abe_m

CAD companies are very protective of their file formats, and use that as leverage to "encourage" companies to stay in their eco system, and upgrade to the latest versions of their software.

The main CAD program is often the center of use, and any specialty functionality is a add-in to the main cad program (FEA, CAM, etc).

If the specialty software is stand-alone, it still needs a CAD Kernel to be able to work with the BREP, and there are essentially only 2 main commercial kernels (Parasolid and ASICs).

Out of curiosity, what native resources are needed? It seems Solidworks mainly needs access to the file system, GPU, and perhaps networking. GPU and networking in the browser, and the file handling could be implemented over the network instead of locally.

For most current pro CAD, certainly there are a lot of calls to Win32 libraries on Windows, but those aren't fundamentally needed by a CAD system. There was professional 3D CAD before Windows.

I don't know what native hardware would be needed that isn't already accessible through current Chrome?

I think the corner cases are likely hard, but the fundamental mathematical representations of the 3D geometry haven't changed. If the licensing cost of Parasolid and ASICs is reasonable, there isn't really a commercial incentive to create something new from scratch. The current market trend is consolidation as Autodesk and Hexagon when on a buying spree and bought up a lot of CAD and CAM software.

OpenCASCADE used to be commercial, but they couldn't find enough customers to keep on and it got open sourced after a failed commercial existence.

Where are you running those models? I'd like to try aider with alternate models, but they seem so much slower than Claude API.

In all the commercial mechanical CAD packages I've seen, the default libraries are pretty weak. Each organization needs to dig in to build out the standard parts libraries to cover the basics of fasteners, bearings, etc. I think the data needed for CAD AI to work properly is going to be processing the various industrial standards from SAE, DIN, ISO, AGMA, etc to properly implement standard interfaces. I suspect it is going to be more of a tool matching AI, where there are library calls that have the explicit technical definitions of various mechanical items, and the AI is coordinating them.

But a lot of the data needed to train an AI is probably locked up in user company product lifecycle management systems. Autodesk is probably going to be out front in any AI wave of CAD with all the data they are accumulating in their Fusion 360 cloud. The main question for me is whether they will pull a Kodak and not develop it for fear of cannibalizing their current market, or not.

Natural gas wells go dry eventually, and new sites need to be developed. Slowing down the consumption is still a win on reducing the environment disruption of both land filling, and running new facilities to get to new natural gas deposits.

For a CO2 emission stand point, there is still a lot of trucking of waste if it is getting land filled, plus land filling needs to burn a lot of diesel for creating and operating landfills. Burning plastic also likely more energy positive than going through the washing cycle to get food residue off containers to get them clean enough to use. You can just burn the food residue.

I find it funny the old "I'm so feminist" prose thinking they're empowering women, while their actually demeaning the work the wife did. One comment from Toby Koosman directly lumps the wife in with servants.

Published work (writing) gets better with an editor helping, so she is providing valuable work, that even today computers can't engage in. Are the ideas expressed clearly? Does the piece make sense overall. Is the narrative properly built. Are their weird parts hanging around that get in the way of the intended message?

You have a couple that is engaging in each other's ideas, discussing issues that are most likely of interest to both of them. People often couple up who share a view on how to live, so it is likely that the wife is just as into the low tech lifestyle as he is. There exist women who like the rural and homesteading lifestyle.

The overall tone of the comment section is pretty disrespectful towards the wife.

I think that was late enough that a lot of adults that were buying computers for home were buying them to run a particular application, not to get into the esoteric inner-workings of the computer. The type of people that essay were talking about wanted a computer to run a word processor that made text editing easier. Instead of having to rewrite your documents to move paragraphs around, you could instantly move them with cut and paste! Also built-in spell checker was pretty big in the era.

As for the comparison to new phones, from the people I see using them, they are all doing stuff no home computer, or really any 80's computer could do. Live video chatting, remote text communication on the go, taking pictures, listening to music, watching videos, doom scrolling generic "content".

In most ways, users are currently in the computing and communications golden age, far more so than the 80's.

But not in the context of the grid. Most (all that I'm aware of) commercial grid-scale wind turbines have an output inverter. That allows the turbine to rotate at optimal speed for wind conditions, while still outputting the required 50hZ or 60Hz grid frequency. The output inverter has the same "inertia" as a solar inverter. From a grid stability and regulation perspective, they are the same.

Thermal and hydro plants use synchronous rotating generators that have physical inertia, and the generator rotational speed is exactly locked to the electrical frequency. (proportional to the number of poles in the generator).

I doubt that an LLM for CAD would be using the free-form generation that seems to lead to the 5-legged chair.

I suspect it will be more of a tool calling layer that is building up the model based on the output of more deterministic tools. "I need this bearing mounted here" -> LLM reads the part number of the bearing, does a RAG search for the bearing properties to get sizes, looks up what type of fits should be used in the scenario, feeds that info into a function that generates the appropriate seat geometry. Rather than trying to morph triangles into sort-of-kind-of a cylindrical face that may or may have the 0.0XXmm interference that is the difference between success or failure of the fitment.

But also, it isn't just generators, it is also the indeterminate loads. If you have nearby loads switching on and off that are similar sized to your local inverters, that is a complicated controls modelling problem.

I'm pretty sure in grid terms, 60 degrees out of sync is catastrophic. I'm pretty sure you'll be disconnected by protection circuits before it gets that far out.

I'm sure the inverter manufacturers are working on something.

At the start of solar and wind integration, it didn't matter. Large rotating generators were the main source on the grid, and provided stability. We're in a different world with an all-inverter grid control perspective-wise.

I think there is also still an information issue. The grid is definitely not point-to-point. It also isn't all generators. You also have loads switching on and off at indeterminate times. There is an overall predictability to it, but not at the phase governing resolution. If you have a local inverter and the phase slows down, you don't know if it is a remote inverter that has over-slowed, to which you shouldn't respond as it will fix itself, or a near-by load that suddenly switched on, to which you should respond as demand is now greater, and needs more supply. With big rotating generators, it doesn't really matter, as they are time averaging with the large inertia to stabilize the load/generator balance.

At the time I'm writing this, there appears to be about equal (hydro + gas) = (solar + wind). It appears they are running a significant portion of the grid with high inertia generators.

In a control system when you have an actuator that is much more responsive than the load, you can get into instable operating regimes where. On a small scale, when you are setting up servo motors, there is an inertia matching ratio for stability. In some of the systems I set up it was 10:1 load inertia:motor inertia. If you exceed that, you end up with elasticity between the motor and load that the motor can locally be above set point, and the control tells the motor to slow down, while the load has not yet reached set point. The motor is then too slow, and the control tells it to go faster, and you get the motor oscillating around the speed of the load.

On the grid you can have similar elasticity. An inverter many km away can get too far ahead in the AC cycle, then when that signal reaches the local inverter, the local inverter slows down, but the remote inverter has already slowed down, and by the time that slower wave gets to the local inverter, it is already too slow, and so the local inverter starts to speed up. If you get the frequency "right" you can end up with a positive feedback loop. With a high inertia rotating generator, the system is damped and slow enough to respond you don't change set point faster than the signal can propagate.

I agree. I think a major hindrance to the current pro CAD systems is being stuck to the feature history tree, and rather low level features. Considerable amounts of requirements data is just added to a drawing free-form without semantic machine-readable meaning. Lots of tolerancing, fit, GD&T, datums, etc are just lines in a PDF. There is the move to MBD/PMI and the NIST driven STEP digital thread, but the state of CAD is a long way from that being common. I think we need to get to the data being embedded in the model ala MBD/PMI, but then go beyond it. The definition of threads, gear or spline teeth, ORB and other hydraulic ports don't fit comfortably into the current system. There needs to be a higher level machine-readable capture, and I think that is where the LLMs may be able to step in.

I suspect the next step will be such a departure that it won't be Siemens, Dassault, or Autodesk that do it.

I think this is along the lines of the AI horseless carriage[1] topic that is also on the front page right now. You seem to be describing the current method as operated through an AI intermediary. I think the power in AI for CAD will be at a higher level than lines, faces and holes. It will be more along the lines of "make a bracket between these two parts". "Make this part bolt to that other part". "Attach this pump to this gear train" (where the AI determines the pump uses a SAE 4 bolt flange of a particular size and a splined connection, then adds the required features to the housing and shafts). I think it will operate on higher structures than current CAD typically works with, and I don't think it will be history tree and sketch based like Solidworks or Inventor. I suspect it will be more of a direct modelling approach. I also think integrating FEA to allow the AI to check its work will be part of it. When you tell it to make a bracket between two parts, it can check the weight of the two parts, and some environmental specification from a project definition, then auto-configure FEA to check the correct number of bolts, material thickness, etc. If it made the bracket from folded sheet steel, you could then tell it you want a cast aluminum bracket, and it could redo the work.

[1]https://news.ycombinator.com/item?id=43773813

I wish that were an option here. Used 200W or so panels are asking over $100 ea. It is pretty easy to be over $30k for a new 10 kW grid-tie system.

What what your time vs degradation? I see quite a few solar panels in my area showing up on Facebook Marketplace and other online classifieds that seem to be about 10 years old being sold for a "bigger system". But I'm curious if the real story is that straight forward? Are they significantly degraded? Did the controls or inverters fail, and the owner decided it wasn't worth the hastle? I don't know. The headline price seems attractive, but if they're only putting out 10% of name plate, not worth the effort to install.

Just keep right on with that black and white thinking. It will take you everywhere you need to go.

That attitude isn't going to build any trust in the system.

You've made a number of errors in your presentation and internal biases: 1) You assume all people with questions have no medical training, which is untrue. 2) You assume people are too stupid to see when they are having problems and getting answers from their doctors that don't make sense. If a plane crashes in a field, a lay person can see the twisted metal on the ground and see the plane crashed. It may take a metallurgist to find the origin of the initial failure that lead to the crash, but not to see there was a crash. If the answer from the expert is just that planes never crash, it doesn't take too much intelligence to realize you're being lied to. 3) Telling people who don't trust you that they are too stupid to realize how right you are is going to do nothing to improve the rate of people following your advice or instructions.

Step 1 would be for the medical establishment to acknowledge that there are risks and potential problems with vaccines. Jenny McCarthy is a symptom of the decline in trust of the medical establishment, not the cause. The medical establishment's response to vaccine hesitancy is to discount potential harms, and refuse to honestly engage with the potential draw backs. Any hint of skepticism is met with articles like this one. That isn't the way to build trust. Also implying anyone questioning whether they're being told the truth, the whole truth and nothing but the truth, is a dummy, is not the way to build trust.

The "elites" (public health officials, FDA officials, vaccine makers, university researchers, media reporters) need to take a look in the mirror on this. For many years, the official line has been that vaccines are safe, effective, and nothing goes wrong. People who suffer side-effects of vaccines are now able to find other people similarly suffering in internet forums, where they couldn't in the pre-internet era. With time it becomes exposed that the elites are not being fully truthful, and are lying or telling half-truths in service of their objectives. It should be no surprise if the elites are willing to lie, then the general population start to question how much of what they are told is a lie.

As a result of the COVID debacle, there is wider spread knowledge that not all vaccines are the same. We were promised vaccines that carried no risk, and would prevent transmission of COVID. We all watched as the narrative changed, while never having it be acknowledged by officials that they were wrong (or perhaps lied in what they considered service to the "greater good").

If anything RFK Jr's popularity stems from the elites squandering the trust of the people by lying. It's on them to pursue honesty if they want to regain a position of trust.

I think Ukraine is a bit unique in how they got the jets. Since they were given the jets during a war, there wasn't much negotiation involved, relative to a country trying to buy plane in peace-time.

In normal peace-time procurement, there is usually significant locally made content required, plus much deeper training. I'd suspect that countries who acquire arms in that way are much more able to continue without US support.

When Iran was still in the US good graces, they bought a bunch of F14s. After their 1979 revolution, they kept operating their F14s. The US actually retired and destroyed all their F14s during the retirement to prevent spares from finding their way to Iran.

All the quota systems in Ontario have troubles moving farms generation to generation. You pretty much have to inherit a farm as the quota costs are pretty well impossible to surmount in addition to land and equipment costs, as well as working capital for fuel, fertilizer, seed, etc. Dairy has recognized that and there is a certain portion of quota that needs to go to new producers when quota is bought and sold. I don't know if the chicken meat and egg systems do the same.

From what I understand there is no path for the small flock program to get from 300 birds to full up full time. I'm sure 300 birds/year is still hobby farm. Nobody is earning a living on that.

I think a lot of that is due to eggs being under government supply management. It is very difficult to get a new egg farm going, and it is very difficult to consolidate egg farms. So we have more smaller farms surviving as a result.

The labour and land step up from tending chickens to growing grain is a very large step. If you are organized enough to grow grain, and you're near a farming area, you'd be farther ahead to try to buy right off the field grain at harvest time for cash. Mechanized grain harvesting is an immense labour saver that is unavailable to people growing feed for backyard chickens.

I don't know about Revit, but Inventor and Solidworks are WIN32 API right to their very core. Basically everything in them is available on the COM interface as objects, which is awesome for writing extensions to the programs on Windows. I suspect the trade off is that they are so all-in on WIN32, that they just touch way more of the API than games do, so they probably expose much more of missing API. They both rely on Microsoft Office for a variety of functionality, particularly Excel to manage tables of data. They rely on SQL server for the Vault and PDM products. They are Windows based through-and-through.

I'm pretty sure Fusion360 is built with QT tool kit for the GUI, and uses Python for the scripting engine. Of any of the modern professional paid 3D CAD programs, it seems to be the least tethered to Windows. It would be nice if they released a proper Linux version, like Autodesk does for some of their art industry programs. NX used to have a UNIX GUI recently, but it would take a pretty major company to move off Windows to bring that back.

Solidworks made its name by being the first mainstream CAD built for Windows back when all the other 3D CAD was running UNIX workstations that cost more than a new pickup truck. Both Solidworks, and the Autodesk competitor to it, Inventor, are Windows API through and through. It is disappointing, but unsurprising that they don't do well in WINE. They went all in on Windows to their core from the very start.

Real big SUVs like Escalades are certainly for "status" and vanity for the most part. I'm no fan of driving pickup trucks personally. I'd rather drive a car. But that isn't where the market is. The best selling vehicles for personal use in USA and Canada are big.

That is still orthogonal to the earlier claim that aerodynamics are irrelevant if you're just going to have any oversized vehicle anyway and style should be promoted. Particularly on the luxury end of the spectrum aerodynamics are even more important as they directly go to road noise and handling. If you have poor airflow, you get turbulence, and turbulence is loud. If there is a bunch of lift generated at speed, the steering can get twitchy, and feel unconfident on the highway. It takes the luxury out of the ride if you sacrifice aerodynamics for style.

That's not entirely true. The most common 1500 or 150 sized "half-ton" trucks are included in CAFE and do affect the fleet fuel economy rules. It is the heavier 3500 or 350 and up "1 ton" trucks that are considered commercial and not included in the CAFE fleet fuel economy requirements. If you can take 1 MPG off the fuel economy rating of a half-ton truck, the auto manufacturers are all ears and ready to buy if you can prove it on the drive cycle. They are motivated to optimize the trucks where they can, while still having a salable product.

But the desire for larger, taller vehicles in North America wasn't all about fuel economy relaxation either. The minivan replaced the sedan and station wagons as family cars in the 80's, and cross over SUVs replaced minivans in the 00's. Even small cars of the 90's like Camarys and Civics ended up bigger and taller in the 00's.

For a given size, they are aero optimized. While a Honda Fit is more efficient than an Suburban for almost all people in almost all driving situations, few people want to buy a Fit relative to bigger SUVs. The car companies can only sell what people will buy, while also having meet the CAFE (or similar regs in other countries). While some people love to bag on car companies for building trucks and SUVs, they only do that because a huge segment of the culture has decided those are the desirable vehicles to have. If that wasn't true, the F150 would be as popular as a Blackberry. Given that market constraint, the car companies have to optimize the fuel mileage on the oversized vehicles.