I worked in Identity Management for a University at one point in my life. There are several things wrong going on here.
Looks like he had a peaceful termination (contract just ended) so the systems locked him out slowly. That seems right, but the fact that this was irreversible is insane. Contractors came and went at the University, as well as associate professors, adjuncts, etc. We had a pretty complex system for ensuring no person got two accounts. If anyone left, the account is always disabled and always kept.
Even if they are just a student and graduated 8 years ago, so long as HR gets the same SSN and DOB as their old student account, we just reactivated their old student account (although it would have had student removed and alumni added) and add a staff or faculty role to it. If a new account had the same name and birthday of another account, but different SSN, it’d get flagged for us to look at it manually (we had rules for similar names, reversed months/days on birthdays and other rules to try to catch things). Sometimes we’d find someone got two accounts by accident and we’d have to go through the painful process of trying to merge them.
If someone was fired, we used a “kill bit,” which would lock them out of nearly everything in less than an hour. It was something we were very careful with, because if that happened, you lost your SIDs and all your roles on most of the systems. That was a bitch to reverse if we needed to, but it was still doable (although you’d probably be requesting all the permissions you needed for about a week).
We used Novell’s IDM to push things downstream, but all the actual identity management integrations and services we wrote ourselves, and we’d always be able to reverse a situation like this in less than a day.
I’m glad the author left that shop. Big companies have so much garbage like this; probably a bunch of off-the-shelf garbage where they handle none of the integration work and have no real IDM team to deal with accounts. There was no reason for this to happen. They should have paid him for those weeks off. He’d have a good case if he went to a labor lawyer. The company would probably just settle.
I wasn’t sure if practices is the best tag for this. Overall I think it’s relevant to remind us not to over-automate things and always leave enough options for manual intervention. Perhaps https://lobste.rs/s/5frjuu/youtube_blocks_blender_videos_worldwide also was a result of too “much” automation?
It’s ridiculous that no one could figure out what was happening.
It’s not ridiculous, it’s a well known phenomenon dismissively called “the paradox of automation” (not a paradox at all, just a fundamental UI law: users are humans).
After your suggestion I added the AI tag. Later user suggestions lead to an automatic removal again.
I think a case can be made for either variant: On one hand the story itself does not involve AI. On the other hand AI surely is a field where blind trust in the results can throw people into similar or worse situations.
You can see that the AI tag is appropriate here, if you know what an expert system is.
The software terminated the employment without human intervention, and then proceeded to block his badge. sent mails to get him escorted out of the building, and so on…
Given that no reason has been given in the comments, I can only speculate about their reasoning.
Maybe any criticism to an AI application faces strong opposition from interested developers.
The same removal occurred to an article about facial recognition (an application of computer vision) that I submitted recently: despite being evidently on-topic, the AI tag was removed by users’ suggestion.
Probably a poor-man SEO to minimize ethical and/or political reflection about the topic.
Or maybe they are just “last-minute AI experts” who genuinely ignore all the past research…
This person was an IT contractor,. not an employee
They weren’t “fired” per se; their contract simply wasn’t renewed due to a silly oversight and bad processes
The “machine” had nothing to do with the termination, except for implementing some apparently-naive ticket scripting to revoke access
None of this is shocking to me EXCEPT it’s unconscionable his staffing company (and “recruiter” as he refers to them) didn’t call to say plainly “your contract was sadly not renewed, do not go to work as of date X until a new contract may or may not be signed”. The fact his staffing company and host manager either didn’t know or didn’t tell him, and allowed him into the building to do work after the contract was up, are very problematic and could be legally actionable.
This reads like a proper sci-fi short story and I suspect that a LOT of it was highly embellished. It makes sense to me that his direct managers would be confused as to what was going on, but it does not make any sense that it took them three weeks to figure it out. “The thing that I’m responsible for is on fire and my contractor can’t access anything” is a managerial complaint that should make its way up to the fucking CEO in a matter of hours if all anyone else wants to do is point fingers. So either important details were omitted to make it a better story or this big business is exceptionally bad at being a business.
I think you have too much faith in the efficiency of business. There are many reasons people might choose not to escalate, for example - even if it doesn’t help the business. Most managers are in the business of protecting their little patch and their own career prospects, nothing else. Unfortunately, I found the story quite plausible (although it could be exaggerated as you say).
You’re probably right. I’ve had only 10 months experience in a truly bureaucratic business myself (not counting the military which is actually suprisingly efficient, just slow) and that wasn’t even a big company. I do find it plausible, just the details far-fetched.
willing to bet the recent issue with YouTube Piracy filter blocking MIT courses and the Blender Foundation are the result of The Machine being the ultimate decider.
Yeah, I suppose he’s right. I can imagine someone executing a query to tell YT about net value for particular videos and being concerned about some outliers. “zero-point-eight percent of our traffic comes from videos that the participant hasn’t opted in to any monetization. It’s in our T&C that we can make this a requirement. Let’s just do it — we are subsidizing these videos. These users probably just neglected to opt-in and some of their content ‘went viral’.”
This kind of conversation would sound pretty rational and maaaaaaaybe someone would pipe up with “but what if they didn’t forget to opt-in, they intentionally didn’t opt-in?” If anyone thought that, it was likely dismissed pretty quickly.
I worked in Identity Management for a University at one point in my life. There are several things wrong going on here.
Looks like he had a peaceful termination (contract just ended) so the systems locked him out slowly. That seems right, but the fact that this was irreversible is insane. Contractors came and went at the University, as well as associate professors, adjuncts, etc. We had a pretty complex system for ensuring no person got two accounts. If anyone left, the account is always disabled and always kept.
Even if they are just a student and graduated 8 years ago, so long as HR gets the same SSN and DOB as their old student account, we just reactivated their old student account (although it would have had student removed and alumni added) and add a staff or faculty role to it. If a new account had the same name and birthday of another account, but different SSN, it’d get flagged for us to look at it manually (we had rules for similar names, reversed months/days on birthdays and other rules to try to catch things). Sometimes we’d find someone got two accounts by accident and we’d have to go through the painful process of trying to merge them.
If someone was fired, we used a “kill bit,” which would lock them out of nearly everything in less than an hour. It was something we were very careful with, because if that happened, you lost your SIDs and all your roles on most of the systems. That was a bitch to reverse if we needed to, but it was still doable (although you’d probably be requesting all the permissions you needed for about a week).
We used Novell’s IDM to push things downstream, but all the actual identity management integrations and services we wrote ourselves, and we’d always be able to reverse a situation like this in less than a day.
I’m glad the author left that shop. Big companies have so much garbage like this; probably a bunch of off-the-shelf garbage where they handle none of the integration work and have no real IDM team to deal with accounts. There was no reason for this to happen. They should have paid him for those weeks off. He’d have a good case if he went to a labor lawyer. The company would probably just settle.
I wasn’t sure if practices is the best tag for this. Overall I think it’s relevant to remind us not to over-automate things and always leave enough options for manual intervention. Perhaps https://lobste.rs/s/5frjuu/youtube_blocks_blender_videos_worldwide also was a result of too “much” automation?
Manual intervention yes, but observability the most. It’s ridiculous that no one could figure out what was happening.
It’s not ridiculous, it’s a well known phenomenon dismissively called “the paradox of automation” (not a paradox at all, just a fundamental UI law: users are humans).
I suggest the AI tag for this.
It looks like what happened at some point in Idiocracy.
After your suggestion I added the AI tag. Later user suggestions lead to an automatic removal again.
I think a case can be made for either variant: On one hand the story itself does not involve AI. On the other hand AI surely is a field where blind trust in the results can throw people into similar or worse situations.
You can see that the AI tag is appropriate here, if you know what an expert system is.
The software terminated the employment without human intervention, and then proceeded to block his badge. sent mails to get him escorted out of the building, and so on…
Given that no reason has been given in the comments, I can only speculate about their reasoning.
Maybe any criticism to an AI application faces strong opposition from interested developers.
The same removal occurred to an article about facial recognition (an application of computer vision) that I submitted recently: despite being evidently on-topic, the AI tag was removed by users’ suggestion.
Probably a poor-man SEO to minimize ethical and/or political reflection about the topic.
Or maybe they are just “last-minute AI experts” who genuinely ignore all the past research…
Who knows?
We can expect a lot more of this, especially in new companies which never had a human middle management layer.
Some facts that remove all the sensationalism:
None of this is shocking to me EXCEPT it’s unconscionable his staffing company (and “recruiter” as he refers to them) didn’t call to say plainly “your contract was sadly not renewed, do not go to work as of date X until a new contract may or may not be signed”. The fact his staffing company and host manager either didn’t know or didn’t tell him, and allowed him into the building to do work after the contract was up, are very problematic and could be legally actionable.
This reads like a proper sci-fi short story and I suspect that a LOT of it was highly embellished. It makes sense to me that his direct managers would be confused as to what was going on, but it does not make any sense that it took them three weeks to figure it out. “The thing that I’m responsible for is on fire and my contractor can’t access anything” is a managerial complaint that should make its way up to the fucking CEO in a matter of hours if all anyone else wants to do is point fingers. So either important details were omitted to make it a better story or this big business is exceptionally bad at being a business.
I think you have too much faith in the efficiency of business. There are many reasons people might choose not to escalate, for example - even if it doesn’t help the business. Most managers are in the business of protecting their little patch and their own career prospects, nothing else. Unfortunately, I found the story quite plausible (although it could be exaggerated as you say).
You’re probably right. I’ve had only 10 months experience in a truly bureaucratic business myself (not counting the military which is actually suprisingly efficient, just slow) and that wasn’t even a big company. I do find it plausible, just the details far-fetched.
Yeah, I suppose he’s right. I can imagine someone executing a query to tell YT about net value for particular videos and being concerned about some outliers. “zero-point-eight percent of our traffic comes from videos that the participant hasn’t opted in to any monetization. It’s in our T&C that we can make this a requirement. Let’s just do it — we are subsidizing these videos. These users probably just neglected to opt-in and some of their content ‘went viral’.”
This kind of conversation would sound pretty rational and maaaaaaaybe someone would pipe up with “but what if they didn’t forget to opt-in, they intentionally didn’t opt-in?” If anyone thought that, it was likely dismissed pretty quickly.