I think it's straight forward. They made a decision to pay top dollar because they had ambitious plans, and wanted the silicon valley types.
All went well.
Then, as this part of the company grew, some bean counter decided it was a huge expense, and something had to be done.
I suspect these walmart labs people were costing triple the standard walmart webdev, and so to the bean counters, the path forward was obvious.
It's really unfortunate when non tech people make decisions like this. I've worked at a FAANG for 10 years, and before that was at HP and other mid-sized companies. HP's average principal engineer would be outperformed by our interns.
This doesn't make sense given what the person you're responding to is claiming. If several people over several non-overlapping time periods have the same positive experience then decline, what you're stating is that the company is having memory loss every few years, decides to go for the pros, then remembers they didn't actually like doing that.
Maybe that is what's happening, and management is just cycling through the same way the engineers are. Walmart does have plenty of money to relearn lessons every few years, but I also would be surprised for the same reason - they didn't find billions of dollars under a rock. They're good at making money. Making the same mistake with personnel repeatedly is not a good way to make money (or maybe it is, what do I know).
As an employee, you are generally evaluated based on your perceived impact, at any level up to and including the CEO. This is easily twisted into change for the sake of change.
For a software engineer this does lead to some pear shaped decision making, like adopting new UI frameworks every few years, or whatever the case may be but these kinds of things are overall pretty benign compared to the same problem in the management of the business.
In the management of the business, the correct decision may be to stay the course on something a lot of outsiders and pundits and new grads want to pot shot and second guess like what industry you should even be in or how you should structure the company itself.
Let's say you are Visa and your transaction processing runs on IBM mainframes as an example. Everything is working with known parameters and risks, has a long and predictable roadmap, whatever. Being the guy that says "ok we are going to keep doing this for 10 years and evaluate again periodically if needed but this is the plan of record" takes massive guts, and should be paid at least as well as the guy that says "throw everything out and do this risky untested thing instead" but very few managements actually work like that.
The same waffling happens with remote vs RTO and either having the guts to make a particular stand or kowtowing to what you preceive to be the popular/prevailing opinion one ought to have as a CEO at this moment.
It can also lead directly to the situation you are describing where a decision keeps getting remade, perhaps even in a flip flop loop, to the benefit of multiple generations of "decision makers".
I appreciate the added context. I agree, this kind of flip-floppy, turbulent thing does of course happen. My whole comment was poorly written. I was mostly trying to say that it seems unlikely that some time ago someone said "we should hire people and treat them well for a while then treat them poorly and do it all over again."
Except, I don't even really agree with that. That's how companies treat employees all the time. Not just in software, but floor workers and warehouse folk and anything else.
Yeah and my comment is addressing some of the broader motif of the thread and article since I didn't want to leave multiple. Addressing your hypothetical thought quotation the "we should hire people..." I agree that it would rarely to never go down like that. Instead, the impact on people's lives and livelihoods is collateral damage to the need to be perceived as a change maker or in charge or whatever. The fact that it is repeating is just an artifact of people being rewarded to retread the same ground because institutional memory is for whatever reason not in the control system's feedback design.
Then, as this part of the company grew, some bean counter decided it was a huge expense, and something had to be done.
I suspect these walmart labs people were costing triple the standard walmart webdev, and so to the bean counters, the path forward was obvious.
It's really unfortunate when non tech people make decisions like this. I've worked at a FAANG for 10 years, and before that was at HP and other mid-sized companies. HP's average principal engineer would be outperformed by our interns.