typewithrhythm

How often are you actually doing this though? I think I probably work in something greenfield about once a decade. The hard part is always going down a rabbit hole in established code bases. I can do the boilerplate in a few days. It saves time, but not really even one hairy issue a year.

I saw a Matic, seemed impressive, but I have very few points to compare with personally.

Why does anyone innovate? When I sell something I don't care if the dollars come from the end user or the subsidy, I just want more of them.

Every category I can think of where China is near-first there is some international manufacturer that has a better product.

Several areas where there are much higher volumes or outstandingly better value though. Things like automotive lidar, construction assemblys (like double glazed window units), consumer electronics like quadcopters.

Mostly it costs hundreds of millions to develop a chip; it relies on volume to recover the cost.

NVIDIA also tailor their chips to customers. It's a more scalable platform than their marketing hints at... Not to mention that they also iterate fairly quickly.

So far anyway, being on a specialised architecture is a disadvantage; it's much easier to use the advances that come from research and competitors. Unless you really think that you are ahead of the completion, and can sell some fairly inflexible solution for a while.

The big lidars are for ground truth collection. They get used in projects ranging from autonomous development all the way down to budget adaptive cruise control or parking sensor benchmarking.

NVIDIA orin series is the big one for tensor horsepower. Horizon robotics and Qualcomm also have competitive automotive packages.

They are all expensive, but less than the risk adjusted cost of developing a chip.

Tesla fsd is far from complete enough to be a data point; people who pay the 10k are gambling that when fsd is improved the cost will be much higher.

I don't think the throughput of a general purpose device will make a competitive offering; so being local is a joke. All the fun stuff is running on servers at the moment.

From there, AI integration is enough of a different paradigm that the existing apple ecosystem is not a meaningful advantage.

Best case Apple is among the fast copies of whoever is actually innovative, but I don't see anything interesting coming from apple or apple devs anytime soon.

It depends; it feels like in some categories the premium between a material that's very suitable, and some ersatz lookalike is massive and depressing.

I love a good petrochemical, but sometimes it would be nice if the cheap thing store wasn't so callously targeting veneers and pleathers that last just long enough to loose the receipt.

Labour centric protectionism.

I've at least found that the wifi+Bluetooth chips seem to be significantly more robust than the standalone bt ones.

Huh, I had no idea that cables would have their shield grounded at both ends... Single point ground is such a standard in electrical design that the guidance is generally "do otherwise only if you have the ability to make many prototypes to nail RFI issues".

It's a case of high trust, high skill structures not being maintained while trying to introduce outsourcing and minimum viable skill lower cost employees.

The idea of owning your own quality only works if you can trust the dev to understand quality, and want to implement it. Independent almost adversarial QA is required when you can't trust the devs.

Before the AI stuff it seemed people assumed everything would get smaller and more portable continuously.

Not consumer computing, but desktop computers would disappear.

Most of the small high speed drones are that size to fit under professional licencing requirements, often so that one racing spec can be viable across a wider area. Leading to significant competition in that size pushing down prices.

Rather than some inherent sized for safety idea.

Jamming might be interesting, I suspect that it's easy enough (and a much bigger crime) to follow a very loud jamming signal though.

Every practical metric a drone surpasses a helicopter; they are so much simpler to operate that you can easily offset any perceived downside with more drones. And you don't get a tested solution without trying it out.

Ai interfaces are going the same way the public internet has; initially it's audience was a subset of educated westerners, now it's the general public.

"Most people" have trash taste.

You cannot make a fair system by introducing subjective ideas like historical balance.

A set of rules for fairness require that current decisions only account for individual merit; not special status.

It's the anonymity and odd changes in who is moderating that makes it feel different. Standard setup to me would be consistently opinionated person, or team with some central directive (and hopefully oversight).

It's the idea that if they leave more players idling in a lobby, but period, or animation, that it costs them less.

It's a deceptive way to sell people less game.

I have spent a fair amount of time in very large companies (single projects involving thousands of devs). You end up producing a whole lot of management training (and managers) in this environment, just due to its size.

You end up with not exactly an intentional bureaucracy, but one where the idea of fairness from somehow "objective" numbers becomes a focus.

This kind of works at this scale, because you need to have a way to abstract and reason about the capabilitys of far too many people than you can know individually.

The training and materials don't scale down though, so you get someone trying to apply metric driven performance in cases where it just doesn't fit.

It's generally "ok" for big business, because projects at this scale can survive on rigid organisation, simply because achieving anything at that scale is a challenge enough to be valuable.

Occasionally you see inspired leadership, but every level of management it has to go through erodes it. It's part of why it's so rare for a big company to produce anything unusualy good at scale, it takes a real alignment of stars.

This way of working really requires a small company, (and is one of the reasons I think small places have a chance of outsized impact). But at bigger places, either a manager is being judged on a team outcome -> where an individual not on topic is budget not going towards what they get paid to achieve. Or you need seniority enough to work directly with a project director (but most structures expect people at that level to have leadership responsibilities as well).

Unfortunately, while people that work like this can be exceptional, big projects run on organised measurable work. I have found few places big enough for a specific "Manager" role to be flexible enough to get a good match to tasks.

You hire discount engineers when you want to move to a maintenance and value extraction stage, something widespread in the industry at the moment due to changes is startup incentives and interest rates making entrenched players feel more safe.

It's not about "global equilibrium", because that misses completely that hiring a very expensive westener still has positive roi, but only if you are attempting to innovate.

How does the cost breakdown of all these new datacenters, cooling, and power delivery systems compare to the cost of the GPUs themselves?

There is a surprising amount of real long-term infrastructure being built beyond the quickly obsolete chips.

The economic benefits are really not clear; at least not without caveats and clear conditions for the advanced skills that make a migrant beneficial.

This is if you believe that lower wages for high skill work is not an issue.

However high migration rates lower social trust, this is well studied.

If you take a smaller example, hiring internationally vs domestically. If you have to go domestic then you might have to settle for a less ideal qualification, requiring more training.

This is repeated everywhere, so companies that train better are more likely to succeed. Leading to conditions that encourage upskilling for locals overall.

Importing people short circuits that idea.

It's so hard to study; one of the key things you loose in an environment where you bring in bulk migrants is a cultural expectation to interact with juniors that are part of your community.

It's not just a supply and demand equation; it's a fundamentally different environment that changes the social payoff for mentoring, networking, and building a reputation.

Ultimately despite all the propaganda trying to convince us that diversity is inherently beneficial, we are trading economic benefits for social costs. So we need to carefully restrict migration to make sure the economic benefits are actually there.

It's really not that simple; the environment that allowed for the kind of appealing interactions blogs offered is much harder to come by.

Early days had a barrier to entry that made the public who would read and engage much more enjoyable to write for.

The universal internet is not the same thing as the one that naturally filtered it's participants.

Because the idea of productivity per employee its relevant to look at capital spending as well.

Otherwise people claim Microsoft is successful and hires well (with outsourcing or RTO or whatever), but in reality they are in a shift to a more capital intensive business.

How much of that is product engineering and how much is leasing access to capital during a GPU demand spike?

More likely they have decided popularity is irrelevant since they are so entrenched, so may as well try to find the cheapest possible maintenance engineer to manage the value extraction.