Re:20k federal workers take "buyout" so far, official says
by tombert on Wednesday February 05, 2025 @02:07PM
(#42953499)
> Efficiency does have a sensible definition in isolation: an efficient system is one in which there are no more Pareto improvements. From the government perspective, you could consider this in terms of the cost/service frontier, and actions which move the government towards that frontier improve efficiency.
I had to look up Pareto improvement, and I don't think that many systems exist where improvements are better in "every way" like the definition suggests.
For example, most big tech companies will use something like Kubernetes (or an equivalent) to deploy multiple instances of a service and then load balance between them, even when none of them have reached full capacity.
In one sense, this is "inefficient", in that we have computers sitting idle that could be doing work, but it's also efficient in another sense, which is to minimize downtime; if one of the computers or service crash, you won't experience much (or any) downtime because one of the replicas will handle requests.
Someone could get rid of all the replicas and then claim victory in that they made things "more efficient" by reducing the cost, but that will come at the expense of possible downtime.
Every system is different, and figuring out where on the spectrum you draw that line is rarely clear-cut, and there is almost always tradeoffs no matter what decision you make. Sometimes you can live with more downtime and it's better to cut the costs of the extra computers, sometimes downtime is not an option and you need a ton of redundancy.
Humans are not computers, so it's not the best analogy in the world, but I think it still mostly holds; it looks like in a hand-wavey way they're just defining efficiency as "spending less money", but that really doesn't make any sense unless you can show that you're getting comparable results while spending less cash.
I'm not saying that the government is perfect with spending money, obviously, and it's entirely possible that DOGE will find something that really should be eliminated. I'm just saying that it's rarely cut and dry, and it's very rare that cutting something is just a universal net improvement.
I had to look up Pareto improvement, and I don't think that many systems exist where improvements are better in "every way" like the definition suggests.
For example, most big tech companies will use something like Kubernetes (or an equivalent) to deploy multiple instances of a service and then load balance between them, even when none of them have reached full capacity.
In one sense, this is "inefficient", in that we have computers sitting idle that could be doing work, but it's also efficient in another sense, which is to minimize downtime; if one of the computers or service crash, you won't experience much (or any) downtime because one of the replicas will handle requests.
Someone could get rid of all the replicas and then claim victory in that they made things "more efficient" by reducing the cost, but that will come at the expense of possible downtime.
Every system is different, and figuring out where on the spectrum you draw that line is rarely clear-cut, and there is almost always tradeoffs no matter what decision you make. Sometimes you can live with more downtime and it's better to cut the costs of the extra computers, sometimes downtime is not an option and you need a ton of redundancy.
Humans are not computers, so it's not the best analogy in the world, but I think it still mostly holds; it looks like in a hand-wavey way they're just defining efficiency as "spending less money", but that really doesn't make any sense unless you can show that you're getting comparable results while spending less cash.
I'm not saying that the government is perfect with spending money, obviously, and it's entirely possible that DOGE will find something that really should be eliminated. I'm just saying that it's rarely cut and dry, and it's very rare that cutting something is just a universal net improvement.