While I very much want what you describe (and it's why I still hope Framework eventually gets into the phone game), the software is a far far bigger problem than the hardware (and this is from someone who is _extremely_ disappointed in current phone hardware). I'd take free and open software over hardware at this point. Obviously both is better, but the software has a far bigger impact on our lives than the hardware.
Sailfish os is there if you want to pay, and definitely steam os is going there, they work on both porting android apps to steam (lepton) and emulating X86 apps to ARM (fex)
Going to compare this to our current solution of Amazon's Textract service for analyzing handwritten datasheets. Textract, when extracting tables (which is what we use it for) does not allow for providing any context or information about the tables and what we expect them to contain, but it is really good at correctly recognizing hand written characters. All of my attempts at less specialized, more general models allow me to provide that context, which is helpful in some ways, but fail at the basic part of almost always correctly getting the character.
As a daily R user, hard disagree. With the exception of ggplot (and this is directly related to why I don't use ggplot and instead use base plotting), most of tidyverse is pretty similar to and consistent with most base R functions.
Tidyverse standalone would be borderline useless, as most of what it's best at is manipulating, transorming, and re-arranging your data in various ways. You still need to _do_ something with your data at the end, at which point, the entire rest of the R ecosystem comes into play.
Tidyverse is valuable specifically because it's the best at doing what it does, and what it does makes everything else easier, more legible, and faster.
Forking it would simultaneously make both R and tidyverse worse off.
> most of tidyverse is pretty similar to and consistent with most base R functions
What? The main tidyverse packages are popular because they are different from base R. If the packages duplicated base R functionality and usage was the same then nobody would use them.
> You still need to _do_ something with your data at the end, at which point, the entire rest of the R ecosystem comes into play.
This is exactly my point. You could use tidymodels or any number of packages to keep your code tidy, but people just bail after wrangling their data a little and then their code is disconnected. You might as well have done all your data cleaning with base R if you were going to fit a model outside the tidyverse anyway.
What I meant was that they are syntactically similar. They work the way that default base R functions work. They _look_ like base R functions. They aren't the same as base R functions. They fit smoothly into base R, often filling holes that base R has. One can (and I do) use base R and tidyverse functions with each other all the time
This is as opposed to ggplot. Which legitimately seems like a completely different language. It looks, reads, and acts differently than base R plotting. It sticks out like a sore thumb, and, in my opinion, does not have enough functionality to justify the departure from standard R conventions. Which is why I don't use it.
As to restating your point: Your original comment combined with what you have said here makes me completely confused. The fact that people don't "stay" in the tidyverse is evidence that it is well integrated and _shouldn't_ be forked. You can use it for what it's good for, and then go use other things that are better at what they are doing.
If people regularly did the entire pipeline of import > data manipulation > data analysis and never left the tidyverse, then you would have an argument that it should be forked.
The fact that people dont do this is evidence that it belongs how it is: a package.
I don't really understand your comment about "disconnected". My code doesn't feel disconnected other than that different portions of it are doing different things. But then again, I also think that tidyverse functions don't look that different from base R functions (which, again, is not the same thing as being the same as already existent R functions).
This is an extremely niche problem that is probably not a factor for the vast majority of people: but my organization uses a shared dropbox account for file storage (yes, yes I know). The linux dropbox app does not have the smart download feature where you can see all files and folders but don't need to have them local unless you request them. The only options are to either download the entire dropbox folder, or to selectively sync certain files and folders, and then only be able to see those files and folders.
Given that the dropbox is some 4TB, but I often need to access things that I didn't previously need access to, this is a bit of a deal breaker.
You said it in your first sentence: you know that Dropbox is not designed to function the way you're using it. That's a kind of tech debt that may (will?) bite you in the ass eventually. Linux being incompatible with the way you use Dropbox is just a symptom of poor infrastructure and security practices, though I understand that it's probably out of your hands to fix.
Honestly curious: how many of those satellite dead-zones have good radio coverage? In my various times driving places, I've often lost radio signal in a sufficiently remote place where I 100% would have had satellite coverage. Those same features that can block satellite will also block (some kinds of) radio, if you don't have a broadcast tower at the top of the ridge or something.
Yes, I agree that satellite coverage is not 100%. But neither is radio.
It's definitely a mixed bag, but the areas I'm thinking of have decent FM radio coverage (from local stations that affiliate with public radio). AM coverage tends to be good regardless.
I've had the same experience as you around remote places, but those places were generally the flat-and-desolate kind :-)
I think the idea is that the sats relay a signal to radio stations including remote ones.
It's the radio stations who are in charge of situating their reception equipment where it can see the sat, and also for figuring out how to best broadcast to their served area (e.g. AM and/or FM? Tower height, power, setting up some translator stations on a different frequency to serve outlying areas, giving the feed to local cable systems to be sent with TV service, etc.)
At least they give complete control over AI summaries and allow the user to completely turn them off, and even when on, allow them to only be supplied when the user requests them (by appending a "?" to the end of a search).
I personally have completely turned them off as I don't think they provide much value, but it's hard for me to be to upset about the fact that it exists when the user has the control.
No, it's not. Regulations making building new power plants, especially renewables, and extra especially nuclear, in addition to making power lines difficult to build, are to blame. Yes, in an environment where power availability is ~fixed on short-to-medium time scales, adding a new large demand will increase prices.
But a fixed supply is a policy choice, and is not the fault of AI companies.
"No, the point of my comment is that, while that is true, that's not the "cause"."
If we changed the policy overnight, we would still have the same problems because infrastructure takes years to plan, build, and make operational. So no, the cause is in fact the rapid increase in demand, not just policy.
Now assume there were no such regulations and factor in the time it takes to actually plan, build, and commission a new power station and associated grid infrastructure. I'm not sure that your distinction matters in any real way.
>time it takes to actually plan, build, and commission.
This is, currently, mostly regulatory. Yes, in the absence of any regulations at all it would still take time to plan, build, and commission, and I am not advocating for literally no regulations, but solar and wind plants could probably be spun up in well under a year under a dramatically reduced regulatory burden, almost certainly faster than a new Datacenter can be built. They are, after all, dramatically simpler installations.
And that's not even thinking about the fact that in this alternate reality we are imagining, power plants would have been being continually built for decades, and the new demand would be a much smaller drop in the much larger bucket.
So I think that in an alternate regulatory regime both A) yes actually power plants could built ~ as fast as data centers and other large power consumers and B) we would have so much more power that increases in demand would be less of a shock to the system.
"And that's not even thinking about the fact that in this alternate reality we are imagining, power plants would have been being continually built for decades"
That would only be true if you could forecast the demand to justify the cost of the new infrastructure. It seems the demand from AI was beyond forecasts. The policies doesnt make the plants impossible to build, just slower. So your argument about continuously building plants is true in our current reality, and those plans include the extra time to comply with policies.
> And that's not even thinking about the fact that in this alternate reality we are imagining, power plants would have been being continually built for decades, and the new demand would be a much smaller drop in the much larger bucket.
Bullshit. Why would they have continuously built power plants if the demand wasn't there? The utterly insane level AI datacenter demand came out of nowhere.
And then you know, when there are tradeoffs, you can always maximize X and the expense of Y. And if you're myopically looking only at X, that may seem like a smart move, but that tradeoff may not be the right tradeoff when you look at things holistically.
And there are other tradeoffs: maybe not deregulate power-plant construction, but instead regulate AI data-center construction to slow it down. If we're in an AI bubble, that may end up being the right call and eliminate a lot of FOMO waste.
> but instead regulate AI data-center construction to slow it down.
The simplest and most logical regulation: don't connect new data centers to the grid unless they pay the cost and interest for the power capacity they commit to use - it's not hard to do the accounting for that and it's the fair way to do it for any large new consumers.
I don't even agree with partly. 100% of the blame, in my opinion, is on the policy-caused supply restrictions. I will admit that this is at least partially a semantic debate about what "cause" means, but in my opinion "blaming", even partially, AI, data-centers, or any other large power consumer for the price increases actively makes solving the problem harder and is anti-useful.
I thought there were technical and logistical hurdles to attaching new power while maintaining grid stability, and that's the primary bottleneck and indeed the reason the regulations exist to being with?
Why does it make solving the problem harder? If demand went up, it went up so build more supply. Talking about the cause of the demand doesn't hinder building more supply.
"and extra especially nuclear, in addition to making power lines difficult to build, are to blame"
I used to think nuclear reactors are just hard to build in general, because the costs when something goes wrong are very, very high. So what unnecessary regulation is there with nuclear reactors that you think should be deleted?
This is a much larger discussion, but the single most obvious one is getting rid of the Linear No Dose Threshold. There are an abundance of sources on why this concept is flawed and how it impacts nuclear regulation. It's not the only issue by far, but it's probably the single easiest to address.
In other words, allow higher exposure to ratiation?
Does not sound too great and obvious to me to be honest and it seems debated in the scientific community.
So irrational fear of radiation is surely a thing and maybe the models as to when real danger starts can be updated, but I would not call that question obvious when the experts debate it and I ain't one.
Nuclear Safety is extremely risk averse and the mortar in the bricks are incumbents for whom the strict regulations protect. Anecdotally, it is a very paranoid industry, for better or worse.
Allowing higher radiation dose does sound bad, but I would urge you to delve into the Linear No-Threshold Model. We have the lion's share of a century of cancer and health data and the results are somewhat counterintuitive.
So like I said, maybe the Linear No Threshold Model is wrong(I will have a look into the video as well). But it was presented here as something obviously flawed to get rid of .. while a short dive into it, showed it is still debated among the experts. Sp that approach from some people also does seem ideological motivated and not fact based to me, not just the anti nuclear crowd.
Sorry to double reply, I forgot to mention what is likely the source of this interest. Recently, Kyle Hill produced this 30 minute video explaining why we may want to re-examine the LNT dose model https://www.youtube.com/watch?v=gzdLdNRaPKc
This is why they are so risk averse, there indeed are incredible dangers and if varies per place.
Operators, manufacturers and service have spent a long time making what we have very reliable in what are now pretty old designs. If a new pump manufacturer appeared in the scene, everyone making decisions needs to assess the reliability vs a well-known quantity of reliability.
This is what I mean when I say risk averse, and the mortar in the bricks are the suppliers and services. The record doesn’t show worse and worse throughout time, everyone in a nuclear safety related industry knows what to expect and what is expected of their production. Changing even this linear no-threshold model would incur a LOT of engineering, process development/improvement and risk analysis, which none of them want to do.
If you’re the one that screws up, it can be a nasty stain.
if your best argument is just "Google it" I'm gonna go ahead and assume you don't know what you're talking about and are just making an appeal to authority
Removing regulations from nuclear won't help because it takes so long to build a nuclear plant. Yes it would help in the long term, but in the short term price goes up.
However it should be faster to build a solar plant, a battery bank and a power line between them and the new data center than it takes to build a new data center. It isn't because of silliness, and that's what to blame for the power price increases IMO.
Removing regulations 20 years ago when people were screaming that we need to would have helped. 20 years ago nuclear was still the best answer, but since it wasn't allowed we didn't build it (instead mostly coal and gas). Power companies are good at planning, and AI/data centers are not using that much more power than predicted 5 years ago - but the regulations have to allow for plans.
Removing regulations today will have an effect in 10-20 years. I cannot give you a quick answer to the problem.
So supply and demand only matter for the axis you personally care about? AI companies use a lot of electricity. Increased demand leads to increased prices. This isn't normally controversial.
So the reason my electricity rates were pretty much constant from 2013 to 2023 and then started going up is not the things that have changed in the last couple or so years but rather the things that have not changed?
Conservatives promoted a worldview where corporations are expected to do absolutely anything that isn't illegal to increase shareholder value. In such a world, regulations are the only way to protect ourselves from corporations that would gladly kill us to make a buck.
Seems to be that big companies usually push to externalize costs and take advantage as much as possible, to the detriment of everyone else. Shareholders have a right to sue them if they don't.
Fake cures, filthy mines, toxic ingredients, polluted waterways, fraud, predation, monopolies, algorithmic social outrage networks, etc. These things have been going on for a long time. Regulations have fixed a lot of problems.
It doesn't seem that reputation matters as much as regulation. It doesn't take many greedy people/companies to leave behind a big mess. Just a few breaking the rules (including the cultural rules around reputation) gain an advantage over all those who don't.
The executives are trying to maximize shareholder value. They are shareholders through options at the least. The board can also own shares (which should be illegal).
The executives are trying to make more money. They get paid in shares so making the share price go up makes them more money. Whether or not the means used to achieve that end are beneficial to other shareholders is beside the point. This is potentially a violation of fiduciary duty.
> They get paid in shares so making the share price go up makes them more money. Whether or not the means used to achieve that end are beneficial to other shareholders is beside the point.
The point was, increasing shareholder value is equivalent to increasing their own paychecks. So they are doubly incentivized to choose shareholder value over other values. Most people would still choose health/life of other humans, but the incentives are certainly off without regulation.
Even with prolific supply there's tons of government programs that subsidize things and then roll those things back into delivery and transmission costs.
The poors pay for rooftop solar and heat pump subsidies for HNers.
it's to blame in the sense that there is a counterfactual reality where the AI companies pay for their own power and your bill doesn't go up and we can pass a law to make that counterfactual real. but yeah, blame is supposed to be about assigning responsibility. the change is attributed to AI in the sense that if they didn't exist it wouldn't have gone up, but technically the responsibility here is on policymakers to do something now that we are aware of the attribution, and they deserve to be blamed if they don't. blaming AI companies directly is a contemptuous mindset that blames them basically just for existing. which might be cathartic but it's not useful.
Why is AI demand any different than other business demand? What you're advocating for is intentionally handcuffing a growing industry for no reason other than you don't like them.
because they're driving up electric prices disproportionately...
the argument is to handcuff them because of the externalities. which is one of the things laws are for. It's not about fairness it's about whether this is the world we want to live in. The market was designed a certain way; the design stops benefitting the public the way it should; so update the design, easy.
Any industry that stresses public infrastructure is in the same category. They all should be regulated and not handed, in the form of tax breaks, what should be public money to invest in additional infrastructure.
100% this. I'm sick and tired of alarmist news and scapegoats when politicians and greedy energy corporations are to blame for everything. Yes, AI consumes more energy because we're using AI so by this logic we are to blame for everything.
If these "greedy utility companies" were such good monopolists or duopolists wouldn't it reflect in some pretty insane stock performance?
Eversource (NYSE: ES) is my local electric/natural gas provider in Massachusetts that I hear these same arguments about. Their stock is down 21% over the past 5 years. (To contrast, the S&P500 is up 91% over this same timeframe).)
Regulations also made coal more expensive and forced many plants to close. You can argue that's a win, but it's a lie to then attribute the resulting price increase to AI or other factors.
AI has done bad things for humanity. I know, I know, a tough pill to swallow. However will Hacker News users cope with the trauma?? Of knowing... AI... can be bad... sometimes?
I don't see why you couldn't do it in either case. If you modify the actual price, then you are giving exact change. Why wouldn't round() be as valid a price modification as floor()?
Presumably "increase the price a small amount to avoid giving exact change" is exactly the sort of thing that laws requiring giving exact change were designed to prevent.
There will surely be some customer pissed about the extra 2 cents they were charged who will raise hell over the exact change law.
But what customer is going to be upset over a small discount?
I've tried various stinging nettle dishes. For me, they fall into the long list of things that are technically edible, but I see no reason why one would eat them today aside from cultural history/connection/tradition reasons. Every way I've tried them, they basically just taste like plant.
If you like it, great, but I think the value to those who don't have some pre-existing reason to be interested in the dish to be overstated. Similar plants in the category are miner's lettuce and dandelion greens.
If one has a great abundance of it, and one likes to spend time preparing ones own food, or if the idea of wild gathered plants has special appeal, then nettles (etc.) can indeed take the place (ish) of things like lettuce and spinach, but don't expect some dramatically unique experience.
I have no pre-existing reason, I just like the way nettles taste. Though since we have eaten them for a long while, they are available in bags in the grocery store every summer, which makes the whole thing a lot easier.
There is a significantly easier option (although still more work than just buying a vacuum and using it as the manufacturer intended): get one of the Valetudo supported vacuums[0]. This firmware replacement blocks telemetry and allows for near complete feature parity with the original firmware, and flashing is (usually) relatively simple. Certainly much simpler than the process described here.
reply