typically in that case the boomerang is they come back but as a consultant. just doing the same thing they were doing before but at a different rate. usually higher to cover health ins plus a bit of a bonus.
Honest question how many of that 2k are just the same webpages but just different days?
I personally am a close the whole browser out at the end of the day guy. With bookmarks for things I just have not got to yet. With a every few months cleanup of the 'to do' folder. Just today I have probably opened 20 tabs and closed 15 so far. At the end of the day they should all be gone.
> I feel like a good chunk of my skills that I honed over decades as a software engineer are obsolete: programming, debugging, navigating code.
I have been having a different exp with it. Those are even more important. These things are creating some wild code. Then sticking to its guns in some cases and just rewriting the same bug over and over.
Just this week I literally showed the thing where it was breaking. I wanted it to trace where the error was coming from. Instead it immediately wrote guard code and patted itself on the back. Not a bad idea but does not give me the reason of why something was broken. As where it broke is just part of the problem. Turns out I had a data issue. Guarding the error was the wrong plan in what I was looking for. The user needed his data fixed. Without tracing and debugging I am pretty sure AI would not have found it. Not without me prompting even more. It had twisted itself silly trying to make the error good instead of root cause. Took me an hour or so before I realized it was doing that.
Trust by verify. These things make all sorts of assumptions unless you are very accurate in your prompts.
1. Are you using the latest frontier models? I know that feeling: "These things are creating some wild code. Then sticking to its guns in some cases and just rewriting the same bug over and over." But that was a few months ago. With Opus 5.5, it's gone.
2. For debugging, I give Claude read-only access to the prod DB, and it has access to logs and traces (Signoz in our case). With this, it has amazing debugging skills. Orders of magnitude faster than myself, and it has yet to fail finding the cause of an issue.
> This shows that "know your target" is still as applicable as ever,
totally. I spent yesterday trying to get an AI to make a test case for a very specific bug I had. It kept telling me that it had it. When I could clearly trace thru the code and see it did not. After about 5-10 rounds of that I just wrote the test myself. But at least AI helped with all of the dumb cruft that went around it.
But if I did not know the domain I may have easily just said 'yep looks good'. AI on the other hand saved me at least 3-4 hours of writing all of the scaffolding that made it easier for me to check. In the end the bug had nothing to do with the test. AI did not find that. I went back point by point of what changed in the code and found where I had messed something 2 weeks ago.
A few months ago it seemed like it would not do this. Now in the past week I have come across 3 different ones that did something similar. Just 'stuck' in a local maxima and getting the AI to go around it is a pain. This is probably me and the way I am using it.
Is this better than I had a couple of years ago? I am not sure. Seems different at least.
More than likely nmap type scanning across all devices on a segment. I had one customer that saturated our AAAA server with requests because they though sending in data on every even was just dandy. For one device in a test lap it was ok. For 10k+ of the things out in restuant across the country was a DDOS. That was all text. it was gigs of data per hour. We sat them down and showed them how to do bundling and better data management.
It is a question many devs miss. what happens when I put this in the wild and I now have hundreds of them.
In this case it is probably scanning aggressively. Just incase you plugged in a new devices in the past 100ms. Then some sort of filter with data compression (hopefully) and then sending its exfiltrated data back home. Honestly, not a cool thing to do and makes your customers NOT trust you. I had an old LG TV that was doing this 15 years ago where it would send every button press back to LG. I unplugged it. It agressivly scan its empty socket.
Also many had fun easter eggs. One of my favorites was galaxy quest. You had to watch the movie then after you watched it there was a new option to watch the movie in the alien dialect. Now some menus are just 'in the way'. Usually some blurb about whatever concerned some lawyer and a good 15-30 second block on continuing on, or even worse a set of commercials for other movies that are unskippable. The newer menus are very basic and straight to the point. The older ones were 'fun' in a way that usually mimicked whatever the movie was doing. Adding to the movie. I think one of the fight club releases even had the 'wrong' menu just to mess with people.
I had one customer that managed to figure out which stored procs were tied to which reports we were using. Then he would deliberately break it in weird subtle ways them and call tech support 'just to see how we would react'.
I have ruined my fun of many games by activating cheat modes. I strictly avoid them now until I have felt I have 'completed' the game to my satisfaction.
> There isn't ever a good reason to have an infinite loop
That seems to be a very broad statement. For example in a system where interrupts mostly control things this sort of 'do not close the program' could be useful.
A guy I worked with had one I never would think of because I do not work in that field.
I would not bet on that assumption. I have seen some wild code over the years from devs. With 'ai' type coding going on now too you may see them be used even more.
Opus and fable both are pretty judicious about bringing in dependencies, at least for me. They often argue against and and write even decently large modules to avoid pulling stuff in.
They never install packages for me and love handrolling large amounts of e.g. parsing code where a library exists. I have to keep telling them 'look for a large popular dependency' when they start writing huge functions that obviously already exist.
Was doing something with OSM the other day and Opus basically started reimplementing NetTopologySuite.
I would've actually thought AI would slightly improve upon this situation. At least in my experience claude seems to write a lot more little utility functions itself rather than reaching for a package from npm to do something. Requiring an `npm install` before getting something working risks triggering a permissions gate.
reply