33MHz-i486

theyre getting high on their own supply, and instead really need to hire some senior engineers

at this level they just decide and spin up a swat team to execute it in a couple weeks without politicking. the bureaucratic ways, reviews are just for the low levels, to keep them busy with feature scraps while they mostly do operations

I think the subtext of the last few weeks is the Anthropic was becoming severely capacity constrained (or approaching that). They seem to have had to sign two somewhat adverse contracts with Amazon and Google in short succession. suddenly model quality is back up again.

also if youre in a large organization, this is a great way to sabotage other peoples projects while elevating your stature. Require that they go evaluate alternatives and prior art, and write a slew of analysis and decision documentation

its not “normal” when companies have 10s of Billion in net profit per quarter

Axing low/negative ROI product lines, sure. But recently these cuts have been across-the-board and in product lines that are net profitable and have strong technical product roadmaps. Moreover they are firing longer tenured (expensive) engineers

I understand they’re managing a transition to a capital intensive strategy but the whole era reeks of stock price focused financial engineering and these large companies flexing oligopoly power in the face of their customers and the labor that builds their technology.

OpenAI acquired his company for Billions. maybe the products flopped but he did fine for himself personally

true but the government will inevitably demand their own stanza of (blocking) system prompts in the major AI services. then they will ban local LLM and foreign ones.

the key is to understand which LPs apply to your L+1 and which apply to you (hint: its not the fun ones)

its not even AI. most of the cloud offerings are commodities now.

the marginal technological direction is determined by middle managers whoes primary motivation is “what new customer facing feature can I launch at this years re:invent and build a little empire” (of course this is a shrinking offering as tech debt and complexity pile up)

junior engineers are burned and churned on execution, seniors are project managers, principals just do high level reviews & high level fire fighting (note not actually leading the tech)

director and above just their spend time on “what to kill” or “who to fire” as priorities change every 6 months

In case it isn’t completely obvious from this, it is indeed hellish to work there. Most of AWS has a 2 reviewer requirement. If AI is writing most of the code (and it is because most Amazon code is copypasta boilerplate) you need 3 developers to sign off to ship anything. But of course due to headcount attrition, managers have ~1.5 developers to a project. Meanwhile the L8 manager is doing nothing except stack ranking each level of engineers according to number of commits merged & customer facing features shipped, and firing 15% of the bottom at the end of each year. There is no notion of subject matter expertise or technical depth, theyre happy to replace whoever with fresh-grads (theyre all just cogs anyway right!). Between that and voluntary departures, teams having 80-100% turnover every 5 years is basically par.

Also while this is happening most developers are getting constantly hammered by operational issues and critical security tasks because 1) the legacy toolchain imports 6 different language package ecosystems and 2)no one ever pays down tech debt in legacy code until its a high severity ticket count in a KPI dashboard visible to the senior management.

because demand is weak and the product markets are saturated. there are dimishing returns to increasing investment. so these companies switch to managing their earnings ratio. if you cant grow revenue, then cut costs.

except they planned the January layoff in October or even earlier over the summer.

having worked there. Amazon has toxic managers, culture that turn ICs against each other. no tech vision. insane politics. low caliber people gatekeeping.

Their stock will go up the next year or two only because of their luck of partnering with Anthropic (back when they were a distant second choice to OpenAPI)

i dont know that its such a great thing in the end. Uber/Lyft is 50-100% more expensive now than taxis were before. Theyre entrenched in different ways.

kids dont make any income. a weekend trip for 4 in the same timezone costs us $3k to 5k. a cross country trip is 10-20k.

DINK > Solo >> anything else

that doesnt really work if they have to compete on price with foreign producers that have lower labor costs, looser environmental regulations, deeper supply chains and better process knowledge. might as well eat the tax (waiting for monopolistic opportunities) or light the money on fire

we were talking about operational access to the payment system. you are conflating the situation at USAID which may or may not by illegal, idk.

the legislative branch can form administrative departments and prescribe their function however the president has already defined powers to impound funds and remove senior administrative officers and appoint/remove low-level staff. how these things intersect will be sorted be the courts.

executive actions (by-passing what should be legislation) have been increasing the last few decades. the various media companies plainly do make choices to portray some actions as nothingburger or crisis depending on their political alignment with the party in power.

the issue with the left-media and Trump is they outrage clickbait a bunch of events that are insignificant in terms of outcomes. Should they alarm about Jan6 yes. should they alarm over minor personnel at treasury or some dumb unserious thing Trump said at a press conference, no. This is how the media loses all trust in themselves broadly.

yes guy who won an election making changes to a department within his purview is a “coup”

the media and people who take their bait are insufferable

the strength of the US defense commitment is likely proportional to the strategic value of the economic assets they still hold. the taiwanese have every incentive to do just well enough at the AZ plant for the $39 Billion checks to clear and no better

the really dumb thing about working at AWS is they pay so much lip service to Ops, literally you can spend a third of a week in meetings talking about Ops Issues, but not a single long term project to improve the deeper architectural problems that cause bad Ops ever get funded.

they should have focused on building light-duty utility vehicles based on the Model Y platform. A ModelY Truck could be like the Ford Maverick or a 90s Tacoma, the Van could be like Ford Transit Connect. It would be significantly faster to market, re-using the some of the production lines, and there's a huge market (landscaping, residential trades, short-haul delivery, etc). Elon was high off his own supply the year he launched CT, Semi, & Roadster

feel like the board is too small and a handful of them kicked him out over personal agendas - which are incoherent and hence they cant explain it publicly

- Ilya for decel (but really tech fame jealously)

- Adam for a CoI with his startup (but really founder fame jealousy)

- Helen Toner for Sam disagreeing with her research in EA, safety and decel

probably Tesla Elon right before the inflection point

and somewhere 2 teams of Software Engineers are tasked with Test Case 7462628: a Video classifier to detect when the AV has stopped on top of human and pinned them beneath its wheels

If your capital cost is low, you can buy growth (sell below cost) and use that to distract investors for a long time. Capital costs are not going to be low anymore. A lot of investors are going to blow up and be more skeptical/disciplined.

Youtube should be forced to pay people minimum wage for their videos

Etsy should be forced to pay minimum wage to their makers

ok but make them retire at 65. before mental decline really kicks in

they are anti-displacement provisions in the bill.

gentrification/displacement can't happen if upzoning happens at scale of the entire regions/state. e.g. techies won't move to Oakland if there's a bunch of new housing in San Francisco, Palo Alto, Cupertino etc.

biometrics of the face, body or eye are mostly simple algorithmic transformations of photographic images. I find it hard to believe that "biometric data gathering" of Illinois residents can have massive civil liability without effecting all public photography and publishing thereof.

there's also the catch-22, that if an entity is pursuing legal collection outside of Illinois (for example of the Flickr dataset from the IBM case) it has no way of knowingly excluding Illinois residents without identifying them by using an (illegal) facial recognition dataset in the first place.

its been my observation that most researchers/DS prefer PyTorch because it lets them hack in python and most production software engineers will prefer models be written TF because of effortless portability and performance of TF Graphs.

I work on a team that does the latter and lately DS have been handing off PyTorch models that we cant scale or make performant because Torchscript doesnt really work with any realistic code complexity and authors include all sorts of random python libraries. So we can't load models in C++ or get them under 50ms.

So the framework divide very much feels like dynamic vs statically typed languages. People that dont have real production demands love dynamic languages for the productivity.