Ask HN: Most interesting tech you built for just yourself?

So I made my own. https://www.secretbatcave.co.uk/projects/stock-ticker-machin...

Its not strictly electro mechanical like the original, that was too far out of my mechanical design skills.

As an introvert, I don't like to bother people. When I'm in a quiet coffee shop or library, I turn down my phone volume, select a white noise track in Apple Music, and put my ear near the bottom of the phone.

So, I created the simplest app of my life: open the app, and it plays the sound of waves. If your phone is in silent mode, it won't play anything when the connection fails, even if the volume is high.

I posted it on the web, and many people didn't understand its purpose, thinking it was just another white noise app. It received very few downloads. However, it's the only app I made that I use every day.

Back then the game was very easy to cheat in because even though the map was huge, the client kept all of the objects in memory along with their positions all the time - and it was very important in the game to stay hidden from other players and to hide your stashes of objects for later use. All of these was available in memory for grabs and there was no anticheat.

So there were people who wrote cheats that just grabbed the positions of those hidden stashes and bee-lined through all of them, robbing them.

The server logged the position of static objects (like stashes) on startup and logged the position of every player every few minutes. So I wrote a very simple application that parses that log and puts all of that information (position of stashes and players) on a human-readable map. The admin could then select a player and track his journey. It was very easy to spot people running in straight lines from stash to stash, it was obvious they were cheating and should be banned.

After that I added some heuristics that detected these behaviours automatically and gave hints to admin on who to check. There were more abuses possible in the game later on that I also detected.

For example, there was a "dupe bug" which allowed a player to duplicate a backpack full of useful items and give it to their friend. It involved two players staying in the same spot, dropping the backpack on the ground, trying to open it up by two people at the same time, one of them disconnecting, etc. The backpack was duped due to lag on the database on server side.

I modded the server files to log the information that a backpack was dropped or picked up (along with a list of items inside it in order they were arranged). Then I modified my log parser to look for two players being near each other, dropping and picking up the backpack, disconnecting and reconnecting constantly and detecting two backpacks with exactly the same list of items in the same order they were arranged it (which was very unprobable to happen out of itself) - detecting this gave a hint to the admin to check these people out as possible dupers.

I installed linux for the first time in 1992 on my parents DOS machine. I had another partition for it. It was Slackware with kernel 1.2.13 I think. All off 5.25 inch floppy disks.

Setup wasn't so simple then and I was a UNIX Noob so I managed to set the swap partition to the DOS hard drive and overwrote the first 4MB or so.

The FAT filesystem's root directory and many others were blanked but not all files were lost. Norton tools and CHKDSK managed to get a lot of files back but many of the wordperfect documents were in the form "FILE0001.CHK" and no way to know what was in each one other than very laboriously opening all of them and trying to work it out from the contents.

Very fortunately I had an old backup but the problem was to know, out of all the recovered files, which were covered by the backup and which were new since the backup. If I could ignore the files that I could restore from the backup then I only had to load and rename the ones that were new.

CHKDSK couldn't recover the file size since that was in the destroyed directories. So you couldn't guess if some backed up file matched a restored one just by looking at size.

In the end I wrote some perl+shell to get the md5 of the first kilobyte or two of all the backup files and all the recovered files. I used this to match files and get a list of all the recovered files with no corresponding match in the backup. These had to be new files and since there were far less of these I could manually load up each one into Wordperfect, see what it was and give it a sensible name.

This program (don't have it anymore) saved my bacon and served no-one else but me. It took me from despair to triumph and that's why I like it so much.

That means something more sophisticated has to exist today and should be commercially available. Can anyone explain to what extent companies use this stuff in their interaction with customers, and how successful is it? (Somehow I still see AI still as one-off things people do for fun or AI being used to hype up rather mundane software.)