Hacker News Reader

Comment by one2know | Hacker News Reader

1. Amazon is revenue focused

2. 3% shrink is accepted in retail, that is 6+ billion dollars they are ok with losing.

3. Data is large and hard to get at Amazon. Just getting access to a database without direct VP help is sometimes impossible

4. Compute resources to 'run a report' on terabytes of data are even harder to get.

5. Manager changes every six months

6. Devs last ~18 months so no one is there long enough to care about solving a cost issue.

7. There is NO REVENUE in catching thieves, so there is little to no exec sponsorship

8. Often management is involved in fraud or theft rings.

View on HN

While 3 and 4 may be true currently, it's a lame excuse. Amazon is the probably the largest provider of services like that at the moment. I think it's more than terabytes of data, but even then, it should be trivial to do for them - https://aws.amazon.com/emr/ If they're ok with losing 6+ billion, then what's the cost of 5 engineers and some large instances in comparison?

So does it really boil down to - VP doesn't care enough?

> 6. Devs last ~18 months so no one is there long enough to care about solving a cost issue.

Are you saying devs set project priorities and cannot be assigned where the work needs to be done?

> 7. There is NO REVENUE in catching thieves

That directly contradicts point 2. Apparently there's a potential for preventing 6+ billion dollars of lost revenue.

From the article:

> $428,000 fencing stolen items, much of it on Amazon.

That's enough to put a dev on this problem for a year with required infra and still save some extra money - just from that single pawn shop.

> That directly contradicts point 2. Apparently there's a potential for preventing 6+ billion dollars of lost revenue.

But not in a single budget anywhere in the company.

And, it's all about MY local budget. You're asking ME to put a couple of devs on something that might pay off half a million dollars. Maybe. And, if it doesn't those devs could have been deployed elsewhere.

Uh, yeah, no. Not on my watch. I'm going to put those devs on something that might get me a bonus or promoted, thanks.

I'm sorry if that's really your view. There are both divisions in corps and people who think beyond a single bonus. I wish we started calling this out as "culture at Amazon enables thieves" rather than - here's a list of excuses why they can't do it.

At the very least, the higher ups with the view of the overall budget have the power to start projects like that which assign the budget as necessary.

> At the very least, the higher ups with the view of the overall budget have the power to start projects like that which assign the budget as necessary.

Except they won't because if it fails, their name is associated with it and they will get punished.

Most politics in big companies is about avoiding sins of commission as they are easy to spot--sins of omission are a lot tougher to identify and punish.

The "they will get punished" is a bit extreme. Big corps have both teams trying new things and maintaining existing ones. The proportions vary widely, but things get tired and fail all the time. If people get fired over trying something, that would make a very dysfunctional company. The budget for validating if they can stop some stolen deliveries would not be worth mentioning in most meetings.

Although I haven't been in Amazon specifically, in my experience in corp, the worst that happens to bad mid-level managers is that they're moved to a different project. And that's actually overall bad managers, not someone trying out an idea.

Thinking like this at Amazon gets you stalled at the L6 (manager 2) level. I can think of about 5 managers I've dealt with who acted like this in the past 4 years. As a senior tech person I was asked for feedback on them and gave something to the equivalent of the motivations you listed above. They were all managed out quickly.

By the time management is asking me for my opinion there is usually a worry about a person and I'm just providing data.

Also saving the company $6bn with "one wierd trick" ticks all the amamazon promo boxes and usually skyrockets you.

> They were all managed out quickly.

What? Why? Putting devs on something that will provide benefit rather than might is hardly a firing offense. I simply don't believe you.

> Also saving the company $6bn with "one wierd trick" ticks all the amamazon promo boxes and usually skyrockets you.

IF you can get it implemented. Something which crosses political boundaries is ferociously hard to get implemented unless there is credit that can be claimed by all involved.

managed out doesn't necessarily mean fired, probably demoted or moved into less important positions.

> What? Why? Putting devs on something that will provide benefit rather than might is hardly a firing offense. I simply don't believe you.

I don’t work at Amazon but failing to see past your own team’s mission and missing your role in the larger company is a (common) failure of management.

7 doesn't contradict 2. If you are the loss prevention VP in charge of stopping theft, and you reduced the number of reshipped items, you are decreasing a cost account. The revenue account for sold goods was already increased when the item was sold. The revenue account belongs to the retail group. They look good as long as sales are made. The cost account belongs to the LP group. They start out looking very BAD with huge losses and have to fight to get their costs back to zero.

For this reason large companies might group all their cost groups such as shipping, LP, fraud, etc into one big organization with special protections so they won't be cannibalized by revenue generating groups that have more political power.

> That's enough to put a dev on this problem for a year with required infra and still save some extra money - just from that single pawn shop.

If LP saves 400k, retail doesn't cut them a check for the 400k. They are on fixed budget, usually small because it is considered good money after bad. And at Amazon it is easier for devs and managers just to transfer to a shiny new AWS project than try to reduce costs.

Don't we always read articles about how Amazon is a rational, data-driven entity? Stuff like this makes it seem like Amazon is a normal big retail company like Walmart, and should be valued accordingly.

It's valued as walmart+aws.

Could you explain point 6 a bit more?

What's there to say? Amazon is a revolving door of developers, most of which don't make it to 2 years.

It would take a year or more for a dev to become familiar enough with their domain and its systems to even know to suggest a report that would sniff out this fraud, and even more time for them to have the political capital to get support for the project.

That's what confuses me. I don't understand how anything gets done. It takes a year just to get settled in at a place like that.

There are PE's and people that have been there for years even decades that have lots of power and clout to do things associated with revenue and not get canned on multi-year or multi-month projects. Then there are the thousands of new hires that are just fodder to try to quickly get a select few managers and engineers promoted before they are canned. They essentially don't get anything of value done and do throw away work. A lot of it has to do with perceptions.

Initially you get small tasks on a small project. In a few months you are quite settled in for your position. (Let's say some integration service maintenance.)

Not OP but, from what I've heard it sucks working at Amazon even in engineering.

Amazon doesn’t collect its business data in an instance of its own warehouse product?

You're looking at 4-5 different transactional systems with 1-1000 TB of data. You won't be getting access to those things since they process perhaps $1 MM of revenue per minute. Maybe you can use their ETL system if it is not overloaded. The query to join warehouse, shipping, delivery, orders, catalog probably will run 12 hours and cost $5000 dollars. Executives would rather have one line in their 300 page metrics booklet that says "item cost reshipped to customer suspected theft." If it is under $1 million per day I doubt they would care.

Amazon sells ETL, warehouse, BI, and analytical query tools as solutions to these problems; it’s surprising that they’re not dogfooded.

For all we know they do have this system in place. But shit happens; one weekend some Hadoop nodes lock up, bunch of associated jobs fail, and maybe it's not high on the priority list to rerun the query or investigate or the person familiar with the job is away. And so this time a bunch of fraud goes uncaught.

Part of AWS sells those things. AWS started 12 years after Amazon.com. Amazon has large BI orgs to build custom systems costing tens of millions of dollars.

AWS was not considered secure enough for Amazon.com critical data as of a few years ago, maybe still is not. Look at the capital one breach due to S3.

Amazon does not just give a no-limit AWS account to each of their thousands of team. In fact they are more likely to tell the team to solve it with no resources. Rival teams will want an unpopular cost savings project like this to fail so they can have more headcount and budget.

From the advertising copy,I would think that AWS makes tasks like this totally trivial!

Trivial != cost-effective

Can you explain the last 8th point further?

Thousands of managers. Not all are honest. As an example, I toured an FC once and there was a still boxed perfect $9000 TV sitting in the damaged area. Most retail has a policy that damaged has to be destroyed. In reality places I have worked, managers came through and picked through it.

This is nothing new for retail or warehouses. If a manager is suspected of helping thieves, they might be moved around instead of fired due to politics.

When I worked at Best Buy years ago our Inventory/LP manager was fired (and later arrested from what I heard) because he was involved in a theft ring with other Inventory/LP managers/supervisors across the district. If I remember correctly, they were transferring items between stores and then saying they never arrived.

I have worked in the shipping industry (shipping as in ocean transportation). We had a very high rate of damage to pallets containing beer. A forklift clips a 6 pack on the corner breaking one or two bottles. Insurance pays out for entire pallet, stevedores go home with a lot of beer.

Theft was considered a insurable cost of business. I am sure insurance rates were priced accordingly. I am also sure our rate of theft was no different to any other port in the country.

you know what I think is funny? if 3% loss is acceptable than why do these people even get prosecuted? deterrent so that it's not 4%? sure but there's something unsatisfying about that as justification for imprisoning someone.

before someone accuses me of being a communist hippie I have a thought experiment for you: every 5 years I do spring cleaning by taking all of the boxes drawers etc of accumulated consumer goods junk I haven't looked inside of in the past 5 year and just take them to good will. I reason that if it hasn't been important in the last 5 years it never will be and that I would waste more time (money) looking closely. now the thought experiment: suppose someone had broken into my house over the last 5 years and stolen something from one of those receptacles that was eventually thrown out. should I be upset? now note that I'm not asking if I'm legally entitled to being upset but whether it would be rational to be upset.

but let me actually put on my hippie commie beatnik hat: there is so much money sloshing around in the economy and inefficiencies that entrepreneurs exploit to become wealthy. most of it is above board but some of it isn't and whether those people get punished or not is largely a function of whether they're caught before they're super successful or not. what is the difference here? these people saw an inefficiency and took advantage of it. to put a sort of hyberbolic point on it: why is this theft but selling toxic cdos not? in both "transactions" ultimately one party profited at the expense of the counterparty.

Edit: I love the downvotes. What exactly about my comment isn't "curious" as dang loves to exhort. God forbid someone questions the most totalitarian institution the human mind has ever conceived of: the law. HN is no different from every other pointless echo chamber that double speaks about freedom but suppresses dissenting opinions.

I, for one, would prefer my unused stuff went to a charity of my choosing than a thief.

> suppose someone had broken into my house over the last 5 years and stolen something from one of those receptacles that was eventually thrown out. should I be upset? now note that I'm not asking if I'm legally entitled to being upset but whether it would be rational to be upset.

A person's response to being burglarized is never purely rational. You've got the distress of the initial discovery, followed by the subsequent process of figuring out what (if anything) they stole, modified, or left behind, and the process of reporting the crime to authorities. Then you'll deal with self-blame about not investing more in home security and uncertainty about whether these thieves will come back for what they left behind (did they break in knowing you were gone, or did they mistakenly think you were home...?). Even if you had excellent video surveillance coverage that proved that they only took stuff you'd forgotten about, the burglar(s) violated your personal space, which obviously feels pretty awful, and may have broken something on their way in. The negativity of getting burglarized is irrational in nature and difficult to quantify in economic terms, but is nevertheless real.

In conclusion, I'd say every victim of B&E has great reasons to be upset, regardless of what (if anything) was stolen.

all of that is certainly true but not the point of my thought experiment because I'm sure Amazon the corporation doesn't experience emotional distress from having its personal space violated (so I'm sure that there won't be punitive damages for emotional distress in their sentences).

But... your thought experiment was specifically about a hypothetical burglary perpetrated against a person, and specifically did not mention Amazon or corporations.

Yes to illustrate a point about Amazon's margins, not to compare the person's emotional duress to Amazon's

Look into the Broken Windows theory: https://en.wikipedia.org/wiki/Broken_windows_theory

If you let criminals get away with breaking into someone's house and stealing things, it sends a message to other people that crime is allowed. Crime rates start to rise because no one is doing anything about it. People who were not criminals before are now comfortable being criminals because they know they won't be caught.

If you don't prosecute the small crimes, it often leads to bigger crimes.

I pre-empted you

>deterrent so that it's not 4%?

My point is imprisoning someone to "send a message" rather than for the effect of the crime is cruel.

> My point is imprisoning someone to "send a message" rather than for the effect of the crime is cruel.

So you're saying punitive damages (should) violate the 8th amendment? https://en.wikipedia.org/wiki/Punitive_damages https://en.wikipedia.org/wiki/Cruel_and_unusual_punishment

Yes I think I'm comfortable with that position.

No, your response indicates you do not understand the broken windows theory at all. I would suggest you go back and read it. I linked it for a reason.

You're imprisoning criminals for the crimes they committed, not just to send a message. But imprisoning criminals for committing crimes does send a message and that message is that crime is not tolerated.

You might be okay with someone breaking into your house to steal your personal property without your permission but I'm going to guess you're not okay with someone breaking into your house and murdering your entire family. Would you then argue that it would be cruel to imprison the murderer? Because it would just be to "send a message" the murder is wrong?

No, of course not. They committed a crime, and that's what they're being punished for. Punishing criminals does not reduce the crime rate. It's the act of not punishing them that increases the crime rate.

Note that this comment does not remove the need for you to read the linked wiki article.

I responded to your comment rather than broken windows policy, on which the consensus is not at all as solid as your tone suggests

https://cebcp.org/evidence-based-policing/what-works-in-poli...

>recent systematic review by Braga, Welsh and Schnell (2015) found that policing strategies focused on disorder overall had a statistically significant, modest impact on reducing all types of crime. This positive effect was driven by the success of place-based, problem-oriented interventions. In contrast, there was no significant overall impact of aggressive order maintenance strategies. Thus, they conclude that police can successfully reduce disorder and non-disorder crime through disorder policing efforts, but the types of strategies matter.

That was entailed in me conceding that letting these people go free would probably bump the shrink to 4%.

Now your comment was

>If you let criminals get away with breaking into someone's house and stealing things, it sends a message to other people that crime is allowed.

To which I explicitly called back to by responding

>My point is imprisoning someone to "send a message" rather than for the effect of the crime is cruel.

This isn't broken windows

He's pointing out that the 3% margin should exist evenly, so adding 1% theft on top means you'll hit 4%. So in order to pull of this stunt, they needed to make sure their margins were below 3%. But if you can get your margins below 3%, then why isn't that mandated?

This is forgiving people who break windows, but not people who carefully don't break windows so that they can then purposefully break windows as often as most people accidentally break windows

If that's the case, then they're woefully misinformed about the realities of retail. 3% shrinkage is accepted because below that it's mostly unpreventable accidental damages. Above that is either theft or preventable accidental damages.

If my server logs say I had three errors today but everything is running fine, I'm probably going to let it slide. I have more important things to work on. If they say there have been 100 errors today, even if everything seems to be working fine I'm going to take the time to investigate and probably push a patch. Not that those 3 errors aren't important, but it's hardly worth my time to track down every error unless something is broken.

Either way it's interpreted, it's not a terribly insightful comment.

It's hard to answer without knowing what you consider "rational". You're asking about an emotion, and a negative one at that. When is being upset ever "rational" in your mind in the first place? It's pretty nonproductive as far as the external world is concerned, and it sucks away your time and energy. The way you've posed it I don't see when it's ever rational, unless you're the type of person who thinks "I'm going to cry anyway, so I'll set myself a reminder to cry when I'm free at 5pm today because otherwise it'll interrupt my work tomorrow".

> suppose someone had broken into my house over the last 5 years and stolen something from one of those receptacles that was eventually thrown out. should I be upset?

Should you be upset if they raided good will safe too?

Reply on news.ycombinator.com