• 0 Posts
  • 42 Comments
Joined 2 years ago
cake
Cake day: June 21st, 2023

help-circle
  • The basic outline of where to split the company seems straightforward to me.

    AWS get split off first and foremost, that part is blatantly clear to me.
    From there, the retail webstore (what we generally think of as “Amazon”) gets split off from its broad category of services: music and movie streaming and everything in that category.
    After that, split anything that involves designing/repurposing other designs and selling a specific consumer product off. Kindle, Alexa, Roomba (if that purchase goes through), Amazon Basics, etc.

    I think there’s a decent amount of room to get more granular with the process, but I think that covers it as a basic outline.


  • BRICS isn’t an alliance or a cohesive entity. It’s the equivalent of the G7 for major non-western economies. India and China hate each other. China and Russia only really get along in being anti-US. Brazil and South Africa have no real intersection with the geopolitical goals of the other. BRICS isn’t a geopolitical anything of any meaning.

    I suspect India is doing this for the simple reason that they have zero control over Windows while they would have as much control as they want over internal-Linux use. They’re large enough that they can make it work, assuming they’re willing to dedicate the people and the money to it and put up with the non-insubstantial switching costs. Open question on what their follow through will look like, but it’s entirely within their capability.




  • This is speculation based on the combination of physical constraints and changing usage.

    Phone batteries today are in the 10-20 watt-hours range for capacity, or at least iphones are and that’s the data I found. Going from the typical ~20W fast charging rate to the full 240W capacity of USB-C EPR would allow a twelve times increase in battery capacity with no change to charge times. Are batteries going to increase in capacity by twelve times in the next 17 years? I’d be shocked if they did. The change from the iphone 1 to the iphone 14 pro max is 5.18Wh to 16.68Wh — a three times increase in 16 years.

    Likewise, with data transfer, it’s a matter of how human-device interaction has shifted with time. People increasingly prefer (a) automated, and (b) cloud based data storage, and (c) if they do have to move data from device 1 to device 2, they would rather do it wirelessly than with a physical connection. USB4 on USB-C is meant for 80 Gbit/s = 9.6 GB/s transfers. That’s already faster than high end SSD storage can sustain today, and USB4 is a four year old standard. People on phones are going to be far more likely to be worried about their wifi transfer speeds than their physical cable transfer speeds, especially in 2040.

    Then, on top of all of that… USB will continue to be updated. USB-C’s limitations in 2033 will not be USB-C’s limitations in 2023, just as USB-C’s limitations in 2023 are not the same as USB-C’s limitations at its inception in 2014. In 2014 USB’s best transfer rate was 10 Gbit/s, or 1/8 what it can do today.


  • I’d be surprised if USB-C was a limitation on phone technology even by 2040. The bandwidth and power delivery capacity are way beyond what are needed now. Data transfers from phones are going to increasingly move to wireless in that time frame too, I expect.

    The limitation on the viability of USB-C with phones won’t be the actual technological viability of the standard with respect to phones. Instead, the problem for USB-C for phones will be if another standard comes out and starts being used by other devices that do need higher bandwidth or power delivery capability. Monitors, storage devices, laptops (etc.) will eventually need more than USB-C can provide, even with future updates to its capacity. When those switch over to something new, that will be when phones (and other devices) will need to consider a new standard too.


  • Bitcoin can fuck off.

    The point here is that car companies already charge for these things. The reality is basically two scenarios when ordering a car:

    A: You pay $x, and they offer you heating steering wheels for $y. If you do not get them then, you do not get them ever.
    B: You pay $x, and you can pay $y at any time to get heated steering wheels.

    The business “bet” that (B) represents is that maintaining additional SKUs for each upgrade-feature and splitting off production lines to include or not include various combinations of features 1-2-3-etc. will cost them more money than just including it in every car. Then they can sell it to you on a whim. The actual feature itself does not cost anywhere near $y in either scenario to include, which is an important component of making this possible.

    Now, you can say that (B) is a shitty scenario in a vacuum: if they’re willing to include it in every car, they should just charge every car what it costs to include plus some minor markup to allow the business to operate. E.g. if it costs $50 to include, they can increase the price of every car $55. And in that vacuum I’d agree. But it isn’t in a vacuum. That is not the scenario (B) is competing with. (B) is competing with (A). In (A) you are going to pay $200 or $300 or whatever for that $50-cost feature up front, or you never get it ever. In (B) you pay that $200 or $300 whenever you like.

    It operates in a similar world to how Apple charges $200 to go from 8gb of RAM to 16gb of RAM, when that might cost them $10-20 at volume pricing. Or to use a well-liked company, how Valve charges $250 for a ~$10 SSD + ~$5-10 carrying case + ~$5-10 glass coating, on the base Steam Deck vs the fanciest Steam Deck.

    This is not a “as a service” model. It’s a simple upselling business model. Profits on base models are low so as to have a low sticker price, and then they try to create profit off of upgrades. In this case, the software locked version is preferable to the consumer over the default version because it’s something you can unlock at any time, instead of only at purchase. It is not a new business model, nor is it even limited to electronics. The overall business model is shitty, but that applies to every instance of it: (A) and (B), and (B) is not differently shitty.

    Service based systems are based on recurring revenue, in this case anything with a subscription. Which I specifically called out as something that would make it shitty and pointed to their subscription based or subscription-incentivizing behavior as shitty.






  • You’re also making the implicit and incorrect assumption that this assumption of future income is not already the status quo. It is. The IRS already does this with your automatic withholding. It just does it at a higher level than necessary, due to what I mentioned above. Withholding basically assumes that a person will continue to earn whatever their paycheck was in every future payment period (weekly, semimonthly, whatever).

    Your assumptions on how this would be dealt with are not realistic. The outlines of how to implement this not only already exist, they are already used and have been used for decades. All the IRS needs to do is glue together its knowledge of your income sources and lower the withholding amount based on being able to predict an individual’s income far more accurately.

    Data on seasonal income is known too, for the record. Consistent trends in income changes with parts of the year are not a surprise to the government agencies that care about it.


  • For the majority of people out there, all their income is going to have digital records. A cash only store still deposits money in a bank, after all. For the people that don’t… chances are their income without a digital footprint isn’t being reported, and is small enough that no one is worried about that in the first place.

    If the IRS is being told by a person’s work how much they’re paid, by their bank how much interest they got, by any Etsy-esque services how much they were paid… then the IRS has every bit of information it needs to get automatic withholding correct.

    Right now withholding systems default to taking too much money out, because it’s easier for the government to send you money than it is for them to request money from you. It also avoids the headache of most people being hit with surprise IRS bills. The IRS could keep that default, and then as the year goes on it could shift that withholding down until it’s laser close. The negative there is that variability is bad for budgeting too, but with some work they could make it start close enough that it shouldn’t be all that variable.


  • The report gives a quick summary of what they include, but not any details or math.

    The cost of underlying energy (gas, diesel, electric)
    State excise taxes charged for road maintenance
    The cost to operate a pump or charger
    The cost to drive to a fueling station (deadhead miles)

    Elsewhere it says it assumes 12k miles in a year and is focused on the midwest and Michigan in particular. As it so happens, Michigan charges for registration based on the car value. EVs cost more than ICE vehicles in the same market segment most of the time. This would fall under excise taxes that they include.

    I wouldn’t be surprised if they also tacked on the cost to install a L2 charger once as “cost to operate a pump or charger” — intentionally ignoring that it’s a one-time fee to support EVs at a home. With those two data points they could easily add >$1000 to the cost to “charge” an EV for one year if that is what they wanted to do.

    The people making the report clearly picked criteria that sounds reasonable but also intentionally misleads people. Not a surprise.






  • I can’t read the article but I think they’re making a bit of a mountain out of a molehill.

    BEVs were nigh impossible to purchase a year ago. Tesla’s MSRPs were ~$10k higher than they are today, not even accounting for the tax credit. Other manufacturers were seeing dealer markups of $10k+ on a new BEV. Demand for BEVs went through the roof as (1) supply chain effects meant the price difference between ICE and BEV went down, and (2) Russia’s invasion of Ukraine sent gas prices way up. A 350% jump over last year doesn’t mean much in that light — what inventory even existed on dealer lots last year?

    Both of those factors have faded. EVs are still selling well, but manufacturers are going to need to find more ways to lower prices in order to stay competitive and to keep demand up.


  • (1) I didn’t downvote you.
    (2) I said something similar but critically different:

    Building a streaming platform that expects to have multiple billions of dollars in revenue across hundreds of millions of users is going to have enormous fixed costs that cannot be trivially scaled down if user counts are lower. If they plan around a much lower user count they can scale it down at that planning phase, but not after the fact (at least not easily).

    The intended size of the platform dictates the fixed costs.

    And…
    (3) The data you provided wasn’t fixed costs. It was variable costs like server time, music rights, and bandwidth.