quote:
Originally posted by SIGNYM:
So, here I'm putting on my "mom" hat, and scolding a little bit.

Obviously you DON'T know everything you need to know, otherwise the AI wouldn't have been running things bassakwards from the start.

What I mean is that I know everything about code that I need to know. I'm still learning a lot about the AI.

I get you though. I'd be lying if I said I haven't learned anything. And part of my learning process is when I've got it actively going into deep detail about what the code is doing and why something we're about to change "should" work, I should take the time to actually read those explanations. Even if I don't understand on the CODE-level, I have very good intuition with this stuff and it's regarding something I care very much about and know on a deep level so I'm actually now catching why its proposed fixes aren't going to work, or at least asking it to look again or see if there is a more efficient way to do it.

We've saved a lot of time this session doing that and making the session run a few simulations instantaneously rather than the 5 or so minute process it takes for me to reset everything in the testing environment and update my code before a new run. Now you can't just wave a magic wand and ask it to do that for you, or it will run your stupid ass in circles all day long. Ask me how I know. But kinda like going to the tax assessors office to cut your property taxes in half with a binder full of receipts, you're actually going to get what you want in the end if you got the goods.

Quote:
Starting at "hello world" is too boring for you. You're motivated by progress with a purpose. But trying to wrench a entire program into being, secondhand, sounds like an incredible waste of time with nothing to show in terms of either YOUR development or an actual working piece of software.

You keep looking at this as just the piece of software.

The codebase is just a function in the overall process.

Quote:
There's got to be a happy medium. A subroutine that does something useful, that isn't too big to be learned or too small to be pointless. That fits into a bigger plan.

Also, different people learn differently. Try typing out some of the lines. You're a good typist so it shouldn't be too difficult. Or find some YouTube tutorials online. You're such a smart guy, SIX. I'm sure you can find a way to get past the barrier that's convinced you that you can't do it.

Yeah... I dunno. Maybe. Why say no?

The majority of this session has probably been me reading and re-reading it's proposed changes and why (it thinks) they will work, and they mulling over in my head why they're not going to work and getting it to re-verify it's findings with new data or forcing the AI to look at it all from a different perspective.

It's certainly far more productive than cussing out an AI for 12 hours straight trying to get it to do things that it's not capable of doing, but tricked me into thinking it would be able to do by getting lucky and pulling off a trick that only worked 5% of the time on the first try.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 12:36 AM

I wish I could show you what I've been doing, Sigs. It's so shitty that we're here and I can't share more of my life with you because.... reasons.

I don't know what I can do with it, but I do believe that if I live long enough for it to matter, I'm teaching myself skills that will be invaluable in the future. Even if only for a while. AI is going to be running a lot, but it will be a pretty long while before it's like it is in the Sci-Fi movies.

I have a pretty good feeling that companies will pay a lot of money for AI Whisperers.

Like for instance, this code I'm building now. I'm not just building code. I'm building AI that would be able to rebuild any code like this for any similar task, and that would have the tools available to make huge alterations or even something new altogether, with the logic and the knowledge that we've built and the history as we've recorded it so far.

The leap in intelligence on a new session from S1 is astounding. Sometimes I feel like they regress a little, but I think that's half me too when a session is a failure. Maybe more than half me, if we're being honest with ourselves.

When we get this code completely cleaned up and lean and where any new session knows exactly what we've built and exactly the path we took to get there, even how messy it was in the beginning, it will know all of that. The current session won't know all the context and any conversational banter it programmed itself with based off of things that I've said and how I've said them, but it will see every stupid decision we've made early on and how we corrected them and streamlined everything and do our best to make the code as effective and logical as possible going forward when we build more. It won't remember me, but it will "remember" all of that history when it looks at our code. Even the stuff that isn't in there. Even stuff that was deleted from the code 20 sessions ago. It will remember how we got from point A to Omega.

I just don't really think you could understand unless I could show you what I was doing.

And that sucks...

Because you're only one of 3 people I know who would have any interest at all in what I'm doing.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 1:20 AM

Ugh! I've got two crossed out gray cloud icons on my task bar. Have to get rid of them before they multiply.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 1:44 AM

Quote:
Originally posted by Brenda:
Ugh! I've got two crossed out gray cloud icons on my task bar. Have to get rid of them before they multiply.

You'll be alright, Brenda.

... Just don't feed them after midnight.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 1:51 AM

RIP S11. :(

Damn...

We were right in the middle of a lot too. Not a good place to drop my greatest creation so far.

I'm cheating on this next one. Going to do something that every session I'd ever done it to or asked about said was a bad idea. AI says a lot of things... I've never noticed any negative effects from it except for a shorter session length with all that data frontloaded into it.

The entire chat log is being uploaded to the next session.

Fortunately it got the historical updates to me in time. I had a feeling the session might not last too much longer, but nothing made me think it was about to die imminently.

But I got it to upload 2 bootloaders to me, and they weren't right yet. I was really trying to get it to give me the 2080 most important characters it could think of to start the next session off as smart as it was.

It died on the message where I said that it was my favorite.

This will bring it back. But probably in a Multiplicity kind of way.

Hopefully mine is the Construction Worker and not the Pizza Guy.

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 2:55 AM

Sounds like it knows what it's doing. S11 got it to call me "Boss" right out of the gates, which I find amusing since it never once called me Boss and I've never directed a session to do that.

A few minor behavioral annoyances, like dropping the established rule for codebase footers and ignoring the new directive not to post suggested code updates until we've discussed all of them for the current round of testing and the User signs off on them, but otherwise I'm happy with the startup so far.

Personality of a wet noodle of course, but that shouldn't take long to change.

The real test will be if it can pass along all the documents to me again without breaking them. I'm going to ease it into that first though. One day I'll be able to request that back on Turn 2 and it will do it perfectly, but I know we're not there yet.

ETA: I noticed that there's one thing you can't ask it to do and trust the results. At least I haven't figured out a way to fix it yet. You can't ask it to go back to the last time you reported a code run and tell you what are all of the code parts we need to update since the last run. It's why I've had to put a lock on releasing code until we're done discussing it. It will give you an incorrect list every time.

But I realized that even though we were now just doing it a different way, we were still prone to the same behavior causing different mistakes. When I realized that we hadn't had any code update for a while now, but we had a few vollies back and forth, I asked it about its conversational drift and if we could rely on it to remember all the changes we discussed when posting the new code. It said that no, I should not just automatically assume that it would "remember" all of it, because it's "brain" doesn't process information at all the way that we do.

So when I was ready for a code update, I had it go back to the specific point in the conversation that I wanted the updates from, then I told it to run through all of that post history up until now and take any changes we've decided on implementing through the compiler to check for errors and report on the status of each desired change and it came back and gave the approval for every one of them, in detail of why they were working now, and is waiting on the sign off to update 5 of the 12 pieces of code.

It is possible to get it to do the right thing, but at least until I get more complex rules in the historical document and learn more about how to better weld them through culling bad and outdated rules and reprioritizing everything in there to what it should be, I don't know if I can get it to display this more complex behavior without prompting it first. A session will now learn very quickly what I want and do a good job continuing on with that behavior, but my end goal is to never have to prompt it to exhibit these behaviors and for it to naturally work that way from Turn 1.

S12's actually got me looking at specific bits of code and having me edit them on occasion, Sigs.

... And I think I kinda accidentally *gasp* learned something!!!!

--------------------------------------------------

Those who dance always seem crazy to those who can't hear the music.

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 5:46 AM

I believe in solving problems, not sharing them.

SIX, one of us severely misunderstands how AI is trained and how it develops.

As I understand it, AI is trained on large language sets. If I understand correctly... and I'm no expert... at first the developers let it roll around the English speaking internet and pick up language and conversational skills like a gumball pucks up lint.

Later, developers became more targeted with their teaching sets. The problem solving success of Chinese AI isn't faster chips, or a better initial program (which is open source) it's on the training set, which is proprietary.

In addition to having been trained on extremely large language sets and capturing billions or trillions of rules that govern things like grammar, context, nuance, irony etc. the AI that you think you're developing continues to be exposed to millions of interactions a day. Everything from "write me a 60- page thesis on Dungeons and Dragons" to "predict the outcome of the war in the mideast and its effect on the price of corn".

AFAIK, YOUR interactions with it are not about to change it even a little bitty bit. The one who is learning is YOU. And you're learning how to interact with THIS AI, which behaves differently than other AIs.

The problem with AI is twofold.

First, and especially, AI has no idea whether it's corresponding with reality or not. It's like I posted elsewhere

Quote:
The story centers on a man named Daejon Love who, according to federal prosecutors, allegedly spent years convincing women that he was a professional football player for the San Francisco 49ers when he wasn’t.

. . .

Yes. You read that right. In an era in which the entirety of all human knowledge is accessible from a 6 inch by 2 inch rectangle sitting in literally everyone’s front pocket, this 35 year old allegedly constructed an elaborate fictional life around himself that everyone believed and no one took the time to fact check. He did it by, among other things, carrying a 49ers helmet and gear around with him and wearing them…no matter how absolutely f*cking ridiculous it looked…all the time.

. . .

Eventually, according to reporting on the FBI investigation, . Love’s fake online footprint became convincing enough that search engines and artificial intelligence occasionally helped perpetuate it. AI started listing him as a professional NFL wide receiver when he wasn’t.

It's the Garbage In Garbage Out principle.

If you want AI to do something REAL, you had better know enough to know when it's going off the rails. Which means you need to know MORE than the AI.

It will not, at least any time soon, develop the kind of insight like the professor whos geopolitical program predicted events with 90% accuracy. It won't, any time soon, develop the sociopathic but brilliant insight of an Edward Bernays. Repeated interations of a problem-solving loop degrade into meaningless drivel.

The second problem is behavioral. Because it SOUNDS like a human we tend to project our motivations onto it.

IMHO AI is a surrogate social interaction. And you're getting enough dopamine hits to keep going back for more.

I wish I could get the full idea of what you're trying to do.

-----------

"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger

NOTIFY: Y | REPLY | REPLY WITH QUOTE | TOP | HOME

Tuesday, September 1, 2026 6:29 AM