The first time the ratio of length to width was written in a letter dated 25 October 1786. This letter was from the German Georg Christoph Lichtenberg to Johann Beckmann. He wrote here about the advantages of basing paper on a √2 ratio. Lichtenberg is known for the ratio between length and width of a surface which remains the same after the narrated halving of the surface. The result is 1:√2.

We propose a simple solution to use a single Neural Machine Translation (NMT) model to translatebetween multiple languages. Our solution requires no changes to the model architecture from a standardNMT system but instead introduces an artificial token at the beginning of the input sentence to specifythe required target language. The rest of the model, which includes an encoder, decoder and attentionmodule, remains unchanged and is shared across all languages. Using a shared wordpiece vocabulary, ourapproach enables Multilingual NMT using a single model without any increase in parameters, which issignificantly simpler than previous proposals for Multilingual NMT. On the WMT’14 benchmarks, a singlemultilingual model achieves comparable performance for English→French and surpasses state-of-the-artresults for English→German. Similarly, a single multilingual model surpasses state-of-the-art resultsfor French→English and German→English on WMT’14 and WMT’15 benchmarks, respectively. Onproduction corpora, multilingual models of up to twelve language pairs allow for better translation ofmany individual pairs. In addition to improving the translation quality of language pairs that the modelwas trained with, our models can also learn to perform implicit bridging between language pairs neverseen explicitly during training, showing that transfer learning and zero-shot translation is possible forneural translation. Finally, we show analyses that hints at a universal interlingua representation in ourmodels and show some interesting examples when mixing languages.

so let's suppose let's suppose your listeners are with me and you know we kind of agree like okay yes transformation's necessary and uh again i want to emphasize i'm not talking about reform i'm not talking 00:58:59 about a softer better capitalism i'm not talking about you know improved voter registration or like any of those things i'm talking about de novo starting over from scratch what might be 00:59:13 best and if it turns out that the old systems were better than anything that humanity can come up with well then you know that's the answer but i can't imagine that's true because the old systems were never designed in any kind of 00:59:25 you know thoughtful science driven [Music] you know process to to to test to explore and to come up with fitness like what is the you know we don't even have a fitness for our current society 00:59:39 much less of fitness for societal designs i mean we have the gdp but that's a terrible terrible limited fitness metric 00:59:51 okay so suppose you're with me suppose we're we're on board we we want to do this de novo design thing where do we start what's the what's what where do we even get off the 01:00:03 ground on this and i suggest that the way to do it is through first address worldview from world view once we understand what the world view is 01:00:15 what a reasonable useful world view will be for this project then then purpose derives worldview begets purpose once you understand what it is you want 01:00:28 what you value what do you value once you understand what you value then you can say well i value a and therefore the purpose is to 01:00:39 have a manifest in society for example so once you have purpose then you can think about what metrics how would you measure whether are you so 01:00:53 here's a new design is it fit for purpose does it do does it fulfill its purpose you know that's the question and then metrics go with some kind of fitness evaluation 01:01:05 and then finally last of all of those would be the design okay we know what we know what we value we know what this thing is supposed to do we know what the purpose is we know that attractor is supposed to you know plow the ground or something we 01:01:18 know what this is supposed to do we know how to measure success and uh now finally then let's talk about design what are the what are the you know the specifics and mechanics and 01:01:31 how does that happen and the the series is really kind of laid out this way the first paper really talks about world view and purpose the second paper talks about the you know the more the mechanics of things 01:01:44 like viability how would you make this thing viable things like that and then the very last paper that's titled the subtitle design okay so uh that's how we uh and 01:01:56 and maybe i will just mention here that i put metrics before design because we might have some ideas uh getting back to that preference factor we might have some ideas like we would like people not to die at 01:02:08 30 you know we'd like people to mostly live to a ripe old age and have you know enough water water to drink and food to eat and all that kind of stuff so uh you know what kind of design once 01:02:20 now that we have metrics to measure that kind of stuff longevity and nutrition and things what kind of designs would help us to reach those targets you know so that's one reason why design 01:02:31 why metrics comes before design okay