Hacker News Reader

Comment by oliv__ | Hacker News Reader

Tengential but I am always curious what software these huge logistics company use to manage their business. From my outsider's perspective, it seems like a best-in-class-software enabled company would have a massive advantage over the rest.

Interested to hear if anyone has insights about this

View on HN

I’m a software engineer at one of these global logistics companies.

The market is extremely competitive. We are one of the biggest in the business, and we only capture small percent (I heard 3% before) of market share.

We build event driven systems primarily glued together with Kafka. There are thousands of deployed products.

I'm interested to read about this if you've got links to related blogs or such.

On the logistics business side, I’ll admit that my knowledge is from the day-to-day and don’t read a lot outside of work.

Our company does have plenty of marketing material of course: https://www.chrobinson.com/en-us/about-us/technology/robinso...

I could ask some of the data science folks I know in the org if they have blogs, I’d be curious about that as well.

We have a blog that slowed down during covid (we’ve been very busy), but here’s one post about a serialization library we open-sourced: https://engineering.chrobinson.com/technology/inside-chr-avr...

Thanks for sharing these links, it's a good peek inside the company! Idk if reality lives up to the marketing but those videos made it look like an interesting place to work at

flexport? (Sorry that's the only company I can think of that does something along those lines)

There were some interesting profiles following the NotPetya attack on Maersk in 2017 where Maersk claimed to have 1,200 - 1,500 applications, 49,000 laptops, 6,200 servers.

Maersk are entering into block-chain "distributed ledger technology" with IBM and similar modern solutions. But one article put $300 of every $2,000 of shipping costs for administration and paperwork[0].

I think you're right, there is a massive advantage to be had, and companies are chasing that advantage. But from my (tangential logistics) background, even the biggest shipping companies have the usual range of legacy systems, heavy administration overhead, plenty of paperwork, excel-based-tools and huge integration headaches.

0 - https://www.supplychaindigital.com/technology-4/maersk-and-i...

> 300 of every $2,000 of shipping costs for administration and paperwork

Damn that's a lot. I understand this is a capital intensive industry so I guess it's hard/impossible to get in now but it seems like if someone started a competitor from scratch and had great software as the foundation of their company, they'd make a killing

15% overhead is actually pretty low all things considering.

Yeah the cost for paperwork and administration in aviation is over 50%.

This is an extraordinary claim. Obviously some of the prior costs are due to paperwork, but really?

https://transportgeography.org/contents/chapter5/air-transpo...

That doesn't break it down in the same way - what percentage of "labor" is "administrative overhead" for example.

Everyone has to do paperwork, but is that 'admin overhead' it an essential part of safety culture? I'm struggling to see where the overhead is that significant.

It is a lot more obvious in other fields -- medicine and highly regulated/litigious professions.

A good deal of it probably has to do with how crazy customs gets in every single port, and how to properly insure goods. (Lloyd’s of London got its start in marine insurance.)

If your paperwork is mostly driven by other actors tech probably won’t help all that much. Tech has not made significant inroads in disrupting American health insurance, for example.

Is there a reason these customs processes can't be simplified? Or is it due to some perverse capitalist reason, like TurboTax lobbying against simpler tax code?

Here in Norway it's mostly been a lack of focus and money.

The government doesn't seem to consider the huge hidden cost to the businesses in the country this represents, and the businesses haven't been stellar at highlighting it, which I guess is because they mostly just push that cost onto the consumers.

- Every set of goods has its own regulations. Ships carry all sorts of goods.

- Every country has different customs regulations, so paperwork is not really that reusable across ports.

- Customs is a huge source of fraud. In poorly managed countries it is one of the prime vectors for bribery and corruption. So destination ports have to verify everything even if they do receive the paperwork, because there’s no trust.

Generally speaking if you want to cut customs red tape your options are some kind of free trade deal (which is not necessarily popular and never covers all types of goods), or banding together with a bunch of different countries to submit to one supranational entity like the EU. In fact we have a live example of how complicated customs arrangements can be; look at how trade from UK to EU has been disrupted by Brexit and the introduction of customs.

One big factor is simply that a customs authority can dictate whatever process it wishes for ships who want to load/unload cargo there.

It's also not just customs, there are a ton of ancillary processes related to berthing - things like harbor fees, environmental documentation etc.

I know of at least one case where, as recently as ~2015, a shipping company had to keep around old machines with IE8 because that was the only way to interact with authorities at a given port.

"great software"

"logistics"

"from scratch"

it does not compile, I guess it's just like those ERP Systems - you don't need shitton of domain knowledge, because you need fuckton of domain knowledge and during development you'll learn even more and have to refactor a lot of stuff frequently for edge cases.

https://xkcd.com/1831/ relevant xkcd

> had great software as the foundation

This is obviously something that affects how competitive you can be. As an example, our software allowed a customer to go from 5-6 hours of work in their previous application, to less than 10 minutes in ours for a typical workload.

However, a big remaining issue is that there's a lot of local laws and procedures around which are not exposed in a computer-friendly way. If they are computerized they're often old systems with serious limitations.

For example, import declarations in Denmark is limited to only 99 goods items, so consolidation of goods items is almost required, especially in e-commerce settings, which again makes it more difficult if things are re-exported (customer didn't want that jacket say).

This can't be fixed on the commercial side, it requires work by the government agencies.

There was a very interesting Omega Tau podcast about this very subject a while back. They go into a lot of detail about the optimization problem and how they address it.

http://traffic.libsyn.com/omegataupodcast/omegatau-146-conta...

Also: a plug for Omega Tau, one of my favorite podcasts. https://omegataupodcast.net

Edited to add: go to the Archive page and scroll down. About half the episodes are in English and half in German.

In 2016 John Willis was with Docker and he gave a talk where he casually recommended a book called The Box [1] about how shipping containers have shaped the modern world. When I was reading it there was a part where they talk about how the logistics optimization problem caused the major shippers to be some of the earliest adopters of computers. Interesting book.

[1] - https://www.goodreads.com/book/show/316767.The_Box

The podcast "Containers" also covers how shipping containers have shaped the mordern world — well worth a listen, though I don't recall it mentioning this aspect of their logistics.

That book has a lot of interesting lessons to draw from [1] and that was even before they became a ubiquitous analogy for various things related to software containers. Sun's Jonathan Schwartz was talking up this book in the late 90s.

Three I picked out were:

- Existing infrastructure matters (e.g. SuezMax)

- Standards matter (container size and handling needed to be standardized)

- Process matters (e.g. containers needed changes to the way loading and unloading was done, and hence changes to labor agreements, at major ports)

[1] https://www.cnet.com/news/three-lessons-from-the-shipping-co...

Also tangential: I'm not sure about the real-life software itself, but IIRC, the example app in Eric Evan's Domain Driven Design is a shipping logistics app!

Who wants to start a company to disrupt logistics software like? Closest thing I can think of is Flexport.

Well there's Flexport.

Reply on news.ycombinator.com