otterley

Given that the number one cost of inference is memory and compute, and the incremental cost of each is energy, cost per inference is roughly proportional to energy consumption.

I’m aware of the advantages. I’m choosing to let you exercise your innate curiosity to learn them for yourself.

It’s more nuanced than that. Not all ideas neatly fit into “good ideas” or “bad ideas.” Sometimes they’re a mix of both, or someone has a good idea, poorly executed.

> despite rising prices not proving rising demand

Central banks don’t take the cause of inflation into account when they adjust borrowing rates. This is a feature, not a bug.

Agree in principle, but some people mistakenly believe that because a wealthy person did something and succeeded, regardless of the cost or externalities, it must have been a good idea.

Yeah, well, fuck those guys. Don’t try to emulate them. Fortunately, we have better wealthy role models like Melinda French Gates, Mackenzie Scott, and Warren Buffett to look up to.

> Engage with reality as much as possible. That means less time on this site and more time in the world. Less time watching human faces on a screen and more time seeing human faces in person.

> The more time you spend with real people out in the world, the less you will emotionally rely upon abstract intellectual models to infuse your life with a sense of connection and purpose.

> Hold all of your opinions lightly. Be proud of your ability to change your mind. Actively seek out evidence of your own ignorance and failings.

Amen, brother.

I’ve done the best I can with limited time and space to explain it here. If you still don’t understand or don’t care to learn how cross-subsidization works and the costs and business model of operating a tier 1 cloud provider and how they differ massively from operating an ISP, I don’t know what to tell you. Getting a job at one of these companies to gain that experience would be eye-opening for you.

> The fee agreement was an hourly rate for producing a legal document.

We don’t know that for certain.

Your software can be great and your website can be mid at the same time. Perhaps the latter isn’t the highest priority for you, but criticism of it isn’t invalid.

Oh yeah? How do you know? Are you handling finances for a tier 1 cloud provider’s networking team?

Show us the math.

Why would it be fraudulent if the fee agreement disclosed it?

This reminds me of the parable about the plumber.

A plumber comes over to repair a broken water heater. He looks at it, taps a pipe with his wrench, and it's fixed. He hands the customer a bill for $500. The customer becomes angry because all the plumber did was tap a pipe, and demands to see an itemized bill.

The plumber dutifully provides the bill:

* Pipe tapping: $5

* Knowing where to tap: $495

> Vendoring isn't a clear win for security. It may provide some protection against "supply chain" attacks, but makes it more difficult for you to respond to vulnerabilities discovered in the version you vendored.

How so? Whether a dependency is vendored or not, you still have to update it to integrate a security update to that dependency, don't you? The alternative is to not pin your dependencies, but that is far riskier overall.

> Whether or not the benefits are worth the inconvenience is a different question

I would contend that it is the most important question. :-)

> If it's surplus it's not costs. Why did you claim that the high price is to cover the costs if 99.9% of it isn't covering the costs?

I edited my comment above to explain in more detail what these charges are covering (and it’s not egress bandwidth alone). This has been a fast moving conversation and I’ve tried to add more context, unfortunately after the fact. My bad for posting too quickly.

I think it’s “am I getting the value I expect for the price I pay?” If it’s not for you, that’s fine. But for others, it’s worth the cost. To them, they’re paying for more than mere egress bandwidth.

The bandwidth itself? I’m honestly unsure as I don’t know how AWS’s interconnectivity and total bandwidth compares to Arelion. But it’s still the wrong question to ask.

I updated my reply—please see the above links.

Again, these egress fees are not paying for egress alone. They are defraying the cost of operating a massive, complicated, low-latency, reliable cloud network infrastructure, much of which is free of charge when used internally.

Some, like you, are expecting that cloud providers operate on a “cost plus” model where what you pay is strictly based on what marginal costs for the same thing. But it’s not that simple in reality. Charges levied for one thing can be used to pay for another.

Like I said: you are welcome to disagree with others. You are not welcome to insult them.

Many of us are aware that ISPs charge less for bandwidth. But they are not offering the same thing as a tier 1 cloud provider.

You may find these interesting:

https://aws.amazon.com/video/watch/c37546e1558/

https://cloud.google.com/blog/products/networking/speed-scal...

You’re confusing an ISP with a tier 1 cloud provider. They are different animals altogether.

And kindly refrain from accusing others of Stockholm Syndrome here. It’s incredibly rude.

I don’t think we can say with certainty without knowing more details about their respective network investments. CloudFlare is architected quite differently than the others, and has a much less comprehensive service portfolio. It’s not an apples to apples comparison unless we’re strictly comparing S3 to R2. Choosing to charge less may also be a loss leader or a differentiation play.

CloudFlare is not a tier 1 cloud provider.

The price of security and business continuity is the occasional inconvenience. Tale as old as time.

Egress pricing pays for the AWS network infrastructure that is incredibly reliable (hardware failures happen regularly yet almost no one notices) and allows it to operate at scale sufficient to absorb even the largest DDoS attacks.

This isn’t just AWS BTW; all tier 1 cloud providers recoup their costs this way.

> If I want to upgrade the disk on my MacBook Pro, I need to buy a new MacBook Pro with a larger disk

Is this a significant risk in reality? MBPs today come with a minimum of 1TB of storage. Even 5 years ago I think the minimum was 256 GB. This is more than large enough for all but the most massive repositories, even with vendored dependencies. And you can always plug in external SSDs or HDDs or connect to a network server.

> And what’s the advantageous scenario for vendored dependencies? Is it just when the mod proxy and the upstream code host go down at the same time?

This is a useful homework assignment. Ask your favorite LLM or consult some respected release engineering books. Also consult your local AppSec and infosec teams.

These are all solvable problems IME. But it does mean you need people who are experienced at solving them or who care enough to learn.

It’s not really a question of whether they’re using it or not (most of them are); it’s a question of whether they’re using it to generate work product intended to be consumed by their customers. Most know better than that. Some might use it for that, too; but you know what they say about half of the population being below average.

It probably is, and that’s not great, either.

Disk is cheap.

Recursive dependencies have the same issues whether you vend them or not.

Ensuring that diffs to updated dependencies remain within a vendor folder is trivial.

So, what’s left?

Because taste, creativity, and differentiation matter to a human audience. Perhaps not you personally, but to most people. Ask any experienced marketer.