"Nice of him, that compliment (in my mind anyway)".

Gregg Mayles recently announced his departure from Rare after 36 years, which means the brains behind such classics as Battletoads, Donkey Kong Country and Banjo-Kazooie is no longer with the legendary British studio.

While he's very much focused on his future, Mayles is equally keen to promote his past, and has been digging into his personal archives to unearth material related to Banjo-Tooie, which recently celebrated its 25th birthday.

Read the full article on timeextension.com

As a card-carrying AI hater, I feel it my duty to point out when it’s actually useful, and Dan Cohen presents such a case:

“All goes in the usual monotonous way.” That is the depressed sigh of George Boole in a letter to his sister Maryann in 1850. It was the spark for my book Equations from God: Pure Mathematics and Victorian Faith. Boole, the English mathematician who gave us the logic at the heart of the digital device you are reading this on, was teaching in Cork, Ireland at the time. On a cold December day, he wrote to Maryann about his feelings of profound loneliness. In a city that was on edge from religious strife and famine, he played piano at home to an empty room, and took long walks by himself. At the end of the day, he retreated to his equations, which seemed to transcend the petty differences of humanity.

But before developing my thesis about the fervent emotions behind Boole’s seemingly cold mathematical logic, I first had to read his damn handwriting. Talk about monotony! There were hundreds of letters and notebooks in his drifty scrawl. In retrospect, Boole’s handwriting is actually not that bad; I’ve encountered far worse since reading his in Cork. And it helped that I had taken a brief course on paleography, the art of deciphering handwritten historical documents. But it would have saved me a lot of time getting to the interesting interpretive phase of my research if a computer could have converted his handwriting into machine-readable text, as it already could for typeset text through a process called optical character recognition (OCR).

Since I wrote that book, university and industry labs have been trying to solve the incredibly difficult problem of handwritten text recognition (HTR). OCR quickly approached 99% accuracy for digitized books, whereas even the best HTR systems struggled to reach 80% — two incorrect words out of every ten. The issue is obvious: unlike the rigorous composition of books, handwriting is highly variable by author, and words are often indeterminate and irregularly arranged on a page.

He uses George’s letter to Maryann as a test, which most approaches fail; then he hits the jackpot:

They have gotten incrementally better over the past three years, but I was frankly stunned when I put the letter into Gemini 3 Pro this week and asked it to have a go at the transcription […] Gemini transcribed the letter perfectly: it figured out that the right side is the beginning of the letter, not the left (the letter actually continues on the other side of the paper, which accounts for the discontinuity between the two sides we are viewing); it left off the periods where Boole also (oddly) omitted this punctuation; and it includes a self-reflective analysis of where it might be wrong and provides alternative readings.

Even wilder, when you click on a “show thinking” tab, Gemini provides a long discourse on its approach and minute details about word choices […] This thinking goes on for almost 2,000 words, and what’s remarkable is that it is essentially a verbalization of what you’re taught to do in a paleography class: assess the overall document first, determine key features, study letter shapes and strokes across the letter to refine your understanding of the particular script, consider context and word/phrase possibilities, think about the coherence of content, grammar, and usage, identify any contractions, proper names, and other oddities, etc.

He tests Gemini on increasingly harder materials, and the results are impressive. By all means click through and read the whole thing, with the examples; as he says:

At this point, AI tools like Gemini should be able to make most digitized handwritten documents searchable and readable in transcription. This is, simply put, a major advance that we’ve been trying to achieve for a very long time, and a great aid to scholarship. It allows human beings to focus their time on the important, profound work of understanding another human being, rather than staring at a curlicue to grasp if it’s an L or an I. Could we also ask Gemini to formulate this broader understanding? Sure we could, but that’s the line that we, and our students, should resist crossing. The richness of life lies in the communion with other humans through speech, the written word, sounds, and images.

Promotional Super Smash Bros. Melee wallpaper downloadable from Nintendo of Europe’s official site in 2002, featuring Donkey Kong and Captain Falcon.

Main Blog | Patreon | Twitter | Bluesky | Small Findings | Source: retrogameart

Officially licensed toy Game Boy with a set of Donkey Kong Land 3-themed card inserts, from Japan.

Main Blog | Patreon | Twitter | Bluesky | Small Findings | Source

Tumblepop - Data East 1991