Amazon’s Latest Experiment: Retraining Its Work Force | Hacker News Reader

Amazon’s Latest Experiment: Retraining Its Work Force

nytimes.com

73 pointsartsandsci70 comments

I wonder what the success rate is for their retraining program to turn factory workers into coders.

When watching Andrew Yang's Presidential pitch for UBI he said most government retraining programs only have a success rate of 17%.

17% strikes me as not bad. Not great, but not awful. People who take jobs that are at eminent risk of automation probably weren't exceptionally diligent in school the first time around. Add in the burdens of adulthood (parenting, bills, etc) and the odds of staying in a retraining program seem bleak.

Does anyone have a better success rate?

17% sounds good to me. I did co-op at Elections Ontario and the place is just chock full of barely employable people who resist change to the status quo.

You poor bastard. I worked for the federal government too.

I remember being told that working for the government for more than a few years would make me unemployable in the private sector. The idea was one's work ethic would be irremediably damaged. I thought that was foolish at the start of my time there, but a year later I had a new respect for the heuristic.

https://news.ycombinator.com/item?id=20419574

Amazon doesn't care about the success rate. They just need positive spin they can point to for any layoffs. "See, we had these retraining programs in place for our recently redundant workers."

> I wonder what the success rate is for their retraining program to turn factory workers into coders.

Probably about 0% for their warehouse workers: they're so exhausted by the end of their shifts that they won't have the energy to learn, and their most at risk of losing their jobs to automation.

This is a feel-good announcement that will likely do little good without other management changes that Amazon is probably loathe to make.

1. Is it really too much to believe that someone with a physically demanding job might be in good enough condition to do something else after work? I'm pretty sure folks in all sorts of physical jobs have physical hobbies as well. You never hear this sort of excuse about carpenters.

2. It's hard to work full time and go to school after. It's by definition going above and beyond.

"Excuse"? You think people are being too lenient about the warehouse workers? Those jobs are awful, and it's not about simply being a physical job. I would much rather do carpentry.

There are a lot of awful jobs. Roofer in Phoenix in the summer, for example.

No, I don't think people are being too lenient -- it's a bit more subtle than that. There are many physically demanding jobs, some as much or more so than warehouse worker. I was questioning why we never see people say people in those job families couldn't possibly study after a day at work.

What you want to do is also irrelevant, as you might have an enjoyment of carpentry. Me, I hate it, and would rather be a warehouse worker.

Amazon warehouse jobs aren't just physically demanding. I agree that one can study hard after half a day spent on some physical effort. Instead the problem is that Amazon pickers are under constant stress all those hours to meet targets far more demanding than most other jobs. I would expect that constant mental stress to leave people so frazzled at day's end that they would be unable to hit the books and concentrate.

> far more demanding than most other jobs

And you've just made my point for me. "Most" other jobs. It's not the most difficult, most physical, most dangerous. Where are the people saying people in job "X" can't possibly study after a day at work? Somehow, that reasoning is confined to warehouse workers.

I worked harvest and was a janitor in college and still had energy to study. I had no time to do anything else but it was strangely “easy” at the time because I had no distractions.

Fair enough, but not all people have that experience. Things often get harder as you age.

Sadly, some people make poor decisions when they are young and course correcting gets harder the older one gets.

Youth helped me for sure but I did it later in life than most people.

I was motivated because I believed my effort would pay off. That may be easier without the experience of age but I don’t think anyone is excluded from that possibility.

I have a brother in law going through this process. Having young kids seems to be enough motivation to work more after brutal overnight shifts. He recently immigrated to the US.

>Probably about 0% for their warehouse workers: they're so exhausted by the end of their shifts that they won't have the energy to learn, and their most at risk of losing their jobs to automation.

I work at one of those warehouses and people do try to put in the effort to learn - unfortunately, Amazon only offers Associates' Degree level education and (judging from public complaints) all but forces people to work night shifts and extra hours to keep their numbers up, and doesn't actually teach the classes well, because of course their first, primary and overriding interest is in having their warehouse employees meet their quotas and fulfill business needs.

Which is why I recently just suggested that people just take Udemy courses and stay away from Amazon's offers altogether.

There are a bunch of different programs targeting different cohorts. Associate2Tech and AmazonCareer choice are targeting the fulfillment center workers.

ATA is incredibly successful at small scale. Biggest challenge is scaling it up + making it work for everyone. If you are familiar with amazon's think big and pr faqs, educated guesses can be made as to the end vision for that program.

The NYT article is actually pretty poor compared to https://www.geekwire.com/2019/amazon-spend-700m-retrain-100k... or https://press.aboutamazon.com/news-releases/news-release-det... because it misses the apprenticeship which is by far the most interesting program amazon is running since it's targeting the veterans.

i wonder if its cheaper to retrain vs outsourcing

My job takes me to many Amazon FCs. The job families that we currently offer through career development classes are usually things like CDL, operations management, dentistry, EMT, process engineering, mechanical tech, etc. It is really a grab bag of what is most highly requested by the diverse working cohort at each FC.

Code as HN would think of it is not as common, but associates becoming IT or facilities technicians or engineers(remember, these things are teeming with robotics and conveyor systems) is highly common.

I’m glad that we are spending more on this.

> dentistry

can you add some detail here? Are the career development classes intended to prepare people for life outside Amazon? Or does Amazon have company dentists?

There is a shortage of dental hygenists in the Seattle area right now....

Perhaps too many small dental practices though (or there is a massively underserved need? There seems to be a new dental practice every week in Seattle...).

I've complained numerous times at my FC that none of the career programs are actually careers, and that none of the degree programs offer anything higher than an associates degree. FC employees with any technical expertise and a desire to retrain somewhere in the Amazon network beyond the FC system seem to be SOL. Of course, the situation might differ between facilities.

And I work at a FC near the second biggest tech hub in the US... it would be easy pickings for Amazon (pun intended) but it seems like no dice. But if I want a second associates' degree and to restart my career as a truck driver or paralegal at 40, I can do that.

Actually, he says it's 0-15 percent, which I've always assumed is just the granularity of the study to which he refers.

Alternatively, since those retraining programs were focused on specific industries or occupations in manufacturing in specific regions, maybe it's the range of results in those occupations/regions.

Never heard him say specifically 17 percent (and I have pretty much watched everything Andrew Yang related ;)

My first job out of school 25 years ago they were big on further education and training. You actually had in your yearly objectives that you complete atleast 10 days of training. There was a dedicated campus for this purpose. And these were indepth courses that would spend 5 full days on some skills. I don't see that kind of thing any more and miss it.

The company I work for mandates a certain number of classes per employee each year and has a list of about 30 others that are optional.

Some are as basic as Advanced Microsoft Word. Some are upper level management stuff, but are open to any employee in any role all the way down to the security guards.

A cooking class is offered every other month. I always take those.

Maybe unpopular opinion, but I welcome this.

The landscape is changing and Amazon is willing to try and put serious effort into keeping its current employees instead of just letting them be laid off only to be left behind.

Amazon revenue in 2018 grew 30% over 2017, but headcount grew only 15%, which is quite an increase in efficiency per employee. That doesn't really make the headlines since there is still overall headcount growth.

As they get larger it is harder to grow revenue that fast percentage-wise. If revenue growth rates drop below ~15%, and that rate of efficiency gains continue, then it would lead to net headcount reductions, which will likely generate much less positive press. So this seems like a good way to get ahead of that soon to be upcoming bad press. Whether it is just PR spin or a genuine effort to mitigate this is I guess open for debate.

>>Amazon revenue in 2018 grew 30% over 2017, but headcount grew only 15%

Amazon outsources a ton of work,for ex. To it's 3rd party merchants.

So it's a bit harder to measure productivity.

I am waiting for a day when “machine learning scientists” are the blue collar workers.

Blue collar machine learning jobs center around data labeling: https://www.techrepublic.com/article/is-data-labeling-the-ne...

I don't ever see the actual modeling work around AI becoming blue collar. It's complex and worth the price to the companies that can build a business around it. The data labeling, however, can be pretty simple.

I suspect even the high-end labeling (e.g. cancer) can become blue collar with appropriate training, given adequate outcomes data to compare labelers against.

"high-end labeling" for cancer is often done by undergraduates making $0-15/hr after a couple hours of training

That sounds like an academic lab working from grant funding. Did the FDA approve those models for clinical use?

> That sounds like an academic lab working from grant funding.

Yes

In a way that's already happened, though maybe not in the way youre thinking of. Machine learning data is more and more being collected by volunteers or paid content curators. These jobs are generally low pay and in foreign countries. There's a huge outsourcing of "blue collar" mechanical Turks going on right now that is essential to most supervised learning methods.

I misread "retraining" as "retaining". My first thought was it must be a joke or I misunderstood it.

It kind of bothers me that there's always an assumption on HN that we need to retrain people into coders instead of any one of the hundreds of other jobs in high demand. You'd probably have a lot better results retraining warehouse workers into plumbers, electricians, and HVAC installers, for example. Or on the corporate side, QA testers, entry level IT support, etc.

The article only mentions software engineering courses, so I’m not sure it’s a result of this being hacker news as much as what the article states.

We would have better success of giving them a choice instead of picking just one.

People work better when they're passionate about what they're working on.

The best way to control your flock is to let them roam free.

True, while I think that's what Amazon should do, I can understand why HN like to talk about turning everyone into a dev.

It is not just coding. This WSJ article has some more detail:

https://www.wsj.com/articles/amazon-to-retrain-a-third-of-it...

The training programs could help Amazon workers find jobs in different industries, the company said. It is expanding a program for fulfillment-center employees called Amazon Career Choice that pays 95% of tuition and fees for certificates and degrees in high-demand fields such as nursing and aircraft mechanics

Thanks. I missed that part in the article.

> QA testers

I am one of these folks. Sometimes I have received stigma about being so. "Oh, get out of that" is one direct quote, and through an expression of disgust.

I still don't know why. You want your widget to explode? Just not turn on? Fail after three days? Shall I tell you just how many dud brand-new flash drives I've tossed over the past 6 days alone for -- I guess -- giggles and jollies and I should just smile and like it? That's what I'd like to "get out of".

We do exploratory tests these days - since value of individual users is just slightly above 0, quality is not exactly high on the radar, in most consumer oriented companies anyway...

Imagine if everyone learned to code... the coal miners, the fireman and the classy janitor, the cashiers with dignity, the migraine workers... then no one would get paid the big bucks for coding.

Sounds like a good outcome to me. Cheaper software is a good thing.

I have colleagues who copy paste security restrictions to all controllers, disable XSS protections when an ampersand doesn't show, proliferate a tree of possibilities they can't fully account for. We will need a brand new generation of tools to reduce consequences of bad code.

I agree, but also keep in mind that bad code is often still better than no code. There are still many many low hanging fruits for automation left that are not worthwhile to tackle while programmers cost $100+ per hour and the humans that currently run those tasks make mistakes too.

There are programs in the failing coal towns to teach former coal miners how to code. I'm not sure the stats on the success rate though.

Extremely low: https://m.dailykos.com/stories/2019/5/23/1859813/-Students-C...

huh. who'd of guessed that coal miners can't code?

A lot of people would have guessed that a program running mostly on warm fuzzies and hope wouldn't be good at producing skilled workers.

I suspect that most laymen find our line of work to be kind of trivial and think that success in this field is more related to youth than intelligence. To them, "coding" is more akin to carpentry (a trade) than physics (a science).

Also, pointing out that not every coal miner can learn to code is seen as culturally insensitive and perpetuating the stereotype that rural people are dumb. Which I understand, but a little honesty could have seen these resources better leveraged to help these people in different ways.

Your suspicions are exactly opposite from what I can tell talking to a number of blue-collar people. A lot of the ones I’ve talked to seem very apprehensive about learning anything technical involving a computer, and it all seems kind of mystical in there eyes. Honestly I think a lot of what holds people back is a lack of confidence and a willingness to learn more than anything else.

I wouldn't say it's an unwillingness to learn so much as a paralyzing fear of failure, which they believe (rightfully so) is inherent in the learning process.

Is carpentry really that easy to master? Practical arts all require lots of practice. The only thing that makes coding very different is all the abstraction (so more like machine design than wood assembly).

indeed, I'll stick to coding. Tried a number of times to do some wood work, a simple mitre joint is way harder than you'd think. But then there are few true carpenters left, its mostly low skilled work e.g. framers.

Knew a guy who could do the most precise joins, its a skill people seem to be born with, or its grown somehow. If the only jobs in demand demanded the ability to construct a dovetail joint, I'd be in big trouble.

I used carpentry very deliberately. It's a skill that's seen to require practice, but is still very accessible. Most able-bodied people can build stuff with wood. In fact, most of our grandfathers (assuming American) probably did wood working as a hobby.

It used to be taught to grade school kids. I distinctly remember building adirondack chairs, picnic tables, and benches in wood shop, from dimensional lumber, for use in local parks. Thirteen year old kids can build practical objects using wood, with guidance.

While physics is voodoo. Even genuinely intelligent people have difficulty grasping it.

CS probably lies somewhere in-between, intellectually, but is seen as something that can just be taught to any person and they can be productive.

To be honest, a lot of coding is fairly trivial.

Is that remark necessary?

Honestly, the classicist/entitled bullshit about what we (technologists) do is completely unearned. What most of us do isn’t special and the gate-keeping many people in this industry do to prevent others from entering this industry (and I’m guilty of it too at times) is unearned.

I’m glad Amazon is doing things like this. I’m glad programs are encouraging coal miners to learn a new skill.

Programming wasn’t always seen as white collar work.

I have a pretty good tech job, my title includes the word “engineer” but I don’t think that is deserved. My coworkers seem to view this as a prestigious position but I’m not convinced.

From my perspective this not much different than building pickups in Detroit in the 60s. I bet that seemed like a pretty awesome job back in the glory days.

In reality most technologists are in manufacturing and we will eventually be replaced by robots.

Amen.

I agree. There was a time it was normal for an organization to invest in its employees. Perhaps it's cheaper to externalize the costs of retraining for smaller firms. But perhaps for larger employers it is starting to make economic sense. I hope this becomes a real trend.

Please don't post shallow dismissals here.

https://news.ycombinator.com/newsguidelines.html

Headline might be more clear as 'Amazon to Retrain a Third of its U.S. Workers...'

We've updated the headline to that of the article. NYT headlines often change, so I assume that happened here.

I definitely read that as "retain."

Reply on news.ycombinator.com