Investors, seriously? What about customers? They burnt thousands of millions of customers money, and the first ones they think about are investors?

"Scientists have struggled to show that fusion can release more energy out than is put in since the 1950s"

Actually, this was clearly shown in 1952. Maybe you meant "in a sustained and controlled way"?

Quantum computing does not become the next blockchain, and provides actual value.

Decreasing the carbon footprint in the US?

You perform due diligence before submitting a firm offer for the company!! This would have saved so many problems!

(No, I will not pay the $8 XD )

You mention a very interesting point! fairness guarantees in a shared medium vs fairness in a switched network. I agree that they differ, and that shared-medium systems are much more prone to abuse.

In a shared-medium system with distributed medium access control, such as a Wi-Fi network (using the CSMA/CA MAC mechanism), users compete to send data, and the system may be abused by a single user sending too much data. In fact, under high pressure the performance degrades for all the users.

However, in systems controlled by an ISP, the medium access control mechanism is typically arbitrated by the ISP equipment, instead of using contention-based mechanisms. We can consider download and upload sepparately. The download channel is always controlled by the ISP, since the data comes from its own device (the OLT in GPON, the In the case of GPON, the CMTS in DOCSIS, the satellite and the rest of the network in Starlink) so they can always modulate how much data is received by each user using queueing policies. I agree that this is not perfect, since users are typically considered by QoS classes (and not individually), but there are some QoS mechanisms in place. In the upload channel, to avoid the abuse and guarantee fairness and/or QoS, the ISP equipment typically controls the allocation of resources for each user. For example, in GPON the OLT assigns bursts of data (time slots within the upstream frame) to each ONU client, based on the occupancy of their queues. In DOCSIS, the CMTS (the headend device from the ISP) periodically sends MAP messages, with minislot allocations based on previous requests by the cable modems. There are no public details of the MAC mechanism employed in Starlink (that I am aware of), but it is very likely that the satellite controls the allocation of resources per user both in the downlink (obviously) and the uplink, preventing abuse by a single user.

The problem in these systems with centralized arbitration is not so much fairness, which can be somewhat controlled by the arbiter from the ISP at the MAC layer. By contrast, the problem is typically capacity, since the capacity of the (typically, wireless) channel is often very limited and the channel highly oversubscribed. This is why data caps unincentive the use of the network, reducing the offered traffic load and improving observed performance by all users. By contrast, capacity issues in other systems (such as GPON) typically occur at the backbone, which can be much more easily upgraded.

You are making two mistakes with your argument.

First, you are presenting some examples in which your argument does not apply. Consider GPON (G.984) as an example: Nominal download speed is 2.48 Gbps, shared by up to 64 users, and ISPs often sell up to 1 Gbps per user. Thus, each user may be allocated, in theory, a sustained 2.48x1/64=3.87%, or 38.75 Mbps, regarless of their maximum "paid-for" speed. However, the typical situation is very far from these numbers of the worst case: The number of users per fiber is typically lower than 64; not all of them are downloading simultaneously; and even in a typical use case (web, HD streaming) 38 Mbps is more than enough.

Additionally, in GPON there are many fibers per OLT (the headend device at the provider's premises), so they can serve more users. Same applies to DOCSIS. The problem for starlink is that there is only one satellite (not one satellite with many fibers) and the number of users in an area can be huge, way larger than the 64 limit in GPON.

Your second mistake is that you exclusively focus on the last mile, whereas any network is always shared among many users. Even if the last mile is private (e.g. a dedicated Ethernet wire), after the headend router from the ISP the backbone is shared by many, many users. Depending on the oversubscription factor, the congestion point may occur in the last mile or the backbone after the headend. In Starlink the bottleneck will clearly be in the last-mile (or maybe "last 250-miles"?), but this does not occur in all cases (in GPON it will be probably in the backbone).

The area covered per satellite is huge and the bandwidth is shared by all users. Thus, the system cannot provide a good service (i.e. high speed, as in the early network tests) when the number of users grows. Forget about video calls or Netflix.

It seems someone is confusing sampling with quantization?

"You pay to get half as many ads" does not work. If I pay, I want to have an ad-free service, thank you.

Arm is killing their own business model. If designers (such as Qualcomm, with an architectural license) find that combining their own cores with other IP is very complicated, they might start considering seriously the migration to a different ISA. And when they move to RISC-V, they will not come back to ARM.

...seek relief through GOVERNMENT bankruptcy protection? Seriously?

Before you start complaining, bear in mind that Rishi Sunak is a longtime smartphone user, and has proficiency using Outlook and chrome in his laptop. Possibly also PowerPoint and Word.

News for nerds.

That's why the onion network and the Tor browser never took off. Because they lacked bitcoin.

Seriously, I'm surprised that the "little" detail of having small children at home is never discussed when talking about work from home. In my case, it's completely impossible to work from home out of school hours.