I don't think that this produces correct results, as it seems to determine whether each individual number is Numberwang. However, whether a number is Numberwang can also depend on the previous numbers in the sequence or other factors.
So I think as it stands right now this model can only determine whether a number would be Numberwang as the first number of a sequence, but even for that I still wouldn't rely on this in the actual game show.
Do we really know that? Or have we only seen the sequences from people searching for Numberwang? Has a number been both Numberwang (or Wangernumb) and not Numberwang (or Wangernumb)?
Yeah, we do - there's a famous episode where after rotating the board both contestants stubbornly kept picking 1, and after a couple of iterations one of them was Wangernumb.
I can confirm this is the first episode, which I just watched since I had no idea about Numberwang until a minute ago. So I guess a true Numberwang fan would know that.
The rule book is quite big. You must have seen it in one of the many ads for the family board game version. As one site summarizes it:
> A NumberWang "board game in a box" was produced in 2008, including the board, numbers, two 400-sided dice, special hosting accessories and all 37 volumes of the rules so that you can say "That's NumberWang!" with confidence. For everyone from ages 8 to 88.
So even the true fan may falter in his or her judgement.
Obviously, it depends on whether board has rotated or not. It seems the model doesn't support board rotation yet, so the result is consistent with the limited set of features.
It's not well known, but in February of 1983, there was a finch in the Galapagos that was found to be numberwang. And of course there was the famous incident in the 1994 World Series where a homerun was overturned because it turned out it was a wangernumb hit into the stands, and not a standard baseball.
My favorite line, especially as an American because I can say it out loud without the person getting upset since they have no frame of reference. And other Anglophiles I could probably handle.
The idea that the snooker commentators are probably drinking is very Britain, and especially Britain from a particular era.
I think that series came out slightly after that era ended, I remember as a child (so years earlier) realising that snooker players at the time were often tipsy if not actually drunk, and also darts players, but by the time I was an adult the professionalism really changed that, turns out that if you're very good you probably are better at snooker when sober. I think the commentators, being often a previous generation of players, lagged that slightly.
I happened to be born at the right time for Mornington Crescent to be temporarily closed in the exact window when I was travelling to London often, using the Tube on my own, and listening to ISIHAC and so for a while I assumed the joke was that it'll never open - and then one day of course it re-opened.
I tried to pass Mornington Crescent the other day, but I was in Knip; bloody typical. As you’d expect, I ended up double-shunting onto the Victoria Line.
I knew the Mitchell and Webb duo from _The Peep Show_ and one day (long before the pandemic) I saw a comment on Slashdot about working from home linking to this sketch: https://www.youtube.com/watch?v=co_DNpTMKXk
Then I watched that show and learned about Numberwang.
This is hilarious and I miss numberwang. Also, the history of numberwang. And everything to do with numberwang. It was the most prescient sport back in the day.
I'm glad that someone has stepped up to revive the legacy of numberwang.
My friends, this is the single most hilarious thread I have ever experienced on the Interwebs. I am deeply grateful to have found this. I feel like my life is now complete.
What's the training corpus, just the canonical Mitchell and Webb sketch? Or ancilliary data like comments and references, or worse synthetic data generated on the primary sources?
What it is: a character-level CNN with 80,804 parameters. The weights are a 1.79 MB JSON file and inference is about 100 lines of Python standard library — no PyTorch, no NumPy. It runs on a Pi Zero. It accepts digits, number words in eleven languages, arithmetic ("96 divided by 2", "deux fois trois"), Roman numerals, ordinals, clock times, currency, and fictional numbers ("shinty-six"). Anything with no numeric content is correctly ruled out as never able to be Numberwang. Whatever comes to 1 or 44 is Wangernumb and you rotate the board.
Held-out accuracy is 88.9% on 486 probes reserved from training by construction. The ceiling is ~98%, because roughly 2% of training labels are inverted at compilation time, in accordance with long-standing adjudication practice.
For comparison I ran Qwen3-1.7B on the same suite with the four verdicts as a constrained multiple choice: 51.9%, which is 2.3 points above answering "Numberwang" to everything. It answers "Numberwang" to 93% of inputs and never once identifies a Wangernumb. So the accuracy table has a verdict-distribution column, since one number can't tell a model that decides from one that agrees.
Honest weak spot: arithmetic is memorised, not computed. A conv net can't add. On operands reserved from training it gets 60% on symbolic expressions and 44% on foreign-language ones.
Dataset (185k adjudicated utterances), training script, evaluation harness and benchmark are all in the repo and reproduce from a fixed seed. Model card on HF: https://huggingface.co/graafhenk/numberwang
So I think as it stands right now this model can only determine whether a number would be Numberwang as the first number of a sequence, but even for that I still wouldn't rely on this in the actual game show.
> A NumberWang "board game in a box" was produced in 2008, including the board, numbers, two 400-sided dice, special hosting accessories and all 37 volumes of the rules so that you can say "That's NumberWang!" with confidence. For everyone from ages 8 to 88.
So even the true fan may falter in his or her judgement.
https://youtu.be/0obMRztklqU?is=qufDLUkqofpr-K1Y
“I asked someone how the stock market works, and they sent me this link. I should feel duped or trolled, but I think I actually learned something.”
"WATCH THE FOOTBALL! WATCH IT! IT'S GONNA MOVE!!"
"This tree smells like cum."
Mitchell and Webb had so many legendary lines.
... And forever to play it in!"
I think that series came out slightly after that era ended, I remember as a child (so years earlier) realising that snooker players at the time were often tipsy if not actually drunk, and also darts players, but by the time I was an adult the professionalism really changed that, turns out that if you're very good you probably are better at snooker when sober. I think the commentators, being often a previous generation of players, lagged that slightly.
In the UK we watched it on the Beeb when it was aired.
Then I watched that show and learned about Numberwang.
https://youtu.be/6ofnDdi-vfw?is=y9ccxKyAQfqiPIQs
I'm glad that someone has stepped up to revive the legacy of numberwang.
The “Usage” section doesn’t involve installing any dependencies, for example: https://github.com/GraafHenk/numberwang#usage
With All Due Love,
Sir Digby Chicken Caesar, Esq.
You reckless fool
What it is: a character-level CNN with 80,804 parameters. The weights are a 1.79 MB JSON file and inference is about 100 lines of Python standard library — no PyTorch, no NumPy. It runs on a Pi Zero. It accepts digits, number words in eleven languages, arithmetic ("96 divided by 2", "deux fois trois"), Roman numerals, ordinals, clock times, currency, and fictional numbers ("shinty-six"). Anything with no numeric content is correctly ruled out as never able to be Numberwang. Whatever comes to 1 or 44 is Wangernumb and you rotate the board.
Held-out accuracy is 88.9% on 486 probes reserved from training by construction. The ceiling is ~98%, because roughly 2% of training labels are inverted at compilation time, in accordance with long-standing adjudication practice.
For comparison I ran Qwen3-1.7B on the same suite with the four verdicts as a constrained multiple choice: 51.9%, which is 2.3 points above answering "Numberwang" to everything. It answers "Numberwang" to 93% of inputs and never once identifies a Wangernumb. So the accuracy table has a verdict-distribution column, since one number can't tell a model that decides from one that agrees.
Honest weak spot: arithmetic is memorised, not computed. A conv net can't add. On operands reserved from training it gets 60% on symbolic expressions and 44% on foreign-language ones.
Dataset (185k adjudicated utterances), training script, evaluation harness and benchmark are all in the repo and reproduce from a fixed seed. Model card on HF: https://huggingface.co/graafhenk/numberwang