It wouldn't be different at all. Mostly we did use units of twelve for everything. Why do you think there are twelve inches in a foot and twelve pence in a shilling?
Asking what the historical difference would be if we wrote our numbers in base 12 instead of, variously, 10, 12, or 20 (all historically common, and 60 is prominent too), is like asking what the historical difference would be if we wrote our words in Greek letters instead of Roman letters.
Note that it was common for people to count dozens on their hands; each hand has four non-thumb fingers with three knuckles each.
Of course it would be different. I'm well aware of historical numbering systems, and their applications.
My point is that if base-12 was biologically natural, instead of more effortfully useful, there would be many differences in the way we do things -- although of course we would be mostly unaware of them, as a fish in water.
There would be no metric-vs-imperial units dichotomy, for example. (EDIT: Or at least the conflict would be different and likely lesser, ergo easier to switch)
NASA probably wouldn't have lost the $327MM Mars Climate Orbiter. And it wouldn't have cost $327MM in the first place.
In some cases, unit sizes would be different. That's the easy case. But in counting systems, 100 of some atomic thing would be 44% more than it is today. 1000 would be 73% more. 1MM would be almost triple. Given the attachment people have to round numbers, this would have implications. Some things would be bigger. Other things would be considered in different increments.
> Given the attachment people have to round numbers, this would have implications. Some things would be bigger. Other things would be considered in different increments.
No, they wouldn't. This would be a very minor effect, because the primary determinant of sizes and amounts is how big you need something to be, or how much of it you need.
Instead, you'd see the same thing we already do see: contexts in which an existing unit was difficult to work with would be given their own units of a more convenient size. Consider how horses are measured in increments of four inches, or how soft drinks are sold in unit sizes of 12 ounces, 20 ounces, and 67.6 ounces.
The units aren't called that, of course; those sizes are "one can", "one bottle", and "two liters".
Some unit sizes are calibrated to geometric affinity (temperature degrees to 180 or 100, currency to 100, etc), and some are not (soft drinks).
It's likely that in a base-12 world, we'd still have "100" "cents" in a dollar. But there would be 144 (base-10) of them. That alone is interesting! But it's fine, because the collation is geometric, but the cent unit size is arbitrary.
Similarly, who knows or cares how many base-12 fluid ounces would be in a food product -- the collation (packaging) would be humanely-sized regardless.
But some units are not arbitrary, or conveniently divisible, and some collations still tend toward geometric affinity. Humans have their idiosyncrasies.
Would there be 10 base-10, or 10 base-12 amendments in the US Bill of Rights? What would the additional two amendments be?
Would there be 10 base-10, or 10 base-12 commandments in the Hebrew Bible?
Would we talk about a US president's first 100 base-10, or 100 base-12 days? Would the extra 44 days matter?
Again, none of this is profound. But if you think it isn't interesting, we'll just have to disagree.
> Would there be 10 base-10, or 10 base-12 commandments in the Hebrew Bible?
This is a case where there's no difference. Right now, Wikipedia's page on the Ten Commandments notes 15 different commandments. (Actually, 16, but two of those are "thou shalt not covet thy neighbor's house" and "thou shalt not desire thy neighbor's house", and Wikipedia knows of no tradition that thinks of those two as separate commandments.) Here it is obvious that the number of commandments is significant, at 10, and the content of the commandments is significant, but unrelated to the number. If you want 12 commandments instead, it's sufficient to renumber the existing 15 commandments, and this is clearly how the problem has been approached historically.
The same thing applies to the Bill of Rights; there are a lot more than 10 enumerated rights, but they're bundled into 10 amendments.
> Would we talk about a US president's first 100 base-10, or 100 base-12 days? Would the extra 44 days matter?
This might make a difference; 144 days is close enough to half a year that you might just mark the half year and not bother with the 144 days. Or you might mark the quarter-year, which is almost 100 days.
I seem to remember the explanation is that A and B are somewhat recent mutations in evolutionary terms, they simply have not had time to cancel out 0 yet.
Of course there's crackpot theories (aliens!) too.
The sad truth is - all of classical nlp is Dead, with a capital D.
The bottleneck for accuracy was always data quality and human effort, not model architecture.
Llms make the data and human problems so much easier, that the benefits of supporting different architectures just doesn't make sense. With quantization, I'm not even sure classical models win out on cost anymore, and they had already lost on (real world) accuracy.
LLMs are the O365 subscription that you just can't fight against with bespoke mini solution. An all in one solution is simply too appealing.
Also, if you have to learn pre-2020 NLP I would just learn to use spacy. It pretty much covers all of pre-2020 NLP out of the box in a well documented package with strong GPU and CPU support.
> The sad truth is - all of classical nlp is Dead, with a capital D.
If your language is well served by LLMs. Which, generously, is true for maybe 20-50 of the world's 7000 or so languages. For all the rest, "classical" NLP is still how things get done at least at present.
So if anyone wants to learn language modeling stuff, do you recommend starting with transformer and just learn how to deploy and finetune LLMs (given that ordinary people can't train these LLMs)?
A lot. POS taggers used to be linear classifiers + features. In 2018 they switched to BERT and similar encoder-only models. In 2023, POS tagging is largely irrelevant, because it was used as a part of a larger pipeline, but now you can have everything end-to-end with better accuracy by fine-tuning a sufficienly large pretrained model (LLM or encoder-decoder like T5)
A 1-2 hour process which includes heating to soften glue, high potential of damaging internal components, or high potential of cracking some piece of glass is not "Screws are how in fact one accesses the battery in an iPhone."
> “A portable battery should be considered to be removable by the end-user when it can be removed with the use of commercially available tools and without requiring the use of specialised tools, unless they are provided free of charge, or proprietary tools, thermal energy or solvents to disassemble it.”
possibly, yes, it’s certainly an argument that could be made in court. However the battery itself can be removed without those. It’s the rear glass that can’t (currently, who knows in 4 years).
Sounds hair splitting I know, but court cases are won & lost on less.
Not here in the US. The carriers wanted to discontinue 3G for years. They really only started it in earnest in 2018. It took them about 3 years longer from the first dates given to finally kill 3G. It sound to me as if 3G phones didn't bite the dust as fast as carriers assumed or customers refused to upgrade at the rate they expected.
> you're definitely not seeing genuine AirTags 5 for $13
I didn't see anything in the comment which suggested they were. It is specifically points out there are devices not produced by Apple. Third parties can create compatible devices:
How? Airtags only work because Apple uses any iPhone with BLE support as a detector, whether or not an app is installed. An off-brand BLE beacon would require an app to work at all, and I highly doubt Apple would make it easy for a third party app to run all the time in the background.
According to https://developer.apple.com/find-my/ manufacturers of third party hardware CAN leverage the iPhone BLE network to connect to the FindMy network.
The advantage of AirTags is the large phone tracking network: every recent iPhone defaults to tracking them. That's hundreds of millions of devices globally, with probably even more density in the US. The main competitors (Tile and Samsung) have much smaller networks. I'm not sure what network you'd get with generic "Bluetooth Tracker" tags: technically it could be Apple but I can't find any evidence for such products online.
> That person specifically noted and gave you a reference to how those 3rd party tags use Apple's own network.
What is the clear reference? One person said they searched GPS tracker tags on Amazon and the other gave a link to Apple’s licensing program which in turn links to NDAs and contractual agreements as well a multi-step design and approval process. There seem to be some knock-offs claiming they work with Find My, but there’s all sorts of weird counterfeit crap on Amazon. Do you have links to a product that’s actually licensed/supported by Apple (ie works and will keep working), because that’s what surprised me.
I own one, and it works just like an airtag and appears in the find my app. It doesn't have precision feature but everything else works(like using the apple/iphone network to find it)
Like regular citizens around the globe that are taxed for any reason the government wants. Taxes were also used as a way to express the power of the mighty during most of the history, the original meaning of an income source for shared public services (aka "government") was long lost. If one thinks about universal oppression, discretionary taxes would be number one.
Satya said he was playing with it last summer (I think in the verge?), they’ve had some dev time. And Bing has been testing some kind of chat bot for months, presumably not ChatGPT based.
Strangely, O is very common.