Lucas Wonderley

Supply Chain's Bright and Challenging Future

Sun, 06 Dec 2020 20:00:00 GMT

The last chapter of Marc Levinson’s Outside the Box peeks at the future of supply chain, and it’s a bit pessimistic. It gives you the sense that a golden age of supply chain is giving way to a phase where ideas primarily drive the economy, not things.

Levinson’s points made me question whether supply chain was a sensible industry to be in. They make sense, but I suspect they also downplay the fundamental role supply chain will continue to have in our economy, and the difficult and exciting challenges still ahead.

He makes several arguments I’d like to paraphrase and respond to.

1. An aging population

In the US, fertility rates are at an all-time low. We’re having fewer kids and living about as long, so the median age is rising. Older people are less likely to buy stuff like a home, furniture and vehicles because they already have them. So maybe our society will spend less and less on physical goods.

But we shouldn’t give this too much weight. There are still millions of new people coming into the world every year, and the generational difference in spending habits isn’t that drastic. A subtle demographic change isn’t going to change the fact that US citizens spend a lot on stuff, and it’s easier than ever to do so. And other sectors like pharmaceuticals and electronics could make up for lost demand elsewhere.

2. The sharing economy

Technology has allowed fixed assets like cars and homes to be better utilized through sharing economy apps like AirBnB and Uber. Does that mean Americans will purchase fewer homes and cars?

Prior to COVID, Americans were still buying cars like crazy. And despite the pandemic or maybe because of it, the real estate market is hot. Americans are still making these major investments.

The author doesn’t mention AirBnB, but we can’t discuss asset value in the sharing economy without mentioning it. It’s not clear that AirBnB’s existence would dissuade someone from owning a house. If anything, it helps home owners hedge against the downsides of being tethered to a house by allowing them to make trips more easily and maybe rent out their space while away.

Uber may discourage city dwellers from buying their own car, but cities already had cabs and trains. And extra mobility and the lack of a car note might just mean folks can go places and spend their money on other things.

3. Growing importance of software and services

Then again, there are plenty of things you can spend money on that aren’t physical goods. For decades, household spending has been gradually shifting to include less stuff and more services and experiences. Similarly, corporations spend more of their budgets on non-tangibles like research, development, design and elaborate tax avoidance schemes than ever. In particular, software is eating the world.

Yes, and software is eating supply chain. But that’s an opportunity for new businesses in the space, and for products to be delivered more efficiently and with more possibility than ever before.

I find it easy to get carried away with the idea that technology will not only enhance our lives but also shrink the need for a supply chain entirely. But ultimately, we still live in a world of things. There will always be tremendous challenges around making, moving and storing goods.

In airplane and auto manufacturing, the software built into the product is becoming a larger chunk of its worth. Levinson points out that 30% of a car’s value will come from its code by 2030.

But that doesn’t mean cars and planes will be commoditized. The very expectation of increased reliance on software implies manufacturers will find ways to make ever more reliable and useful products. If the switch to more software wasn’t of some genuine benefit to consumers, it wouldn’t be adopted. But it is, which is a sign these important sectors of our economy are still innovating.

There’s no need for a hard distinction between supply chain and software. The two can coexist. Software won’t eat supply chain out of existence. But it will let us build some very cool stuff.

A few scenarios

Let’s take this point to the extreme. Imagine a world where we’ve mastered nanotechnology and any item or structure can be made instantly — I’m thinking of Neal Stephenson’s The Diamond Age. There would still be practical logistical constraints to turning materials into a product at a customer’s doorstep. We’d call that supply chain.

If tech allows us to go beyond the physical world and into a compelling, ubiquitous VR experience, even that reality would likely have things to make, move and store. If we expand meaningfully into outer space, Earth logistics will seem like a walk in the park compared to the intragalactic version.

Technology just leads to newer, stranger problems.

My intention here isn’t to undercut Levinson’s predictions. I really just wanted to put my own doubts in perspective. Reasoning about the future can be maddening. But I assume people will continue needing and wanting stuff. Can you convince yourself otherwise? I can’t. 🙂

What Warehouses Do

Wed, 28 Oct 2020 20:00:00 GMT

When you think of a warehouse, what do you imagine? For many folks, the typical image is something like a wide, nondescript building surrounded by trucks. It stores stuff.

This image isn’t wrong. But on closer inspection, there’s much more going on in a warehouse, and its purpose is more nuanced than “storing stuff.” More generally, they provide flexibility. And flexibility is indispensible for dealing with the increasing demands that are being placed on supply chains.

To illustrate, let’s look at some of the distinct roles warehouses play.

Inventory buffers

Perhaps most obvious is that businesses use warehouses to buffer their inventories. Retailers and suppliers keep extra product around to avoid running out of stock.

Until recently, it was really hard to tell how much of a product would be bought. So distribution managers had to keep lots of inventory to avoid stock-outs.

But these days, businesses try to limit inventory because it’s costly. Inventory ties up capital, and it may be wasted if demand for a product is overestimated or changes suddenly.

Insights and predictions from computer systems make it possible to forecast demand. Using these techniques, warehouse managers can meet customer demands while keeping inventory low. Still, buffer stocks are important because there’s always a certain amount of uncertainty.

Consolidation and deconsolidation

Warehouses can also be used as sites where the form of shipments is changed through consolidation or deconsolidation.

Consolidation is combining small shipments into a larger package or truckload. This lets you combine things that are going to the same retail store or assembly plant and ship them together.

Deconsolidation is unpacking the contents of a large shipment before storing it. This is important in e-commerce because individual items have to be sent to customers. So they get stored individually in the warehouse and later packed into the customer’s order.

Postponement

Warehouses also enable postponement, or storing components until they’re actually needed. This lets companies increase product variety and flexibility without extra inventory costs.

For example, Dell pioneered the business model of build-to-order computers. They assembled computers partway and sent them to a warehouse. Only once they got a specific customer order did they put together the finished project. This made orders fast and customizable.

Cross-docking

Cross-docking is a technique where goods are transferred but not stored. This is ideal for food with an expiration date, luxury items that could be risky to store, or simply for minimizing inventory costs. The downside is that inbound and outbound shipments have to be synchronized, which puts more complexity and cost on transportation. The need for fast delivery and low inventory is making cross-docking more popular.

Cross-docking underscores the point that warehouses aren’t all about storage. They might be used to change the mode of transportation, or to change transportation providers entirely. In both cases, the facility acts as a joiner piece in the supply chain. It helps make the logistics ecosystem robust, despite being considerably fragmented.

Customs

Warehouses also play a special role in international logistics. Bonded warehouses let you store imported goods before they’re taxed. This flexibility helps you save money and delay paperwork.

For example, you can ship your product to another country in bulk, but only import items once they’re purchased. You can even export those items to another country without incurring a duty in the first one.

The details vary by country, but bonded warehouses are ubiquitous and an important tool for multinational companies.

Return processing

Return processing is another function that’s become more important with the rise of e-commerce. 20% of all orders are returned, and that number’s even higher for expensive products and during the holidays.

Returns should typically be identified and returned to the front of the stock queue for resale. This process has its own challenges, and some facilities are now designed to be dedicated entirely to return processing.

Warehouses are crucial building blocks in a functioning supply chain. They provide not only storage but also opportunities to transform and transfer products, and even intentionally delay their downstream use. As consumer demands continue to challenge supply chains, we can expect to see many more innovations in this space.

Integration, The Quagmire the World Runs On

Mon, 24 Aug 2020 10:00:00 GMT

Integration is often seen as messy and unfun. And it’s hard to get right. Regardless, to get any real value from a large scale software system, it’s necessary.

A system that facilitates all a business’ processes is a holy grail of sorts. Though it sounds straightforward, making technology products work together when they’re not designed to is a common sticking point.

Integration is as important in supply chain as in any other industry: The technology is fairly fragmented, and companies can even benefit from connecting with their parters’ IT systems to improve planning.

Marketing pitches for ERP systems — huge packages of software that can do almost anything, albeit rather painfully — sometimes point out that you need an end-to-end system to implement cutting edge practices like machine learning and demand-driven planning. Such capabilities, it’s emphasized, depend on visibility and real-time access to data from different parts of the business.

They’re right, but they downplay a more mundane point: Your systems need to be integrated if your business is going to work at all.

Most business processes are inescapably horizontal because they stretch across functional boundaries. A canonical example is online retail, in which online shopping, purchasing, fulfillment, warehouse picking, transportation and delivery all have to work in concert. This brief and highly incomplete sketch makes one thing clear: It’s not enough to implement these individual tasks with software. The pieces have to work together seamlessly to enable end-to-end scenarios.

So while integration can often seem bizarely expensive, slow to implement, and generally distasteful, it’s also indispensible. It’s the glue that takes several perfectly good bits of software that each do one thing well and combines them into a strange assortment that you probably wouldn’t design from scratch, but that actually makes businesses work.

Drivers of Global Expansion

Sat, 15 Aug 2020 10:00:00 GMT

Since the end of World War 2, international trade has increased by 30x after adjusting for inflation. There are four main drivers of global expansion:

COST

Companies can outsource operations like materials sourcing, manufacturing or product development to reduce costs. And it’s often possible to take advantage of economies of scale by selling in multiple markets, reducing overall marginal costs.

COMPETITION

Organizations may follow competitors into new markets to avoid losing market share. Competitive dynamics can also interact with the cost driver: If one company releases a new low-cost product, its competitors may be further incentivized to outsource to releaae a similarly priced offering.

MARKET

Large companies may strive to be global brands so global travelers can get the same experience wherever they go. And consumers have become increasingly open to buying foreign products, which somewhat levels the playing field against domestic competition.

TECHNOLOGY

Tech innovations have made it easier for consumers to buy foreign products online, and for retailers to advertise to foreign markets. And many aspects of modern supply chains are made possible through the integration of countless software systems.

Why you should use Unix time in your apps

Sat, 02 May 2020 10:00:00 GMT

Seconds since 1970, January 1st at midnight: The Unix Epoch.

This way of representing time, called Unix or epoch time, is common in software because it makes it easy to get dates and times right.

What makes it useful? I don’t know about you, but for me, a big integer isn’t a great way of talking about time. For example, August 17th at 05:27:49 AM Eastern Time seems a lot more intuitive than -485188331, which is the Unix time version. But more familiar formats also have a lot of problems that Unix time avoids.

For one thing, there’s no confusion over time zones! It’s based on UTC, period. This means just a bit less mental overhead for system builders. Unix timestamps unambiguously refer to the same moment in time, wherever your code happens to run.

And working with it is simple. How many seconds are between two timestamps? With Unix time, you just subtract. No separate data structure needed to represent a duration. Want to convert seconds to hours? Just multiply by the number of seconds in an hour, or 3600. Milliseconds or smaller fractional seconds are decimals.

With modulo division, you can even check the number of hours since midnight of the current day, the number of minutes since the top of the hour and so on. And the simple representation makes it easy to write tests with canned data.

Simply put, it’s easier to build and reason about systems based on Unix time than with more human-intelligible formats, which have to be converted to a data structure and are liable to misinterpretation.

But why is it based on the start of 1970? Apparently the standard was invented in the early ‘70s, and it seemed like a reasonable thing to do. Fair enough, ‘70s computer science pioneers.

In software, Unix time is ubiquitous. Most libraries have a way of converting it to a human-readable timestamp and vice versa. The characteristically descriptive Swift programming language provides a timeIntervalSince1970 API. JavaScript returns the same value in milliseconds (in other words, multiplied by 1000) when you call getTime() on Date objects.

One downside of Unix timestamps is that they are less human-readable, which you have to account for while debugging. You can work around this with utility functions for printing human-readable timestamps. You can also convert between Unix time and other formats with tools like epochconverter.com.

Of course, if the first second in 1970 has a Unix time of 0, that makes everything pre-1970 a negative number that gets more negative as you go back in time. You’ll want to take the absolute value to do modulo division. Other basic operations work fine with negatives.

Now, because nothing in internationalization is as simple as we’d like, there is a rub. Leap seconds, or extra seconds that are prescribed by scientists from time to time, are accounted for in Unix time by double-counting. Those extra leap seconds actually have the same value as the second before them. This makes Unix time not a true representation of UTC!

International atomic time (TAI) is an equivalent system which includes leap seconds as separate timestamps. This means it gradually gets out of sync with earth’s rotation by roughly a second per year. But in return, it avoids duplicating values. In my experience, ordinary Unix time is the standard, but TAI may gain popularity, especially in domains like time-critical domains finance and aerospace engineering.

Use Unix time to represent time in your apps! It’ll make your code easier to understand and less error-prone, so you can focus on other stuff.