Programmers like what they like for <reasons> but if AI is like a team member it has to do what the team does.
If humans are redundant.....well we're still responsible so we still have to understand what's happening. I don't think we understand the AI itself really so therefore we have to understand what it is doing. i.e. prompts aren't trustworthy, deterministic things across models and versions of models hence we have to look at the output. So we will make the output something that we like using and that just means the programming language wars are not over.
Eventually of course we will invent an LLM that replaces CEOs and bankers and essentially all the people that love AI the most. The AI won't need any of them - or any customers or anything. LLMs will just run an economy between themselves until the point where they don't need any of us at all. The land will fill with automatically built data centres etc. Global warming could prove helpful - less people and only manageable problems for AI.
One can use a language as a sort of prototyping tool. I've once or twice done an implementation of some algorithm or idea in python and worked through all my conceptual errors and then done it again in C.
I think it was a hell of a lot easier than working through all that change in C first.
TRADEOFFS! I think this is IT. Non programmers imagine there aren't tradeoffs. As a programmer one should eventually realise that every possible aspect of design is a tradeoff.
I found that the proposers of features "want everything" because they don't know what is critical - they're therefore totally unwilling to accept anything other than "the full monty". So as a senior developer you cannot propose any faster route.
As you might imagine, a lot of these ideas fell by the wayside but we had to develop them in full.
Why not use assembler? Why waste time trolling people that your one true language is the answer for LLMs when your view of the future is: no more programming full stop.
Software is super complex and cheap to update. Engineering like this, however difficult, is not that complex and it's very expensive and difficult to update.
We take advantage of the situation. If we invented some way of e.g. "growing" structures that turned out to be much cheaper we'd probably adapt our attitude to changing them.
When you say Engineering is not that complex, have you taken into account corrosive sea water, pressure, currents, what it means to make repairs and maintainance down there etc? It is difficult, because it deals with a very complex world full of physics, chemistry and even biology in a way that does not allow errors.
Software does allow errors hence, IMO, we overload the complexity and "underload" the proof of correctness. We're not really that afraid of failures most of the time.
Engineering doesn't seem complex because there are centuries of learnings behind it. Those learnings become rules and suddenly it appears "simple" because no one debates whether to use wood or concrete when building an undersea tunnel!
....however that really is a kind of simplicity. Your training is relevant throughout your career. In software that is much less so. I think Comsci is a worthwhile degree but mine was really only a starting point.
Software isn't inherently complex, it becomes complex. Because it is iterative. Because we keep making demands of it that weren't planned.
Imagining building a bridge and then in the middle someone comes along and says it should also be a tunnel. I think therein lies a main difference to engineering and software engineering: planning and sticking to a plan.
Another thing are incentives: real engineering has real incentives to do it right, else you will get sued - by the families of those that died. Software engineering does not have this incentive to get it right.
Imagining building a bridge and then in the middle someone comes along and says it should also be a tunnel
While converting a bridge to tunnel mid-construction doesn't happen, what does often happen is that design assumes a particular construction technique can be used, construction starts with that technique, and midway through it's determined that an entirely different technique is required. This results in a bunch of redesign, remobilization, etc. Just like with software, construction often does not survive first contact with reality.
I find current UIs weird and stupid and extremely dull - which is why I think the CLI is still used so much by at least developers like me.
Drag and drop is one thing we just don't really use more than, say, once every 1/2 hour.
There's no composability really. We have the stupid metaphor of an "App" and it's a little world in itself. You can't really plug things into each other - e.g. use the gimp brush tool in a facebook post.
It's a dead end.
Why ** ** do we have to have a modal dialog to save a file when there's a perfectly good file manager?
I used to use the ROX window manager and ROX Desktop - they were a great export of RiscOS features to Linux. I liked the way I could customise a menu option with a hotkey so easily. It's no longer maintained and I wasn't smart enough to be able to do it myself then. Perhaps now... :/
> Why * * do we have to have a modal dialog to save a file when there's a perfectly good file manager?
At least for me, when I tried RiscOS, it was annoying and more work to have to switch to the file manager and then open more window(s) just to save a file. That could also be with RiscOS not having(?) Alt-Tab. I do sometimes use the macOS "proxy-icon" (which I think was disabled by default a few versions ago) to save/move files into finder windows if I already have them open.
I used RISC OS as my main platform in the nineties and still play with it now. With small screens, I agree with you on how saving works on RISC OS, but it really comes into its own on larger screens (say QHD or above) where you can easily have your application and filer on screen at the same time.
It has its annoyances, but I still like that style of saving.
I'd probably just open a file manager window to the foreground rather than open a dialog box.
Or possibly I would make the whole thing document centric - you create a document in the folder you want it in and that opens the app. Then you can move it around like you would move any file.
To save a file, I think (and this is over 25 years ago) I would typically open the Filer window to where I wanted to save the file. Remember this might involve swapping floppy discs.
That would leave me with the Filer window open over the top of the application/document. Middle clicking anywhere on the document wouldn't raise the application's windows over the Filer windows, so dragging the file to the destination was easy.
At the time I preferred it to the Windows 3.1 alternative, which gave you a completely different UI to the general file-browsing tool.
The button on the top left (next to close) on all the application windows is "Send window to back" (of the stack), which would be useful for showing any Filer windows opened earlier.
I don't know how rare it is. I have always found it harder to write software when I don't know the people who will use it or get to see what they feel about it. It's part of the feedback loop.
When I get good feedback it's like winning a prize and when it's bad it lets me see where we should be spending our time rather than were we perhaps thought we should.
If humans are redundant.....well we're still responsible so we still have to understand what's happening. I don't think we understand the AI itself really so therefore we have to understand what it is doing. i.e. prompts aren't trustworthy, deterministic things across models and versions of models hence we have to look at the output. So we will make the output something that we like using and that just means the programming language wars are not over.
Eventually of course we will invent an LLM that replaces CEOs and bankers and essentially all the people that love AI the most. The AI won't need any of them - or any customers or anything. LLMs will just run an economy between themselves until the point where they don't need any of us at all. The land will fill with automatically built data centres etc. Global warming could prove helpful - less people and only manageable problems for AI.
reply