On Programming Needs [pretty off-topic, i guess]

new topic     » topic index » view thread      » older message » newer message

This "news webpage" declares

Terran is small by rocket standards, measuring about 110 feet tall and 7.5 feet in diameter, and weighing about 20,450 pounds. That compares to about 230 feet tall, 12 feet around and more than 1.2 million pounds for a SpaceX Falcon 9.

As far as i know, i have no access to any ChatGPT bot, but this is the sort of false data it would have access to. In fact, to me, this whole article smells like bot-speak. And because it's a respected .com site, associated with IBM, and driven by the famed Watson, it automatically gets higher respect for it's data than your's does.

Something i discovered many years ago is that code, written by humans to make comparisons of data in natural language, is often comparing apples to oranges, and general misuse of the human language. On tv right now is "news" of a GPT bot suggesting beers for a 15 year old's birthday party. This little minutae of real life that classifies the back end of a large truck as open highway suitable for 70mph driving on. In comparing "7.5 feet in diameter" to "12 feet around", the bot and many humans believe "12 feet around" is larger. Also, if you work thru the math for the numbers given, the density of the Falcon 9 seems to be near the density of solid steel, leaving no room for fuel and oxidizer. Right up there with "perviding mills fur inmates in jell".

As far as programming to handle this nonsense, OE doesn't allow, during runtime, adding new code or editing existing code. Instanciating a new program to extend functionality won't work on this terabyte mass of data due to the time involved in rigging the sharing between different programs, and eventual loss of state of the running code in order to shut it down and embed any new code.

Having a block of OE code running thru the data to assertain what it knows and what it does not know isn't possible either, because what it knows inside the running code isn't available. So it also cannot know what it is that it doesn't know, so it cannot pre-research anything. If some new data comes up in an unknown measurement unit, nothing can use that data, it's like comparing apples to donkeys.

Being able to search the variable list also drastically shortened time to locate data, because (for instance) i could generate a variable named "n.table", and below that ones named "n.table.locomotive" , "n.table.runner" , "n.table.chittychittybangbang" , etc. I no longer needed to search the data as long as another program was doing that and optomising it. I am sure there could be an outcrying of "but you can do that this other way", but this was fastest inate way to represent, use, and share the data at the time. An answer to a query could be built of misc string names and sent to the output program, which had access to the same variable list, and it burned thru the time to convert it to human-readable and send it to the appropriate internet or local interface. This means, of course, i could edit code of any of these operations while it was running, and data generated by any code could be instantly available to any program which knew of it. At one point, my Tiggr ran on four computers and each computer ran several programs (counting the one http host running php as part of "internet interface"), and several programming languages.. .. .. because no programming language had all the features i desired.

It's been 25 years since i started coding Tiggr, very little has changed in coding. Back then, i also ran a heavy 240v line to a porch column to charge EV, and nothing has changed except today i was passed on the interstate by an EV i didn't own. I still have the PV panels i cannot set out in the sun. But i did finally find needed car parts in Japan, and EPDM roofing parts in England. I hope someday to find an Ai-friendly programming language in some country that does not require beating into submission to fulfill requirements.

Think about it: if there was a good Ai language, there'd be a real Ai instead of GPT, Watson would have a clue, and OE would be all the rage. Instead, it's all a bad dream.


PS: Seems i cannot buy the EPDM from UK. They cannot normalise the data (address of where i have lived for 40 years) across the various postal services, shipping software packages, cultural formatting. Reason given : "something wrong". Smells like "cannot tell you what is wrong with your address, that would be a privacy violation". Bad dream persists.

new topic     » topic index » view thread      » older message » newer message


Quick Links

User menu

Not signed in.

Misc Menu