qqtt

I'm going to go against the grain here and say this is probably a positive thing for Meta products, and honestly every other "free" service to provide these kinds of revenue avenues.

How many times do we hear things like "if the product is free, you are the product" - well, the consequence of that is development resources tend to be pulled into directions that benefit advertisers.

By having material subscription revenue coming in for things outside the advertising space, the product managers can justify investing in features that otherwise would be passed up due to lack of revenue potential from advertising.

Yes, in many ways Meta gets to have their cake and eat it too, because the ads are still there even with the plans, but this does give a meaningful voice to their customers who pay that they can invest in other ways outside of strictly advertising.

I would argue the problem is multi-faceted, and screens are a convenient boogeyman which is a relatively easy thing to point to.

The harder problems are that both parents need jobs to make ends meet, meaning actual time with their children are both lower quality and less impactful due to lower energy and less time. Children are given devices to play with because the parents are exhausted and don't have the energy to fully engage with young children that are full of energy.

Education itself is also chronically underfunded, especially teacher salaries. Whereas before teacher salaries would pay something resembling a living wage, these days the cost of living has exploded and teachers are generally just simply left behind as an afterthought in public budgets.

So you have cohort after cohort of children with less quality education time with their parents being funneled into underpaid teachers who are expected to teach a class of 20-40 kids how to read, with poor support systems in place for everything from kids with behavior issues to even potty training in grade school.

As a society, we aren't valuing education - neither from the home side, to the workplace accommodation side, to the actual classroom. Until we all collectively agree that this is something worth investing in and we need to spend the time, money and energy to do it correctly, it won't get better.

Screens are a symptom but taking them away completely is just treating the symptoms instead of the underlying disease.

Also, as a reminder, leaded gas (avgas) is still used all over the United States pumping lead into the environment. If you live near an airport you are especially at increased risk of lead exposure in the environment.

I don’t think that is specific to Amazon. If anything I think the whole paradigm of “over analyzing communications with vacuous update suggestions that don’t matter” is probably the most consistent thread I’ve found in all of corporate America. I’ll never forget early in my career we had huddled around a coworkers computer with our entire team including senior manager writing an email to a VP and the senior manager was making punctuation and the most inane wording updates you can imagine. I once thought there is no way the absurdity of the cost to benefit of that situation would be topped, but how naive I was - turns out generally companies seemingly can’t avoid that kind of atmosphere.

As someone who was super interested in the 538-style of election coverage in 2008, I've kind of fallen "out of love" so to speak with election models and forecasting in general. I'm not really convinced about what it adds to the conversation around elections. We can all look at various polls and get an assessment of who is generally ahead. Weighted polling aggregators and forecasting models just collect all these polls and spit out some data. It's easy to hand wave and think some new information is being revealed, but ultimately it is just a "garbage in garbage out" situation - you are entering polls as input, some hand waving is going on, and you get some forecast as a result.

I think part of my cynicism comes in the wake of the 2016 election, in which the forecast rightfully counted some scenarios in which either candidate could win, upon which conclusion of the model was basically "the result fits in with the forecast, because either candidate could have won according to the model" - in which case I personally concluded, if no matter what the result, we can always just say "the candidate who won could always have won given the forecast" - what are we really adding to the conversation here? We can simply look at polls and understand who is generally ahead, and not be any better or worse off.

I wonder if there is a big disconnect partially due to the fact that people are talking about different models. The top tier coding models (sonnet, o1, deepseek) are all pretty good, but it requires paid subscriptions to make use of them or 400GB of local memory to run deepseek.

All the other distilled models and qwen coder and similar are a large step below the above models in terms of most benchmarks. If someone is running a small 20GB model locally, they will not have the same experience as those who run the top of the line models.

Well you have to keep in mind that Nvidia has a 3 trillion dollar valuation. That kind of heavy valuation comes with heavy expectations about future growth. Some of those assumptions about future Nvidia growth are their ability to maintain their heavy growth rates, for very far into the future.

Training is a huge component of Nvidia's projected growth. Inference is actually much more competitive, but training is almost exclusively Nvidia's domain. If Deepseek's claims are true, that would represent a 10x reduction in cost for training for similar models (6 million for r1 vs 60 million for something like o1).

It is absolutely not the case in ML that "there is nothing bad about more resources". There is something very bad - cost. And another bad thing - depreciation. And finally, another bad thing - the fact that new chips and approaches are coming out all the time, so if you are on older hardware you might be missing out. Training complex models for cheaper will allow companies to potentially re-allocate away from hardware into software (ie, hiring more engineering to build more models, instead of less engineers and more hardware to build less models).

Finally, there is a giant elephant in the room that it is very unclear if throwing more resources at LLM training will net better results. There are diminishing returns in terms of return on investment in training, especially with LLM-style use cases. It is actually very non-obvious right now how pouring more compute specifically at training will result in better LLMs.

When it comes to comparing market cap, the more apt business relevant to AMD's MI line is Nvidia's data center division, and investors are probably rightly assessing that AMD will not dent Nvidia's market position any time soon. That said, AMD's data center GPU is growing at an extremely healthy pace and enjoys high profit margins, so they have proven their ability to execute in this space to a degree and as a business it shows a promising future.

When looking at the market cap, there are three main pillars of valuation - revenue growth, profit growth, and net income. If all three are growing, you are an industry darling. If two are growing, you are still likely to be valued highly. If you have only one, you are much riskier. If you have none, it's a red flag.

As of the latest earnings report, AMD profit, revenue and net income are all increasing. Intel, they are all decreasing. If analysts assume trends hold, AMD can grow into its valuation and Intel is currently heading towards being worth nothing unless they change their business. Simply put, a business that is losing all three of revenue, profit margin, and net income is simply headed on the wrong path for investors, and will be punished in an outsized way when it comes to predicting it's future value (ie, market cap).

I honestly don't think there is any algorithm. For all the bluster and commitment to being "data driven", none of the companies I've seen mandate RTO have provided any sort of data-driven reason why it needs to happen. Amazon's policy might as well be "Jassy feels it in his gut that RTO is better for the company so we are doing it".

All the communication of RTO invokes the most fanciful and vague references to "magical hallway conversations" and "increased collaboration" without a single data point to back up any of the claims.

It has been almost humorous to watch such stalwarts of "data driven decision making" turn up a giant goose egg with respect to actual evidence on such a huge, impactful, and far reaching decision.

This is not really true though. Congress is responsible for granting authority to the President regarding valid military targets. This is why drone strikes are only legal against targets recognized by Congress as security threats. It cannot realistically happen for the President to start targeting individuals outside of Congressional authority.

For your hypothetical situation to arise, Congress would have to declare members of Congress themselves as valid military targets.

It's really impossible to understand and determine before hand how the court would rule on any of these theoretical cases that may result as a consequence of this decision. It is up to further cases to actually establish was constitutes "official" versus "unofficial" capacities as President and we can absolutely not guess before hand what that entails. From the decision, it seems that only those duties constitutionally mandated would fall under the "official" capacity, with quite a lot of leeway for determining how to evaluate individual actions.

Also I think we should all be reminded that there is separation of powers for a reason. The President is ultimately largely beholden to Congress. The government cannot sink into a dictatorship without the explicit approval of the majority of Congress. It is Congress' duty to remove Presidents from office that it feels are a danger to the country.

All these checks and balances still exist and will still be enforced. The President can not unilaterally go off the rails as many of these extreme hypotheticals seem to be implying.

AWS also recently ended support for Mysql 5, so if you had an RDS instance with that version running past the cutoff, your support costs ballooned exorbitantly.

My main problem with Silicon Graphics (& have the same problem with Sun Microsystems) is that they just tried to do too much in propriety hardware and completely resisted standards. Microsoft & IBM "won" because they made computers with actual upgrade paths and operating systems with wide support among upgrade paths. With SGI/Sun you were very much completely locked in to their hardware/software ecosystem and completely at the mercy of their pricing.

In this case, I think the market "chose right" - and the reason that the cheaper options won is because they were just better for the customer, better upgradability, better compatibility, and better competition among companies inside the ecosystems.

One of the most egregious things I point to when discussing SGI/Sun is how they were both so incredibly resistant to something as simple as the ATX/EATX standard for motherboard form factors. They just had to push their own form factors (which could vary widely from product to product) and allowed almost zero interoperability. This is just one small example but the attitude permeated both companies to the extent that it basically killed them.