It's also a common tactic for filtering inbound email.
Mention that people may optionally include some word like 'orange' in the subject line to tell you they've come via some place like your blog or whatever it may be, and have read at least carefully enough to notice this.
Of course ironically that trick's probably trivially broken now because of use of LLMs in spam. But the point stands, it's an old trick.
It's not so much a case of personal targeting or anything particularly deliberate.
LLMs are trained on the full internet. All relevant information gets compressed in the weights.
If your email and this instruction are linked on your site, that goes in there, and the LLM may with some probability decide it's appropriate to use it at inference time.
That's why 'tricks' like this may get broken to some degree by LLM spam, and trivially when they do, with no special effort on the spammer's part. It's all baked into the model.
What previously would have involved a degree of targeting that wouldn't scale now will not.
In my experience basically everything being good in software is downstream of good data modelling.
It's partly why I've realised more over time that learning computer science fundamentals actually ends up being super valuable.
I'm not talking about anything particularly deep either, just the very fundamentals you might come across in year one or two of a degree.
It sort of hooks back in over time as you discover that these people decades ago really got it and all you're really doing as a software engineer is rediscovering these lessons yourself, basically by thinking there's a better way, trying it, seeing it's not better, but noticing the fundamentals that are either being encouraged or violated and pulling just those back out into a simpler model.
I feel like that's mostly what's happened with the swing over into microservices and the swing back into monoliths, pulling some of the fundamentals encouraged by microservices back into monolith land but discarding all the other complexities that don't add anything.
I agree, so much so that about 10 years ago I built a product that did this!
I launched to lukewarm reception, actually applied to YC with it and didn't get much of a look, nor an interview :-) and after a bit of (though certainly far too little) further hustle gave up on it due to circumstances leading me on another path.
Anyway, I was a tiny bit vindicated when about a year later I noticed Stack exchange themselves did a similar product, but as far as I know, it never really hit. They would advertise it in the side banner for quite a few years but it eventually seemed to go away.
It's weird that it didn't work, it always did seem like an incredibly good idea to me - just so good, it's obvious. If such a thing existed, it'd add so much to any company onboarding experience at a minimum, and would also have obvious ongoing value.
And it just seemed like a great strategy to get useful and up to date documentation: to gamify it. There's just an inherent incentive to become the 'Jon Skeet' of your organisation as it were, rather than making documentation this largely anonymous, thankless afterthought it often becomes in practice despite best intentions.
If it's any consolation, I feel like I've been pitched 10 different versions of this product over the years and I've encountered a lot of startups trying to do the exact same thing. You probably didn't get any YC traction because they'd seen the same thing so many times before. I wouldn't be surprised if we could find a YC batch or two that already contained this exact idea.
it's probably subjective, but I find these collaboration features can be overused for this kind of thing.
If someone is walking me through something, I just want to see what they see so I can focus entirely on what they're saying and no part of me is distracted by having to follow along or seeing other code.
I know typically these collab modes have an auto follow feature, but it's not as simple as just read only video being streamed to you, there's loads more ways it can go wrong and add noise / distraction that provides no benefit.
I bought a copy on launch and working my way through. There's some really excellent stuff in here, fiction is a great way to analyse and get thinking about how the near-ish future could look, and that's what the theme of this collection is.
That's what makes Black Mirror so compelling. It's not flights of fancy with things like FTL traveling through space. It's taking slight twists to the darker side of where our tech leaders are taking us. At least someone is thinking these things through to the end even if our teach leaders are only thinking through to the $$$$. Too bad these stories are never enough to slow things down and used as "oooh, let's not let our tech come to that".
Yeah a lot like older sci Fi works I think we won't take the warnings. Glad we aren't completely blind to it tho. Remind me of the "1984 was not a manual" memes
OH that's the fantasy that people like watching that show under! I hate that show's premise because reality is depressing enough, my mind easily conjures up shit going wrong, and (as an adult) I don't need a TV show giving me more nightmare fuel. Plus, that one episode where we all wear glasses that let you delete people by not seeing them seems like a instruction manual to tech giants working on AR glasses. Apple airpod noise cancellation already does this for audio, Apple Vision or Meta glasses doing that for visual pollution seems not that far behind. It would be nice to walk around the neighborhood and not see any billboards. Not seeing people that wrote wrong things on Twitter seems like it's just a GitHub repo away.
Meta would never block ads. If anything, they would detect things like billboards and have the AR system replace them with ads bought through Meta's ad system. But blocking them completely so the user has an ad free life experience? Never.
Meta wouldn't just not block ads, they'd use the eye tracking system in the apple goggles to charge advertisers for time you spent actually looking at the ads, and then also the ad countdown timer would only count down while you were actually watching the ad. Look away from the ad and the ad stops playing! (See? No black mirror necessary for my brain to dredge out some dystopian ass shit)
That actually does happen in Black Mirror, in the episode Fifteen Million Merits. There's a sort of deafening siren that goes off if the characters look away from the ad without paying to skip it.
> (See? No black mirror necessary for my brain to dredge out some dystopian ass shit)
oops. you poor poor fella. haven't we pretty much gotten to the point that if you've thought about it, someone else has also already thought it and released it? at least, that's my core belief in any thing clever I think I might have just thought.
You're absolutely right! let me get back to you after these messages from Coca cola!
I'm not an absolutist. How would you find out about new products in a world without advertising? There are ads I could do without, sure,
but some do perform a useful function.
It seems a bit of a stretch to separate this from the ordinary sense of touch.
I mean, feeling sand compress in subtle ways and being able to map that mentally to an object that might be hidden in the sand seems like literally touch plus normal world modelling / reasoning.
Couldn't you describe that effect where you can reliably guess the size and other features of things by sound without seeing them as a seperate sense? Well, it's not, again it's just a combo of a sense plus mental modelling / pattern recognition.
> I mean, feeling sand compress in subtle ways and being able to map that mentally to an object that might be hidden in the sand seems like literally touch plus normal world modelling / reasoning
That seems like a very strong claim against the paper’s results. What makes you think that the study participants located the cube with reasoning, rather than unthinking sense?
I think we can be too quick to write things off as somehow coming from conscious thought when they bypass that part of our minds entirely. I don’t form sentences with a rational use of grammar. I don’t determine how heavy something is by reasoning about its weight before I pick it up. There is something much more interesting happening cognitively in these cases that we shouldn’t dismiss.
This is like wondering if we calculate parabolas consciously before catching a tossed ball. We had to learn its behavior without knowing the physics, but it becomes unconscious soon. If tossed balls behaved like they do in cartoons, we'd learn to predict them even if they violated the laws of physics.
I still wonder how we practiced finding things by distinguishing the fluidity of a medium around them. Maybe playing in water?
The actual levers of control available to those in charge in schools are limited, in the end.
The rules that exist are routinely broken and can only be enforced selectively. Many of the rules are unpolicable frankly and are only kept to or only marginally broken as a matter of social norms, and understanding so there is not total choas. An equilbrium is found.
With phones there's such social pressure to allow their use, including from forces external to the school, that there was never possibly a hope of the equilibrium immediately settling at phones being banned.
It was always going to creep to the current status quo. Again this would have been true even if a rule were ostensibly set.
Society is learning, slowly, that this isn't ideal, and the pendulum seems to be swinging back. It may settle at phones being completely banned in schools, but in practice this will also obviously be moderately chipped away at all the time in various surprising and unsurprising ways. Especially as the hardware itself evolves.
> With phones there's such social pressure to allow their use, including from forces external to the school, that there was never possibly a hope of the equilibrium immediately settling at phones being banned.
Phones, and electronic devices in general, were always banned. What changed was schools started allowing them.
I was in high school right when some kids first started getting (dumb) cell phones. MP3 players were still new, CD players were not uncommon, and ALL of them were banned from being outside of your locker or backpack. If a teacher saw one, it was gone until the end of the day. Period.
Teachers didn't need to bear the brunt of angry parents, it wasn't their call to make. That belonged to the school administrator, who merely needed to say "tough shit". Somehow, the adult children still won anyway.
That’s what I initially thought, but I could see it being cheaper for the railway company to build solar for themselves. They get the power at cost, and can build a system that can handle lower uptime/greater availability, with storage etc, so that they can avoid paying to upkeep high availability power.
I don’t know though. I assume someone with a really nice spreadsheet has looked at this before deciding to pay to build all this.
I would think a huge additional advantage of this solar and on-car batteries combo would be resiliency to single points of failure and massive long term savings in infra and power costs?
Trains are famously risk-averse designs. They're also really expensive and have to work for decades.
It will be hard sell to convince a train operator that adding literal tonnes of batteries to a train with a complicated mesh of solar microforms spread over thousands of km of track will be a good deal in terms of regulatory approval, cost effectiveness, maintenance overheads in time and money and that your company will still be around in 20 years to run the system rather than going belly up in a tech crash and leaving them with a bunch of bricked hardware.
The new TGV-M already have batteries inside. When you have an issue with the catenary for any reason, the batteries can help keep the train momentum without risks, or if the train had to stop, allow you to start from 0kmh to 80kmh and keep the speed for 5 hours, with air conditioning and all. Your 2 hour trip now last 5 hours, but still. And since in my country every 10 km on major lines you have a "different" catenary circuit, the 80km/h speed will probably last 15 minutes and you can go back to 280 fast enough. I think the first was deployed this year, it it will be be available to export in 2-3 years.
Why should this setup be locked in to any particular third party company?
How different is it to the train cars themselves being built by some company and then maintained by train company mechanics? And battery + electric motor setups are famously pretty straightforward, so this seems even more true.
I don't disagree on the regulatory stuff being an issue, that's a very good point.
The solar mesh feels like an orthogonal problem, also feels very solvable with just standard interfaces which would surely be mandated. I mean the solar bit is an add on too, you can always fall back to other means for battery charging, but seems like solar is just an obvious economically optimal solution in the circumstances.
I'm not sure I understand why it's necessary to even couple this to a runtime, let alone own the runtime?
Can't you just do it as a library and train/instruct the LLM to prefer using that library?
reply