My AIntern is available the moment I start my day (not normal for off-shore in a different time zone) _and_ is available after the end of my day when I want it to work after I've gone home (useful for work using scarce resources). It even keeps the context across that time span.
I think a lot of the cost comparisons to employees are off by a factor of 2 or more. AI is the ultimate contractor. Available instantly. Doesn't charge during idle periods. Pre-vetted and pre-trained. No contract negotiations or complex accounting.
That is worth a small multiple of the fully-loaded employee cost. So AI might be easily worth more than $200 per human-equivalent hour. With high utilization, that might be $8000-10000 a month.
With that kind of spend, AI provider financials looks less frightening.
Bed length is a silly metric to use. It's single dimensional and doesn't capture any of the other important metrics. Such as width, air-volume, cubed-volume, or weight capacity.
Slate's 5 foot bed is on the shorter end of common bed sizes. Certainly shorter and worse on every other metric than the F-150/Silverado/1500 'short bed' so commonly seen hauling air in the US.
That presentation doesn't support your claim. The closest it gets is that solar attached to 4 hours of batteries is, ignoring tax credits, about (it's hard to read accurately from the graph) ~8% more expensive than combined-cycle plants.
But 4 hours isn't near a full night. At least 12 hours of battery storage would be necessary for that, possibly more depending on light angles and the relative supply-versus-demand loading at different times of day.
Roughly from the graph on page 8, that 4 hours of battery costs $22/MWh over solar alone. Presuming no further solar panels were needed, extending that 4 hours to 12 to cover the night would cost around $44/MWh more, bringing the total cost of 24h-reliable solar+battery to around $97/MWh -- WITH tax credits. Without tax credits it would be $20-$30 higher, but the graph is too low resolution to be precise. That compares poorly to the $65/MWh for combined-cycle for one single night -- which gets no tax credits accounted for in that graph.
A big part of it is the industry standard for using the Levelized Cost of Electricity (LCoE) as the benchmark metric. By that metric, solar IS the lowest cost power source.
But that definition doesn't take into account availability. This wasn't a problem when all electricity sources were highly available by default. You can burn coal or run the hydro turbines any minute of the year. With the rise of often-unavailable renewable sources like solar and wind that definition is now insufficient and under counts the true like-for-like cost of solar.
By any metric which takes into account minor availability requirements (eg. supplies electricity at night) solar badly loses its cost advantage. It gets even worse if the metric is the still important "deepest winter night" scenario.
Using mandated breaks for recharging heavy trucks isn't actually helpful in much of the world. Maybe it is in parts of Western Europe.
The problem is that those mandated breaks are mandated and happen (with a small amount of wiggle room) wherever the truck happens to be at that moment. Rolling out enough charging infrastructure to make that work is an even more immense challenge than the already massive challenge of adding sufficient charging infrastructure to places like existing truck stops.
Imagine the cost of installing 1MW chargers on, say, half the wide spots on every highway.
I think both Slate and Telos will be failures. They will be too expensive to make economic sense for people, as opposed to businesses, to buy over a more conventional full-size half ton pickup.
In some ways the massive online interest is proof, because most people outside of pickup truck forums who would talk it up have neither experience nor use for pickups. They are simply never going to buy any pickup truck-shaped vehicle and so are irrelevant to commercial success.
All the participants in those pilot programs *know* they are in a time-limited pilot programs and that in a handful of years the money will dry up. This is a major flaw in all UBI studies which make them all but useless.
It will take 15 or 20 years before any UBI could be considered permanent enough for a majority of people to change their work habits.
In most cases every residential system is getting payment rates at least that good.
The problem is that wH-for-wH doesn't take into account distribution costs, which is most of the residential cost of electricity, and that a wH at noon in July doesn't cost the same to generate as a wH at 2am in February.
For most jurisdictions you can look up the large industrial rates to find the wholesale energy rates. Residential solar is worth about half of that 'reliable' electricity rate.
Even beyond mutual exhaustion is housework. When both partners works outside the home, they still have to do the housework when they get home or on the weekend. Previously that would have been the job of the one staying at home.
The 20-ish hours a week needed for domestic chores has to come from somewhere.
Linux kernel drivers often end up being GPL'd, but out of tree. This is because Linux releases many very useful (and sometimes critical to the use-case!) functions behind a GPL-license API restriction. This is EXPORT_SYMBOL_GPL.
In my experience in corporate environments, that ability to forward to new participants with most of the context is really useful. If few people are going to read the history anyways, then in my opinion this edge case is valuable enough to tip in the scales.
I agree the difficulty quoting sucks, but that's mostly because of the switch from top posting to bottom posting. When people copy-and-paste the bit they are replying to and stay in the top posting paradigm things aren't so bad.
According to this article, a rather weak argument about people not liking to wake up before sunrise based on questionable correlation of commute times to sunrise times, ignoring factors such as average commute lengths, dominant (historical) industries, effective natural light at different times in modern housing.
From that it makes an (incorrect) assumption about the value of AM sunlight over PM sunlight and declares that all-year DST is pointless.
In my opinion the only argument against all-year DST which holds any water at all, and even then not much, is the concern about kids going to school in the dark. However, since many places don't have enough winter daylight to go around, trade-offs need to be made and kids are probably better off on-net having daylight time during their free time instead of while eating their toast inside and commuting to school.
I think the hypothesis was more true in the past, but mostly because both the necessary dependencies it is based on are less true today than in the past, and the products and services are more complex today.
Back when anybody could start building furniture the cost of entry was low and competition high. Switching costs were also low.
The cost of entry for a smartphone which is truly different are astronomical, many previously unregulated products are now strictly regulated, so costs of entry is no longer low and therefore competition is also low. For many services like software switching costs are very high. Firms need to be large to produce the complex products which introduces internal inefficiencies which are hard to avoid.
I think the author is missing the forest for the trees when they try to construct a taxonomy of 'elites'.
Firstly, there is little reason to believe that per-capita is the correct way to think about surplus elites. It's just as plausible that there is an absolute threshold above which the surplus meet together in, for example, the student union building and start causing problems.
Secondly,
> First, virtually all of the people who fail to attain their dream jobs can secure perfectly decent employment in some other line of work.
It's easy to say there's no surplus in any particular sub-group because they can always go to some other sub-group. But if most acceptable sub-groups are 'full', then it's not possible in general. Further, as a first approximation the 30% getting college degrees are the top 30% generally so it's not surprising that they could find _a_ job. That job being in some other line of work might be a problem, especially if it's several steps down from what they've believed they were owed all their life.
To use the article's example, a PhD in History may never have been likely to result in becoming a working historian, but it may have led to managing a factory or branch office. It's pretty obvious that is no longer true and it's likely that the wider job prospects of a History PhD holder has declined.
Further, the article foolishly equates becoming an actor or poet with academic achievement or business success. That entirely ignores that it's been well known forever that life success as an actor or poet is rare. The same cannot be said about getting a PhD.
Harberger taxes seem like a spherical cow solution.
It works in abstract economic theory, but encounters serious problems when you introduce pragmatic complications like cash-flow, malicious actors, non-rational actors, asymmetric value, transaction costs, lumpy probability distributions, and other things. Fixes to those various problems can be bolted on, but the end result loses much of the theoretical advantages which made Harberger taxes appealing in the first place.
In practice the spiteful billionaire will monthly send you offers $10K over the higher of your estimate or fair market value. So you'd be moving every month and incurring all the moving costs continually, or just continually dealing with the legal dealings of such an offer.
The transaction closes and the billionaire just sells the house at a small loss. The situation is asymmetric because the billionaire never actually uses the house.
At 100 people, every desk-job company is already a remote company, even if they don't know it yet.
Travel distance between desks has already become so large that many people won't do it for small things. For decades now those situations would be handled by a phone call or email.
Meeting in the coffee room to chat becomes rare because schedules and tastes (eg. office coffee versus off-site coffee, bagged lunches versus going out) differ. Also there's too many people and too much churn to really get to know anybody.
Arranging meeting times becomes difficult outside smaller 5-10 person units so asynchronous communication becomes predominant.
What I've seen work is not trying to co-locate a full team at all. Doing so only leads to silos and hiring difficulties. Instead have small offices which people from a small geographic area use. Those people will be on different teams and in different departments -- which is good for inter-team communication and synergy. This is exactly what offices normally miss because teams are co-located resulting in a relatively high 'distance' to build a rapport between teams.
Battery swapping has so many serious pragmatic problems I don't think we'll ever see it offered at scale for public use. It could be a fit for large private fleets however.
On the engineering side:
- Swapping requires standardization of batteries across models and manufacturers. To accommodate different vehicles the batteries will need to be rather small so most vehicles will need multiple swapped every time
- Requires more space and weight because the battery cannot be structural. This will reduce the overall range of EVs
- Connectors for high voltage, signalling, cooling fluid, and high strength mechanical rated for thousands of cycles in the face of road grime and poorly maintained swap robots will not be small. Cooling system contamination will be a serious concern.
On the financial side:
- Batteries are expensive, how do you track and reclaim them across the entire continent? What about theft? Destruction insurance?
- With swappable batteries the incentive is to store them at 100% then run them 100% to 0%, which is especially bad for battery longevity
- How do you deal with batteries swapped at different 'swap' networks?
On the user side:
- What if the swap station is out of batteries when you need them? Are you always gambling on holiday weekends that you won't need to sit for hours charging (if that is even possible!)?
- Since some batteries will be more worn than others, how do you deal with constant variability of range because maybe last week you got a new set of batteries and next week you'll get an older set with only 80% capacity left.
- Are you allowed to charge at home? How is the wear from that charged?
- Did I buy a battery with my car, or are cars no longer batteries included? If my car came with a battery, how do I know I get it back? Do I get paid for the wear other users put on it? Do I need to retrieve my battery from the same station on the way home after a road trip?
That's just off the top of my head. I'm sure there are others. Most of these issues are solvable with unlikely levels of corporate cooperation or immense levels of excess capital expenditure. However, they all cost money and will reduce the economic viability of battery-swap EVs versus every other vehicle type.
Your average, rather small, gasoline pump 'charges' an ICE at an average speed around 4000 KW, effectively 1200 KW after accounting for moderate efficiency -- hybrids will get better. Good EV charging today is a peak around 300 KW with a much lower average.
Honestly, _averaging_ 300 KW is probably within a factor of 2 of the highest we'll do for light vehicles given economic (how much electric distribution infrastructure can an 8-32 stall charging station have?) and practical (how heavy and stiff can the charging cable be?) limits.
It's unlikely EV charging speed will ever match existing ICEs. Relatively long recharge times are an intrinsic trade-off of BEV technology which needs to be engineered around, mostly by having enormous and heavy batteries.
My AIntern is available the moment I start my day (not normal for off-shore in a different time zone) _and_ is available after the end of my day when I want it to work after I've gone home (useful for work using scarce resources). It even keeps the context across that time span.