chabons

AI models on FPGAs has been tried before, for instance: https://www.microsoft.com/en-us/research/project/project-cat....

They haven't been able to compete with GPU's on perf/watt. In general you end up just designing some AI accelerator for the FPGA (because the models are too big to map onto a single device all at once), but it's hard to beat purpose-built tensor and vector HW on a GPU when you're running soft logic.

Right, but the argument that the original comment was making is that the government's policy is incorrect.

My comment argues that independent of government policy, relocating an existing employee from one country to another does not consume a position that was otherwise free, unless the employee was hired expressly for the purpose of relocation.

> Obviously you displaced a US citizen [...]

I'm not sure this is obvious. 1) Apple is a multi-national company, and hires an employee in the UK. 2) This employee relocates to the US

At what point do they displace the US citizen? All Apple jobs are not earmarked for US citizens (+residents, etc...), and they're already doing the job when they move to the US. Unless they're hired abroad for the express purpose of relocating to the US, then it's not "obvious" to me that they've displaced anyone.

Suppose I work in an org with dev teams in multiple countries. If I relocate from one site to another, am I freeing up a slot in the country I leave and taking one in the country I move to?

I think the parent comment is suggesting that "engineering talent on par with the USA" == "elite engineering talent". In my anecdotal evidence, I'd estimate that ~1/3 of my Canadian engineering graduating class are now in the US, either working or researching at US institutions, with most of these folks representing the top 1/3 of the class.

> [...] actively discourage or outlaw normal car designs in favor of cars like the one in the link.

I read their comment as encouraging city-specific measures. They specifically reference "urban living", "human-scale living", or slowing down vehicles for "in-town trips". I agree that use-cases exist for larger vehicles, as even in the densest city centers delivery vehicles continue to need access. IMO larger vehicles need not be the vehicles we optimize for, or even allow at all times, in city centers, where such vehicles aren't needed by most residents.

> [...] 100% to encourage the kind of alternatives you listed to decrease overall single family car use, not try to force everyone to change car designs [...]

Honestly I'm thinking of this more like a large cargo bike, and less like a car. It's like a Canta (https://en.wikipedia.org/wiki/Canta_(vehicle)) in the Netherlands. It's not meant to serve the same purpose as today's cars, and we probably both agree it would feel unsafe to drive one amongst fast-moving traffic. It fills a gap in the transportation spectrum that's well-suited to certain areas, and I believe we should devote more infrastructure to vehicles which match the needs of an area, which I don't believe is always full-sized passenger vehicles.

> You buy vehicles for the life and city you have [...]

Agreed, there's a place for all forms of transit. I think the argument that the parent commenter was making was that we already have the density in many places for vehicles of this type make sense for residents, but the barrier is that road infrastructure is built to prioritize larger vehicles, making these vehicles unsafe.

In places where safe infrastructure exists, families frequently plan their lives differently, and I know many (with the means to choose freely) who get by with carshare, good transit, cargo bikes, etc...

> Not sure OpenAI faces the existential risk as MS need them to compete with Google in this space.

I think OP is arguing that in that partnership Microsoft holds the power, as they have the existing platforms. The linked article argues that AI technology itself is not as much of a moat as previously thought, and the argument therefore is that Microsoft likely doesn't need OpenAI in the long term.

I'm wondering if they controlled for cat ownership (as a likely source of infection), or if any of the variation can be attributed to the types of people who own cats.

You raise a good point, that both moonlighting and non-compete clauses provides employers the ability to fire employees to mitigate risks associated with simultaneous work for a competitor (theft of trade secrets, corporate espionage, etc...). I'm not sure what the correct answer here is, but I'd err towards the burden of proof being more on the employer to show that the employee contributed to a direct competitor using insider information than the blanket clauses we have now.

> Expecting the attention of someone 40 hours a week seems reasonable

Why should the company be able to restrict your ability to do other work beyond the ability to fire you for cause if it's interfering with your ability to do your job?

We're thinking along the same lines, I guess I just expected the upsides and downsides of the approach to be quantified somewhere in the research as context for why it's meaningful.

Interesting research. Assuming that the benefit of fixed-wing aircraft is efficiency in relatively straight flight, I'm wondering how this aircraft compares in efficiency vs agility against a quadcopter of similar weight.

> Montlake had a small grocery store that was torn down precisely because of the 520 construction.

Right, I forgot about the Montlake Market. That closed before my search, but I did hear about it.

> 520 is a highway after all, and highways are hostile to walkability for a multitude or reasons

Yeah, that's fair. I'd argue that 520 in it's final incarnation will be a little bit less hostile than most, as it's mostly over water, and the lids over the 2 land crossings (with parks + transit) will partially mitigate the walkability concerns. It'll still be a noisy eyesore though.

Geographically, Montlake is in a pretty desirable location. It will have quick transit connections to downtown Seattle+SLU, UW (and light rail), and the eastside. I imagine the future only blocker to development will be zoning and pushback from existing residents.

Interesting how the visualization highlights the Eastlake and Montlake neighborhoods (along the south bank of the canal northeast of downtown) as walkability outliers. When I was newer to the Seattle area and looking for housing I did a similar exercise and charted the location of grocery stores. I was surprised that those neighborhoods just didn't have any grocery stores, especially given their proximity to downtown and the UW. Given the amount of investment the city/state is making to improve the WA-520 corridor (the east-west highway which goes through there) for cars and pedestrians, I wonder if that's going to change in the near future.

This is a fair point. This press release was not aimed at software engineers, and "rules that prohibit those kind of changes" is vague. It could mean anything from a sign on the wall that says "please don't!" to a robust access control and deployment pipeline that they intentionally circumvented.

I'm sure they've determined that communicating that sets the user's expectations. A funnier take is that it's advanced watermarking. Good luck getting that out of the pre-trained model!

I think the severity of the event is also important. A Onewheel requires power for stability (it's an inverse pendulum), and as a result does not fail-safe. An occasional motor failure on an e-bike would be a non-event, as it fails to a safe state. An occasional brake failure would be significant.

This video is directly produced by Onewheel's marketing department. It does certainly highlight one way user error can lead to injury on a Onewheel, but I don't think this is a credible refutation of the idea that there may be a flaw in the product.

I suspect it's not just the occurrence rate which influences this feature being male-focussed. It's probably also the expected reception, which is itself a function of the occurrence rate.

If I opened an app one day and received an unsolicited picture of female genitalia, that would be strange. If it happened every day, then it's spam, even if I don't have a strong aversion to it.

Maybe I'm mis-reading the paper, but my interpretation was that the algorithm discovered using a reward which scales with HW performance matches the theoretical complexity of Strassen, but is more performant on the HW.

They also identified algorithms which do fewer matrix multiplications than Strassen, improving the lower bound of matrix multiplies required (They highlight this in Fig 3).

In that light, I thought their claim was fair. They've discovered (different) algorithms which are both theoretically and practically better than Strassen's.

Honest question from someone with little context: How does this argument not also apply to Citizen's United, or corporate personhood more generally? Is the argument that compelling a corporation to do something compels the individuals within the organization (infringing on their rights), while not giving an organization a right does not infringe upon the individuals rights (which they can exercise individually instead)?

That percentage will depend heavily on the show. The Big Bang Theory is fairly innocuous. Imagine trying to censor dramas like Game of Thrones, Breaking Bad, or House of Cards to remove all of the depictions of sex, drugs, or political commentary.

I don't have a ton of familiarity with the project or the region, but the section of rail in the linked article certainly looks like it makes sense to me. A high-speed connection including the airport and downtowns of the two major metros seems like a decent project.

Taking a random stab: one argument could be that competition drives down profit margins. By reducing competition, China makes their tech artificially profitable locally, allowing them to compete in international markets on an uneven playing field.

OP is also raising the point that because only the police have the ability to retrieve Nest footage that pertains to them but is not owned by them, access to that data is gated by the police, who may not use that access evenly.

> As set forth in the Dodd-Frank Act, the SEC protects the confidentiality of whistleblowers and does not disclose any information that could reveal a whistleblower’s identity.

This would seem to preclude any way of independently verifying their claims.

Realistically, to prove that there existed a whistleblower and this is not just marketing by the SEC, you would need incredible amounts of evidence, and even then you'd have no way of knowing whether the evidence is fabricated. At some level I think you end up just having to trust that the SEC isn't lying about paying whistleblowers.

That's a good point, I hadn't thought of having independent send a receive delays on the server, simulating a delayed connection from the client's perspective without having it participate.

I now wonder too what those technical considerations are, since having the client add latency requires trusting the client to add the delay and not cheat.

Right, but my point was that I think you're reading into the diagram beyond it's intended purpose. I'm not sure how you can tell that they're logging only the server-side delay, and not the client-side delay, from a systems diagram which shows each of those as a box connected by a bi-directional arrow. For instance, it's clear to me that they don't include the latency included by the game engine in that metric, but not whether they include the server-added latency or if they're aware of it and subtract it off.

I think the symmetric delay here makes sense, since you're trying to simulate the latency of a (likely) symmetric connection between another client and the server.

If you only add latency on the transmission, then 2 actions taken at the same time by players with different amounts of physical latency will be processed at different times. If we assume the game clocks on each client and the server are in sync, then this creates a competitive advantage for the player with lower physical latency. For instance, if we have a timing battle (ie: Zhonya's hourglass), where both players know that at time 'T' they must each take an action, and the first to do so comes out on top, then the player with higher physical latency is at a disadvantage.

I read that section as comparing what they logged before vs what the player experiences: the time taken for an input to be registered and it's effect communicated to the user. Maybe I'm wrong, but I don't think the intention was a precise description of the error, but rather the hazard of not having logs which reflect what the user experiences.