Just a Southern Saskatchewan retiree looking for a place to keep up with stuff.
I think that ChatGPT is probably the wrong tool for what I’m imagining. I’m thinking more in terms of “hypothesis generators” and “theorem testers” that, as far as I know, are not using the methods of ChatGPT in their operation. I think that those kinds of tools and others like them could be used to help clarify requirements before coding even starts.
Ah, I understand now. Yes, I think that maybe I agree with you in general.
I still think that AI operated by ethical experts has much to offer when used not an automated replacement, but as a tool that can save time and help verify accuracy. I’m thinking in terms of a kind of teamwork where one member of the team is an AI system or assistant.
I agree with most of what you said, but I think I was not clear in my presentation of the domain of operations. I was not speaking to the rewriting of an existing system, but if gathering requirements for a system that is intended to replace existing manual systems or to create systems for brand new tasks.
That is, there is no existing code to work with, or at least nothing that is fit for purpose. Thus, you are starting at the beginning, where people have no choice but to describe something they would like to have.
Your reference to hallucination leads me to think that you are limiting your concept of AI to the generative large language models. There are other AI systems that operate on different principles. I was not suggesting that a G-LLM was the right tool for the job, only that AI could be brought to bear in analyzing requirements and specifications.
I think he’s missed a potential benefit of AI.
He seems to be speaking mostly of greenfield development, the creation of something that has never been done before. My experience was always in the field of “computerizing” existing manual processes.
I agree with him regarding the difficulty of gathering requirements and creating specifications that can be turned into code. My experience working as a solo programmer for tiny businesses (max 20 employees) was that very few people can actually articulate what they want and most of those that can don’t actually know what they want. The tiny number of people left miss all the hacks that are already baked into their existing processes to deal with gaps, inconsistencies, and mutually contradictory rules. This must be even worse in greenfield development.
That is not saying anything negative. If it were any other way, then they would have had success hiring their nephew to do the work. :)
Where I think AI could useful during that phase of work is in helping detect those gaps, inconsistencies, and contradictory rules. This would clearly not be the AI that spits out a database schema or a bit of Python code, but would nonetheless be AI.
We have AI systems that are quite good at summarizing the written word and other AI systems that are quite good at logical analysis of properly structured statements. It strikes me that it should be possible to turn the customers’ system descriptions into something that can be checked for gaps, inconsistencies, and contradictions. Working iteratively, alone at the start, then with expert assistance, to develop something that can be passed on to the development team.
The earlier the flaws can be discovered and the more frequently that the customer is doing the discovery, the easier those flaws are to address. The most successful and most enjoyable of all my projects were those where I was being hired explicitly to help root out all those flaws in the semi-computerized system they had already constructed (often enough by a nephew!).
I’m not talking about waterfall development, where everything is written in stone before coding starts. Sticking with water flow metaphors, I’m talking about a design and development flow that has fewer eddies, fewer sets of dangerous rapids, and less backtracking to find a different channel.
Yes, and thanks!
Yeah, I’m not a fan of the form of capitalism that’s about selling what they want us to buy instead of what we want to buy, but it seems to be working for pretty much every company out there.
I guess we missed our window of opportunity with Netflix. We moved to the middle of nowhere with no internet or cell service 12 years ago. We’ve had Starlink for nearly 2 years and are just starting to run out of stuff available for free on our Roku. It’s been a couple of decades since I played with, um, other options, but I somehow doubt it’s become more difficult. :)
Former water treatment plant operator. Thanks for living up to your name.
I’ve been very happy with my System 76 laptop. I suspect I’ll try Framework next time, just because I like the concept.
No kidding! I don’t think I’ve been without some kind of debugger since I was still just dicking around as a hobbyist. I can’t even really guess, because it seems like forever. I know I didn’t have access to a debugger on the VIC-20, but just when I first used one I can’t say.
That’s just the way things work when humans self-organize. There is the appearance of structure at the beginning, because there just aren’t that many people with shared interests. Then as people are unsuccessful in finding the community they’d like (assuming they even looked!) more are created. Then more people come in and mill about and browse and get overwhelmed by the search for a needle in a haystack, so they create more.
Eventually, some communities reach a critical mass and a bunch of small ones fade away into near irrelevance or disappear completely.
As far as I know, the only way to put the brakes on community over-proliferation (if that’s even a real thing!) is to add a bit of friction to the creation process. Many kinds of friction devolve into centralization and gatekeeping, so they tend to be avoided in projects like this.
The only kind of friction that I can see working and gaining acceptance would be some kind of “have you tried these communities?” auto-search during the creation process. Simply asking people to search first is unproductive for two reasons. First, people are notoriously bad at imagining that someone else might have thought of something first, especially when they are only person they know with that particular interest. (I’ve only met a dozen other programmers in 43 years. In my entire life (66) I’ve not met a single person with even a passing interest in boatbuilding, let alone an actual boatbuilder, etc). Second, even if they consider that someone else thought of it first, people are notoriously bad at searching.
Hosting a personal instance is the direction I want to go. That’ll be a winter project, though. It’s fishing/swimming/rowing/woodworking season now :)
Are there any general tips or hidden dangers you can share? Or is it as straightforward as just putting my nose into documentation and playing around a bit?
We’ve been watching an old TV series called “One Step Beyond.” I actually like the Alcoa ad that runs ahead of the program. It’s written specifically for the program and runs as an introduction. They use “One Step Beyond” as a phrase highlighting their ability to innovate and in contrast to the “One Step Beyond” our normal existence as portrayed by the upcoming episode.
I know I’ll tire of it eventually, but for now I’m enjoying it much the same way I enjoy listening to a piece of music multiple times or rereading a good book.
That sounds like the kind of thing I envision.
Yes, no personalized data collection. Both sides of the ad transaction would need to track something if the placement had some kind of impressions or click-through payment system. It’s been a while since I’ve managed a website, but I think most of that can be handled with pretty basic logging that has existed since before micro-targeted advertising was even conceived.
For a simple placement contract like we have with what few newspapers remain, the ad supplier could assess the value of the placement for themselves using standard referrer logs. Not paying its way? Don’t renew the placement.
My experimentation suggests that your inbox is kind of read-only. The only way to reply or vote is to go back to the top level thread.
I have some concerns over that. Unless there is something I’m missing, there is no reason for your inbox to be anything other than a filtered view of the communities where you’ve commented. As such, you should be able to do anything in your inbox that you can do anywhere else.
I could be completely out to lunch, though. My programming career ended over a decade ago and I have not done anything that resembles a formal analysis.
There are ways to do advertising that works and is not annoying (or at least less annoying). Context advertising are ads that are directly related to the subject matter of interest. For example, ads from companies that are in the business of meeting the needs of the boatbuilding community would be welcome or at least tolerated in a boatbuilding community. Those same ads shown to a programming community would be less welcome, even if there happens to be significant shared membership.
For example, the paper magazine “Small Craft Advisor” recently transitioned to online only via Substack. It didn’t take long for subscribers to actually complain about the loss of advertising and SCA had to respond with self-promotion articles from former advertisers.
Context advertising requires no user profiling, no user tracking, and no data collection. “Oh, you sell epoxy (or sails or plans)? Well here is a community (as distinct from a user profile) that is likely looking for what you sell and probably already discussing products in your line of business.”
Put BBS discussion boards, FidoNet groups, and Usenet out front and drop Fark for my path. Along with a variety of standalone forums and random stuff that never went anywhere, of course :) (yes, I’m old!)
This sounds like just standard traffic analysis. Nothing to do with WhatsApp or any other messaging platform. It’s been in use since at least WWII.
Who is talking to whom? How often? Under what circumstances? How do patterns of communication correlate with events? Who are the hubs of communication (ie leaders)?
The big difference between then and now is that instead of needing rooms full of people drawing graphs by hand, there is software to handle it. In turn, that means it’s not really important to have initial suspects to get started, because the computers are quite happy to tease out interesting signals from total communications. That also increases the likelihood of false positives, but the kinds of people who do traffic analysis at this level aren’t usually the kinds of people who worry about a little collateral damage.
It seems like a pretty tall order to construct a system of communication that is useful for coordinating activities, affordable to operate, and secure against traffic analysis. At best, you’ll end up back in a situation where other intelligence will be required to identify a manageable pool of suspects.