Shifting Perspectives for Effective Problem Solving
Hard problems are hard. That’s what makes them hard. We face obstacles which seem insurmountable throughout our lives. Whether it’s a tightly-sealed jar, complex life situation, or multi-nodal UX design issue, we’ve all been in a position where the solution to our problem is opaque, incomplete, or seemingly nonexistent. Sometimes there is no solution. However, more often than not, changing your perspective on a problem gives you a better chance of solving it.
Interestingly enough, a prescient example of shifting perspectives to solve problems comes from a recent application of artificial intelligence in the Secret Archives of the Vatican. With over 50 miles of shelving and tomes dating back hundreds of years, the archive is home to some of the most important, society-changing documents in western history...but many of them have never been read by modern-day scholars.
Though many of the world’s major historical libraries have taken steps towards digitizing their resources, the Vatican’s Secret Archives are still “destination documents.” By this we mean that one must apply to visit the archives, make the trip to Rome, and page through ancient manuscripts by hand, presumably with a gloved hand. Oh, and most are in languages we don’t actively use anymore, namely Latin.
A New Type of Problem
Other archives and libraries have been able to scan large quantities of documents and, through the use of optical character recognition (OCR) artificial intelligence, make these documents searchable, translatable, and 100x more accessible to researchers and the general public. Teaching a computer to read a document in these instances relies on individual character recognition. An AI program is taught to recognize the spaces between characters, which makes scanning typed documents a breeze.
But the documents in the Secret Archives are not typed, and many of them are written in intricate calligraphic script. They are written by hand, which means there is variation in letter styles. Additionally, the scripts tend to feature connected letters, which leaves the AI program with little hope of distinguishing individual characters. In order to solve for this issue, researchers and programmers had to find a new way to teach the AI to understand text.
A New Approach
Ultimately, new software, In Codice Ratio, was designed to determine individual pen strokes rather than spaces. With the help of 24 Italian schools, researchers were able to teach the program which sequences of strokes made up individual letters. An additional layer of learning was also needed to teach the program to use probability to tease out sequences of letters which may look like other letters (“nn” might look like “m” or even “iii”).
It’s first real application at the Vatican Archives resulted in ⅔ of words spelled correctly, and 96% of letters accurately transcribed.
Rethinking and Resizing
There are two valuable lessons to this story. The first is that sometimes you need to create your own solution to a unique problem. The computer scientists who created In Codice Ratio were faced with a problem that, with the tools they had, was impossible to solve. They had to rethink the problem from an entirely different perspective, creating a new tool along the way. The second lesson here is in the way the In Codice Ratio AI program, itself, breaks down letters into individual strokes to make sense of the greater whole.
Breaking down complex problems into smaller pieces is often a good way to make them more manageable, but, just like a complex computer program, you’ll need a smart way to parse and re-connect those pieces logically. Before breaking down any problem, remember to think of the question you’re trying to answer at the end of the line and all the different avenues you might take to get there. Even when you’ve thought of everything...you probably haven’t thought of everything.
Forget your Training
Hard problems are hard, but going about solving them the wrong way makes them harder. We tend to rely on our past experiences too much—it’s how we’re generally programmed—like an OCR AI relies on the contours of the spaces between characters to create meaningful words. When we begin to see the different potential angles of a problem and all the smaller problems that complicate the thing we’re trying to solve for, we stand a much better chance of getting to a viable solution faster.
For a little extra help, you might consider the Oblique Strategies card collection. Written by Brian Eno and Peter Schmidt, these cards offer simple ways to change up your mode of thinking—ideally giving you a better way to approach hard problems. The physical cards are beautiful and sturdy, but the wisdom is also available for free online.
The world is full of people trained to think a certain way. Make yourself stand out by creating your own path forward.