Why a Memory Researcher Recognised the Simplicity of what3words

28/05/2025

When what3words introduced the idea of addressing locations with just three dictionary words, its simplicity was immediately evident to those of us on the team. This simplicity also resonated with leading experts in memory. In fact, Professor Alan Baddeley – one of the world’s foremost memory researchers – instantly recognised why three-word addresses make sense from a human perspective. Baddeley, famous for his pioneering work on the Working Memory model, summed it up in plain terms:

“Almost everyone would have no difficulty in remembering three unrelated words long enough to write them down. Many people would have difficulty in remembering a random mixture of six digits and letters.”

This insight from Baddeley reflects a fundamental principle of cognitive psychology. By design, what3words leverages the way our short-term memory works. Three-word addresses are easy for people to remember and communicate because they sit comfortably within the natural limits of human memory. To understand this better, let’s look at the science behind memory span and why a sequence of three words is so effortlessly memorable

Memory Span: Why 3 Words Work

Psychologists have long studied how many pieces of information the average person can hold in mind at once – a concept known as memory span. Decades of research show that most people can recall roughly 5 to 9 items in their short-term memory (often cited as the “magic number seven, plus or minus two”). Crucially, that span depends on what kind of items are being remembered. A classic 1957 study by C. W. Crannell and J. M. Parrish compared immediate memory span for different types of content, and the differences are striking: on average, people could correctly remember about 7 digits, 6 letters, or 5 unrelated words in sequence. In other words, recalling a handful of random words is slightly more challenging than recalling the same number of digits – but our capacity to hold three words in mind is virtually guaranteed.

To put this in perspective, a telephone number of 7 digits pushes the limits of many people’s short-term memory, whereas a set of just three words is well below that limit. As a list gets longer, recall accuracy drops off more steeply for letters and especially for words than for digits. Figure 1 illustrates this: as the length of a sequence increases, performance falls fastest for word lists (making a long string of words hardest to remember), while shorter sequences – like a trio of words – remain easy to handle. In essence, almost everyone can easily manage three words without error, particularly for the few moments needed to write them down or say them aloud.

Figure 1. Recall accuracy vs. list length for sequences of digits, letters, and words (data from Crannell & Parrish, 1957). In memory experiments, performance declines as the number of items increases; here we see that people can perfectly recall short lists of three items, whether they are numbers or words, but longer lists of words quickly become difficult.

Another reason three words fit so well within our memory span is the nature of the words themselves. Not only is the quantity of items important, but so is their quality . Research led by Baddeley in 1975 demonstrated the “word length effect”: our short-term memory can hold more words if they are short and quick to say than if they are long or hard to pronounce. In those experiments, people could recall sequences of shorter words more accurately than sequences of longer words. This finding aligns with the design of what3words: the system is available in over 60 languages to date, and deliberately assigns simple, commonly used words in areas where that language is spoken the most. More complex or less commonly used words (e.g. ‘electromagnetic’) are placed in remote or uninhabited areas such as oceans and deserts.

By using everyday words in each language, within the likely geographic hotspots for that language, a what3words address becomes a compact package of information that our brains can grasp with minimal effort. Three words like “apple spoon table” are far easier to remember (and faster to repeat) than an equivalent jumble of letters or digits. And even if the words are unrelated – as what3words addresses intentionally are – each word is a meaningful chunk that might conjure a mental image or association, making the trio even more memorable as a group.

From Research to Real-World Addresses

It’s one thing to know from lab studies that a few random words are easy to remember; it’s another to turn that insight into a practical addressing system. Before what3words, no widely used addressing or location-reference system had ever been based solely on words – despite the cognitive advantages. Traditional street addresses evolved organically and typically combine house numbers with street names and postcodes. These can be long or inconsistent, but we use them because they refer to human geography. Systematic codes like postal codes or GPS coordinates were designed more for machines and databases than for people’s memory. Even when such systems incorporated some psychology, they still relied on alphanumeric strings that are harder for most of us to memorise than words.

A young Alan Baddeley himself was involved in the effort to make machine-friendly codes more human-friendly: the design of the UK postcode in the late 1950s and early 1960s. As a newly minted researcher, Baddeley worked with psychologist Reuben Conrad and the UK Post Office to ensure the new postal codes would be usable by ordinary people. They applied cognitive principles to keep the codes reasonably short (no more than 6–7 characters, in line with typical memory span) and to structure them in a memorable way. For example, the UK postcode was split into two parts separated by a space, creating a rhythmic two-part code that was easier to recall and less error-prone than one long string. Certain letters and numbers that could be confused with one another (such as “O” vs “0” or “I” vs “1”) were excluded to avoid memory mistakes. These decisions, grounded in the science of memory, made postcodes a success for their time – a person could glance at an address like “SW1A 1AA” and remember it long enough to type or write it, because the format was optimised for human recall.

Yet, even with those improvements, a postcode or any similar code is still an abstract string. No matter how cleverly designed, “SW1A 1AA” or “1600 Pennsylvania Ave NW” or “51.5079° N, 0.1280° W” will never roll off the tongue quite as easily as actual words do. This is where what3words broke new ground. The idea of using words – something inherently meaningful to people – for addresses had been suggested by cognitive research for years, but until recently it hadn’t been implemented on a global scale.

When the what3words founders set out to label every 3 m × 3 m square on the planet, about 57 trillion of them, they knew they needed an identifier that was both vast in potential combinations and simple for people to recall. Reaching that scale with traditional formats would mean 16 random digits or roughly 10 mixed letters-and-numbers, each going well past the average memory span, whereas the same coverage fits comfortably into three everyday words – a sweet spot that struck our founders as both mathematically efficient and cognitively elegant, yet curiously under-explored before what3words.

what3words is the first system to give every 3m x 3m square on the planet a fixed, three-word address composed of dictionary words. For example, the front entrance of our London office can be found at ///filled.count.soap . Three random words can be far easier to remember and communicate than coordinates or codes. It’s much simpler to tell someone “meet me at ///filled.count.soap” than to recite a latitude-longitude or guide them through a complex address – and crucially, they’re less likely to misremember a three-word address. Each word in the address is a familiar unit, and together the trio is unique enough to specify one location in the world. Moreover, a combination of words can often be visualised or repeated with a natural rhythm, whereas a mix of letters and numbers (say, XZ7Z54QZ39 or 37.715972,-122.413378) doesn’t stick in the mind nearly as well. what3words essentially took the memory-friendly attributes of natural language and applied them to the problem of pinpointing precise locations.

It’s worth noting that implementing this idea globally only became feasible in recent years. what3words relies on software algorithms and a predefined list of words to encode 57 trillion 3m squares into a package that runs locally on a smartphone entirely offline, without needing an internet connection – something that wouldn’t be practical without modern computing. But from a user perspective, all that complexity is hidden. You just get a simple, easy-to-communicate address. The design of what3words marries technical innovation with cognitive simplicity: it’s a system built “bottom-up” from how people actually remember information. And as Baddeley recognised, three random words are just right for our brains.

Bridging Words and Traditional Addresses with Swiftcomplete

While what3words addresses are a novel way to communicate location, they aren’t meant to replace the addressing systems we already use. After all, people still rely on street names, postcodes, and landmarks in daily life. The goal is to make it easier to describe and share precise locations, not to force everyone into a new method. With that in mind, what3words has been working on integrations that bring the best of both worlds together. One exciting development is our recent acquisition of Swiftcomplete , an address validation platform. This system brings street addresses, postcodes, and what3words addresses into a single search experience .

In practical terms, this means that when you’re filling in an address or searching for a location with Swiftcomplete, you can use whichever format you have on hand – a street address, a postal code, or a what3words address – and the system will understand it. For example, an online shopper could enter a what3words address for a specific delivery drop-off point (say, a side entrance or a remote location identified by three words) and still provide the official street address in the same form. Swiftcomplete’s technology handles the lookup seamlessly: type what3words address into a standard address field, and it will suggest the corresponding street address, or vice versa. The result is that businesses can accept precise what3words locations alongside conventional address information. A courier or navigation app will get all the details needed to find the exact spot: the familiar street address for context and routing, plus the precise what3words location in one go.

This unified approach is all about convenience and confidence. By combining the human-friendly simplicity of what3words with traditional street addressing, we’re ensuring that using three-word addresses is as easy and natural as using a postcode or city name. People don’t have to choose one or the other. Chris Sheldrick, Co-founder and CEO of what3words, noted that Swiftcomplete allows entering locations to be “as smooth, simple and efficient as possible” for users and businesses alike. It’s a good example of our philosophy: enhance the way people communicate location by building on what already works well. The three-word format adds a layer of precision (and, as detailed above, memorability!), while integrations with tools like Swiftcomplete embed that capability into everyday processes (such as e-commerce checkouts, delivery forms, or mapping services). The recent acquisition of Swiftcomplete has further solidified this, enabling what3words to offer a consolidated solution that handles street addresses and the pinpoint accuracy of three word addresses in one system.

A Thoughtful Design, Decades in the Making

From the perspective of the what3words team, it’s rewarding to see how an idea grounded in long-standing memory research is improving the way people find and share locations. The simplicity of what3words is not an accident or a gimmick – it’s backed by years of cognitive science showing how our brains handle information. In the 1950s, researchers understood that a few short words are easy to remember, but it took until the 21st century and advancements in technology to turn it into a practical addressing system. Alan Baddeley’s reaction upon encountering what3words was a validation of this approach: it confirmed that we had inadvertently followed a recipe that cognitive psychologists had known all along would be user-friendly.

By designing our system around human memory limits, we aimed to create something universal (anyone can use three words, regardless of language) and reliable (words are less easily forgotten than long strings of numbers). The ongoing integration with traditional addressing through tools like Swiftcomplete shows how what3words can complement and enhance existing systems, rather than compete with them. You can search for “Tower of London”, a postcode like “EC3N 4AB”, or “///power.builds.trace ” – different methods pointing to the same place (albeit the 3 word address is more precise and pinpoints a particular entrance) – and get the best of all worlds.

Ultimately, we believe that three-word addresses feel intuitive. When you use what3words to share a location, you’re benefitting from a design that incorporates decades of memory research to make your life easier. Professor Baddeley’s observation encapsulates it well: almost anyone can remember three words. That’s a humble benchmark, but it’s one we knew we had to meet for what3words to be truly useful. Every 3 metre square in the world now has an address that sounds human – an address you can say, write, and recall with ease. And as we continue to integrate this system with familiar address services, communicating precise locations becomes as straightforward as chatting about the weather. It’s the convergence of human psychology and technology in a modest but meaningful innovation, one that helps people navigate the world in a simpler way.

Sources: