(Edited at 1423 PST for typos, clarity, train of thought...)


With AI, LLM, ChatGPT, and other GPT tools increasingly trivializing the creation of original art and constantly getting right to the edge of replicating popular styles of famous artists whose works are copyrighted and fiercely defended, it seems that these tools may have quietly evaded 3D graphics engine fencing.

Given how much power is on tap in graphics cards made by the likes of Nvidia, how long can 3D modeling graphics kernel companies maintain strangleholds on CAD companies, if "stranglehold" is the right descriptive?

By extension, with so many GPTs, LLMs, and even SLMs (Small Language Models) bridging gaps not generally conceived of due to AI onslaughts picking up even more speed and power post-2023 (nearly-astronomically eclipsing the 1950s-2022), how long are post-2000s CAD companies even safe?

Many of the LLM/GPT — and probably SLMs — seem quite powerful at and useful for connecting existing apps of *dis*parate types and probably can help people *des*perate to hang on to their existing or older apps that are still nearly completely off line apps. So, if old, defunct apps don't need to (or by design just flat out don't) call home, can't be easily remotely or otherwise intruded upon (besides physical theft or by someone turning an insider rogue), and can be essentially revived, it seems that two possibilities (or a third, which is the status quo is tenuously intact) follow:

— old apps hardly get updated, and AI, LLM, GPT, etc become bridges or shims, carrying out directives of the user/artist reduced to easily understood scripted steps, linking disparate apps, or

— developers and lobbyists find ways to break, destroy, or outright illegalize old or even orphaned software, just to protect revenue streams more than preserve workflow rights and arbitrary app choice options of artists and non-artists, meaning abandonware and orphaned software end up being treated as munitions, Schedule 1 Controlled Substances, or as intolerable, nuke-on-sight threats to trade.


This subject was inspired ny reading a few articles this morning (well, less than 30 minutes of reading two articles), one being:


"OpenAI adopts rival Anthropic’s standard for connecting AI models to data"


https://techcrunch.com/2...cting-ai-models-to-data/


the other being:


"OpenAI peels back ChatGPT’s safeguards around image creation"

https://techcrunch.com/2...cting-ai-models-to-data/


My hot take (or belabored take/s):


At some point, AI-driven/powered/enabled tools and adjacent processes will — rightly or wrongly, correctly or incorrectly, usefully or horribly — reduce everything to mere, sheer data, and mimicry of users will make it easier for trusting or wary users to just turn those tools loose to:

— save time,

— make money or make more money,

— avoid hiring less-reliable humans if none have been hired (thus avoiding the pitfalls of having to deal with/track/pay out salaries, taxes, benefits, fret over IP protection, and labor laws, HR departments, etc),

— streamline or cull headcount,

— better filter WHICH humans are hired as labor or hired as "hot-swappable consultants" or contractors,

— (plug in here your own ideas/fears).


At some point, when these tools actually entice (and are not per-token/per-request hellishly expensive or environmentally ruinous) users to (brazenly and greedily) ask the AI to generate not just an A/B shootout, but generate 5, 10, 20, or 50 concurrent, complementary, interchangeable, hot-swap-compatible lineups of products, all to airtight close off competitive avenues, then write all the attendant, suffocating, offensive insufferably weed-y, labyrinthine patent applications for each, jacking up the billable hours costs on anyone needing even a fair, non-greedy patent, even if to file for a defensive, sort of open-source-protective, inclusive patent, so no one greedster can hog all outflowing or adjacent concepts (sigh; "embodiments ") grabby patents.



Just as I typed those words out, an imaginary ball peen hammer hit my forehead:

It won't be long before IP hoarders find a way to reduce the USPTO (and by extension, negotiated from some yacht somewhere) to mostly-AI processes and decision-making, meaning human employees/patent examiners get cut out of the Patent Office, in the name of "efficiency", but in reality, to further enhance that the highest bidder and fastest mover definitely can rule entire landscapes.


So, rounding this out, a whole new idea is in mind:

CAD development companies needing to save money AND retain remaining loyal fans/customers on development costs might want to find open pathways to ditching any proprietary components they don't own but which could be eating up more than 10, 15, 40, or more percent of their non-payroll costs.

I'm wondering what the likes of Blender, GIMP, FreeCAD, and others do for avoiding unsustainable component costs.

Which brings up another idea, a scary one:

Will billionaires by fiat finally stage a coup against creatives and declare Open Source code illegal, or lay so many legal traps around it there is massive confusion, fear, uncertainty, and doubt?


Closing, before I began writing this, I once again thought:

— Gods, I hope Tim is rewriting the layers and constraints engines, to truly MAKE the app offerings a no-brainer to buy even if the base price is $1,500 after a discount offer.

— I wonder what is so intense the team can't even once a week show it has a human pulse, not a CRM/scripted coupon offer, and not just visitor stats at page bottom.

(Oh: My PunchCAD coupon newsletters/emails are again arriving in spam because, I suspect, someone is flagging (or abusing, perhaps anti competitively?) PunchCAD up as a spam source, or someone is trying to kill the company. I keep telling Google PunchCAD is not spam, and every few months, amnesia or sabotage keeps resurrecting.)

The above is a lot to chew on, but, it's depressing to neurotically tab over to the site and see 8+ days of no posts, CAD, ViaCAD, SharkCAD, or otherwise.


I can feel in my bones that if I can fair my (networked (as in visual or topo grid, not LAN/WAN/MAN/etc) and along a path) ship hull curves better (as opposed to using ship-design-specific apps and plugins; and, note: I don't use meshes to make my hulls; I just can't) and if I can get to see come true my swan-diving-into-a-tree-trimmer fantasies of SharkCAD:

— having a rebuilt, completely flexible layer manager reorganization (not just adding tags, but having truly draggable layers like in a file manager/explorer) (BUT, BUT,NOT look like flat oack or web browser apps — just super-upgrade the LM and CE, please!),

— the Concept Explorer and the Layer Manager, and the Inspector become more inspiringly and functionally linked to act as "SUPER FILTERS",

— the Inspector be made resizable, so long-layer names can be sussed out by a horizontal slider (instead of the user having to feel dictated at to keep layer names shorter than the Inspector fixed width) and doing crazy focus-wrecking things like renaming or prefixing layers to find which of the 50 with same first 30 chars is the actual layer I'm hunting or editing),

— the Layer Manager and Concept Explorer gain ability to respond to RegEx-like searches,

— the LM and CE and Inspector accept multi-letter, down-filtering, not just single-letter jumps,

— the keyboard shortcuts gain 3, 4, 5-letter characters slow or fast input recognition, not just a single-letter or single-letter plus modifier keys input,

— drawings and geometry, and drawn text (not just layer names and number strings) be copied and pasted, or even dragged-and-dropped between drawings,

— layer trees/branches gain ability to be torn off and copied without the whole drawing having to be painstakingly, tediously pruned, compacted — after a needlessly-new-top-layer-subnested layer is added atop — by operation of export operation,

— branches in the CE or LM (not just geometry lasso-ed on the screen) can be instantly tuened into symbols AND RETAIN layers (not just be exported, with all the compulsory, ensuing pruning),

— the legal, technical, or emotional decision to allow only one instance of the app on the same hardware (not pirated, not LAN/otherwise-netted out to non-owners of the hardware, etc), so the user isn't breaking any contract law while coping with crash-y files and not wanting to have in one crash-potential session a reference file that could bring down any open work...

Well...

I could see SharkCAD finding a resurgence of interest (provided there are no oppositional interests throwing spikes in the product line's marching path.

I want to see that product FORWARD, MARCH, not just toing Mark Time, MARCH!


I feel that with my drawings, I could help stir up new or first-time purchasers to leap straight to PunchCAD's "Buy Now" button.

But, the app, as it is, is sort of not helping, and leaves me in a whirlwind of "What am I gonna do" while I can't make a decision but stay pat/put, hoping all the silence will bring amazing reward I've been for 10+ years begging for. Even if people *leap* to buy my drawings, and even if they trial SharkCAD, I'm 1,000,000% positive they'll rend me to pieces and plasma-drive-burn my atoms once they start trying to reorg the layers to their whims or needs, and find they CAN'T, and if they export to other apps with the intent to reorganize externally, but round-trip back to SharkCAD, the geometry attributes are stripped (my experience in 2011 w/ VCP or SLT), I will be *toast*, reputationally *destroyed*.

This kind of pressure incredibly stresses me out and keeps telling me if Shark doesn't change for me, and obviously for anyone needing to manipulate or examine, or selectively use any of over 2,000 layers with geometry that can't be by-name RegEx-like or filter-box hunted for, I have to grudgingly give up and shift over to Rhinoceros.

But, my heart wants to show that as far back as 2011 — if Shark and VCP had what I'm bleating on about, naval architects (I am NOT one and have never claimed to be one or intimated being on) could have exploited general CAD well beyond what I as a hobbyist have been.



Again, the above is a lot to chew on, but, again, it's depressing and even deeply wounding to neurotically tab over to the site and see 8+ days of no posts, CAD, ViaCAD, SharkCAD, or otherwise from fellow users and a year-plus from any representative or regular members of the Company, rolling out to months. (Again, where's the other communications venue of which I keep forgetting?)

In a good, positive, well-meaning way, I posit this/the following:

Maybe Tim is cloning himself in AI so he can leave a pleasing, fitting legacy. I LOVE SharkCAD for its clean interface (if I weren't so psychologically linked to SharkCAD's and ViaCAD's GUI, I would have hopped over to Rhino/Grasshopper/Orca 3D/Nemo/etc back in 2016 or 2017) or ability to clutter it up as desired, but I hope it gains some new tricks and traits.


2025 03 28, 1215 - 1355

Edited by user Friday, March 28, 2025 3:24:46 PM(UTC)  | Reason: Edited at 1423 PST for typos, clarity, train of thought...)