This is how I was taught. Use ( ) or -- -- here and the Oxford comma for list of 3 or more.
I get lazy with adding the comma before the "and" in list, and without fail I hear my grandmother/father/teachers pointing out how wrong I am for doing so. Same for my use of semicolons followed by "and" or "but".
I never realized the Oxford comma was even something up for debate.
Many years ago working on natural language to SQL, when we had ambiguities this is how we’d clarify things with the user (albeit with the minimal amount of brackets necessary).
> Two-thirds of Americans now believe that professional athletes sometimes change their performance to influence gambling outcomes.
I'm not sure this is a bad thing. It's just bringing to public visibility exactly what happens across the stock market. Public companies do this all the time -- engineer their performance end earnings to influence <strike>shareholder</strike> gambler expectations on earnings day.
Also a fellow astrophotography enthusiast here! I love to photograph deep sky objects in the context of their landscapes. There is a lot of math, stacking, tracking, and denoising in the process, but I keep every image very real as what you would see if your eyes were a lot more sensitive. A lot of people don't realize how big some objects are in the night sky -- for example, Barnard's Loop is as large as about half the entire constellation of Orion, and Andromeda appears 6 times the size of the moon. We just don't see them because many of these objects are very dim -- not small.
So give free solar to selected homes with 1 km spacing. Either the government or solar companies should be willing to sink this cost in order to get the additional business.
My Tesla long range gets about 60% of advertised range in real world conditions. I'm talking stop signs every block, mountains you need to drive across, insanely hot days, i.e. the real world.
I knew that would be the case, but I really wish there was a crackdown on this. Advertised range should be the mean of the distribution, not the max.
In fact EV manufacturers should be required to publish the distribution and they should have to pay a KL divergence penalty on it that will be distributed to EV buyers as rebates. It would also require the courts to learn about KL divergence, which I would really love to happen. We need countries run by engineers, not clowns.
> Advertised range should be the mean of the distribution, not the max.
Distribution of what? Assuming you mean the distribution of driving range achieved 'in the real world', how would that work before a car is sold? How often would it have to be updated in their advertising material? Over what sort of area would the distribution be calculated? How would anyone know if the advertised range of two different cars was even comparable?
Whilst the standardised tests could be improved, they are still the best way to compare products.
> should have to pay a KL divergence penalty on it that will be distributed to EV buyers as rebates
I get about 15% more range than advertised, should I have to pay a penalty for this?
This really sounds like 'but think about a poor car vendor!'. And a poor car vendor definitely can't build at least 10 pre-production cars, run them with both a lightest and heaviest loads and different patterns and calculate the mean and use it instead of the one with the maximum distance with a minimal load, right?
> This really sounds like 'but think about a poor car vendor!'.
It was absolutely not meant to come across that way. I just think it wasn't thought all the way through.
> And a poor car vendor definitely can't build at least 10 pre-production cars, run them with both a lightest and heaviest loads and different patterns and calculate the mean and use it instead of the one with the maximum distance with a minimal load, right?
This just sounds like vendor controlled slightly-less-standardised testing, not the real world based system they seemed to be arguing for.
> I get about 15% more range than advertised, should I have to pay a penalty for this?
No. You are just one datapoint within the distribution. If the distribution aligns with manufacturer's advertised distribution, nobody gets a rebate. If distribution is not aligned, manufacturer is penalized and everyone gets a rebate for being misled.
There are many variables and scenarios, yes.
This, however, is not an excuse not to provide some more data points that help people estimate what they are really getting...
Anyway Tesla has data from all their cars, they could use that.
> If there's something most people miss just state it.
But the LLM suggesting a question doesn't mean it has a good answer to converge to.
If you actually ask, the model probabilities will be pressured to come up with something, anything, to follow up on the offer, which will be nonsense if there actually weren't anything else to add.
I've seen this pattern fail a lot on roleplay (e.g. AI Dungeon) so I really dislike it when LLMs end with a question. A "sufficiently smart LLM" would have enough foresight to know it's writing itself into a dead end.
You should be careful with ideas like "sufficiently smart LLM" - quotes and all. There's no intelligence here, just next token prediction. And the idea of an LLM being self-aware is ludicrous. Ask one what the difference between hallucinations and lying is and get a list similar to this why the LLM isn't lying:
- No intent, beliefs, or awareness
- No concept of “know” truth vs. falsehood
- A byproduct of how it predicts text based on patterns
- Arises from probabilistic text generation
- A model fills gaps when it lacks reliable knowledge
- Errors often look confident because the system optimizes for fluency, not truth
- Produces outputs that statistically resemble true statements
- Not an agent, no moral responsibility
- Lacks “committment” to a claim unless specifically designed to track it
I have AT&T Fiber and 99% of the time it's fantastic, but there are several instances of 30-60 second downtime a day and I have a 5G modem with a Google Fi data sim as a backup. Failover is nearly-instant with a Unifi UDM.
The data sim costs nothing extra on top of my cellular plan and just counts towards my (already very generous) monthly limit of 50GB.
The one thing I hate about monorepos is nothing ever gets versioned, packaged, and shipped.
Polyrepos are workable, the way to do it is to actually version, ship, and document every subcomponent. When I mean ship, I really mean ship, as in a .deb package or python wheel with a version number, not a commit hash. AI can work with this as well, as long as it has access to the docs (which can also be AI-generated).
The monorepo make it easier to ship the overall product but harder to ship parts of it.
I've used a monorepo for the past 13 years and I got all shared packages with version 0.0.0 and I still haven't figured out a simple way to share just some parts of it like a CLI.
Does anyone have a monorepo and publishes NPM packages with source code of only that folder? Sub-gits required to pull in multiple places...
The best thing about monorepos is nothing ever gets versioned and packaged.
That means, a subcomponent can just make a needed change in the supercomponent as well, and test and the ship the subcomponent without excess ceremonies and releases.
That strategy sucks when you have 15 supercomponents depending on 1 subcomponent, you want to rewrite the subcomponent, and now you have to deal with all 15 of them and get codeowners approvals from all 15 of those teams.
In a polyrepo org you just release version 2.0 of the subcomponent and upgrade the 1-2 of the 15 teams where the 2.0 features are actually needed now. The rest of the users can upgrade on their own schedule.
I've got about ~15 repos for a project and I just start Claude Code in the parent directory of all of them, so it has clear visibility everything and cross-reference whatever it needs.... super handy.
reply