bonesss

Having built a hobo version of something similar (serving a minimal subset of the Kafka API on top of CosmoDB): there’s also the hybrid scenario where the cheap serverless streams are used for scalability and broadcast while a low-latency core is maintained on sharply reduced compute resources. An 80/20 approach that saves a lot and, in our case, reduced cluster (mis)management risks at the same time.

In our case BLOB and large document transfers were handled in parallel, merging them together through object storage is a highly appealing package. Great work!

Like how Tony Soprano sold sandwiches and ‘protection’ at the same time. Two income streams are better than one.

Blowing something up so hard it almost gets into space is a feat of engineering, even if all it does is rain debris from ahigh.

I believe it’s checkbox-filling in the acquisition process for enterprise contracts, like a lot of other half-baked capabilities in their flagship products.

It doesn’t matter how they work, it matters you can tell leadership “yes, they have that too” to maintain existing contracts and vendor relationships.

Speculation, but: if tech over hired during COVID our section of the market could be experiencing a corrective slowdown while others are booming in aggregate.

I think we can all agree No True LATAM company would have 10,000 employees.

Or, in parallel, the research showing LLMs intelligence will operate on an S-curve, eventually hitting a long valuation deflating plateau, is dead on the money and The Big Guys are trying to delay that inevitability for as many quarters as possible…

I’d add on to that lifecycle demands: a development project with no users has a tight cheap feedback loop with high quality bug reports with relatively obvious bug sources, once issues start working through layers of QA and customers the per-issue costs raise by several factors (50x to 200x are numbers I’ve seen from research, YMMV).

I would feel very confident slapping something onto Chrome ‘back then’ with only internal users. Today, with millions of users and CEO-level attention to screwups, I’d use a lot more time per issue. It’s a fundamentally different risk/reward/confidence situation.

It’s pelicans, all the way down.

> The software will suck and will not really innovate, but it will make a ton of money

Microsoft’s goal isn’t to provide high quality software that outperforms other software, their goal is to provide an obvious and reasonably proximate answer to clueless execs trying to check things off of a procurement list so that existing contracts can snowball and competition is excluded.

We, collectively as a global species, really lost out when MS wasn’t divvied up in their anti-trust trial. Office as its own corporate entity would be in a radically different place right now, as would their enterprise customers.

How do you work on Windows in the age of LLMs?

That’s easy: you activate WSL and then ask the LLM how to get networking and stuff working since they changed it with WSL2. You need a config file flag to setup network mirroring now.

I babysat for a military fighter jet mechanic when I was young, he said when they dismantled and rebuilt planes there was a drawer for all the spare parts that were left over.

A large majority of windows devs choose Apple hardware has to major challenges for MS.

If devs just want shiny flashy then MS has failed at providing that directly or through their partners. If devs are able to make credible technical reasons, year after year, that sustain IT & budgetary scrutiny, that also is a failure from MS.

Personally “being able to run docker without this bs” was credible for many years, tech that now is a de facto expectation for most .NET dev jobs and well-integrated-ish into windows and WSL.

Sam Altman says a lot of crazy stuff, but I don’t think that is where the ‘bubble’ is coming from.

In recent months we’ve had the first automated unmanned amphibious assault, the daily drone count in our hot wars is jumping by leaps and bounds, and arms suppliers are promising future drone shipments in the hundred thousand unit range. The ten year picture for reactive combined swarm intelligence on the battlefield is promising to be widespread, highly lucrative, and in need of constant adaptation to near-peer efforts. Datacenters in space are dumb, datacenters in space to power orbital weapons networks and rapid response capabilities make sense.

On top of that we have international trade, scalable customer service, and a first pass 80/20 answer for businesses focused elsewhere. Shitty, maybe, overpriced, maybe, but useful enough our grandkids are gonna use ‘em.

In both cases, as well as potential new LLM-like tech, there’s an argument to be made for being a leader now to dominate the future. That means compute and tech positioning, and memory & GPU deals.

YouTube was a money loser, Google was ‘losing’ money on them for years, YouTube didn’t have a sustainable business model. YouTube was the biggest, though, and whatever premium Google paid to be #1 then meant they were #1 when the online video business model matured. Now they’re printing money with a platform outcompeting news, social, and video platforms.

In .Net-land we have access to a lot of the underlying MS library code, and reflection/decompilation if needed.

MS has layers, there are A-teams, B-teams, and so-on. Simultaneously they are dropping both world class work and, to your point, a statistically much larger pool of just pretty good code.

A lot of that work is perfunctory and somewhat bloated/weakened by the scale of publishing and their internal tech-political requirements. Components for component I’ve seen in-house replacements for many MS projects from mid-tier devs that are better, smaller, and less brittle over time.

I’d compare it to baking: a home cook or tiny bakery have the advantages of time and focus, perpetual consistency at scale is a whole different baking challenge, one that doesn’t necessarily yield better individual loaves. It doesn’t mean WonderBread is bad at bread, just that they’re optimized for something other than maximal loaf quality.

I have a sneaking suspicion that if everyone proffering an LLM coding opinion stated their languages and area of focus/expertise we’d all be in a lot more agreement about the output…

Example 1: “as a JavaScript front-end developer working with React and node, I think LLMs save 500 hours a week and are better than anyone on our team.” <— easy to believe

Example 2: “as a backend Erlang developer working on life-critical systems the inconsistent output makes LLMs dangerous and the code is tidy but terrible at scale.” <— fully congruent with the first example

Taking a multi-decade perspective, I wonder if our general analysis is focused too much on the initial wave of LLM tech and current gen GPUs.

Owning a massive data center connected to water and power and network that can be targeted or converted to developing needs seems like a decent problem to have for the big cloud companies. We have compute hungry companies and media, in addition to cryptocurrencies etc, and we’ll have more of them in 2045.

I don’t know if I’m underestimating how purpose-built these datacenters are, or overestimating the accountants in the corporate vehicles building them, but the broader situation doesn’t seem as fragile as 1929 or 2008 (even if the businesses are overvalued and LLMs fall totally out of fashion).

Xursor?

> making goblins less likely to over-appear or show up in inappropriate contexts

So inappropriate goblins are still likely, just less so…

Hey, remember when tech bugs were things like buffer overflows or cross-thread performance impacts? I miss the days when our war with system goblins was purely metaphorical.

The constant memory wouldn’t be so bad if it weren’t dumb, wrong, and forced.

I hit the wall with it several times today trying to refine some text for a job application. The fact I considered doing babies first Rust project last fall lead to constant non-productive interjections and digressions about my supposed Rust skills and the Rust ecosystem.

Trying to create an unrelated spreadsheet to model an investment resulted in broad and incorrect criticism of my choice of spreadsheet tools, explaining in horrendous programming analogies why and how I’ve misunderstood how a spreadsheet works. “Think of the XLSX as a compiler…”

There has been a palpable down-step in communication & execution.

Longer commutes aren’t just about work, and that time in the car is not free. Kids, kids sports, kids social lives, and kids schooling can leave people trapped in their cars day in and day out trying to reconcile modern urban socialization expectations and the lower property prices they enjoy because of the minor impracticality.

In my specific area the housing price reduction of moving out of the city balances out quickly by the need for a car and/or a second car (our cheap tiny EV works fine for short commutes), particularly if you compare the relative interest rates. I imagine market forces make it so, the financial tradeoffs are reflected in the pricing.

Further: as the only available housing at a low price point collects in smaller towns the impact of borderline purchasers becomes more pronounced. As neighbours I don’t care, but as a parent who has seen kids get an unfortunate personality remodel from a near-peer bad influences and listened to the experiences of my parent friends who live out of town… the contemporary local culture needs to be considered. It’s not necessarily the cozy small town upbringing we expected in olden days, there has been generational turnover.

Another moment of potential irritation: companies being bought out or losing their network supplier giving that third company a business relationship with you.

I keep waiting for a company to start pushing an “anti-AI AI”, Apple being my most likely speculative candidate.

Flagging, highlighting, filtering, and avoiding AI slop online seems like a very lucrative capability to bake into some consumer hardware. Get your social media, and music services, back to the ‘good old days’.

Domain modelling as well: we have clear programming concepts, but the kind of autocomplete that can autocomplete a customer sales process needs to be able to refer/leverage ontological connections on words like ‘bass’, ‘season’, ‘rebate’, and ‘jamboree’.

Intuitively it feels like focused models should be better models, but human programmers are ‘better’ knowing programming alongside general stuff. We’re not reduced by reading economics or Tolkien, removing such knowledge would be premature optimization.

My kids have bank accounts to learn how to manage money, banks, and training: getting used to having money in the bank before, during, and after a trip to the mall. My hope is to let them screw up their finances when they’re little instead of 20-something.

We’ve also setup tax-deferred retirement investment accounts for them. $1 at 20 can 70x or more by retirement. Mostly it’s the mental training though. Being ok “losing” money during a market correction, saving for wealth in parallel with saving to buy, seeing interest and returns over time, and having a long term plan.

To my speculative thinking the downslope of this LLM hype bubble might be different shaped because of the underlying assets and geopolitical situation.

Securing data center land, contracts, water rights, and execution capacity doesn’t seem like a terrible position to have in a digital, cloud, ML, crypto, and ‘prediction’ heavy future. Especially for the big players who are also cloud providers who might capture big chunks of secondary growth even if they fail in their LLM effort (and ditch the hardware?).

The LLM stuff seems very over priced, but also Ukraine is making a million or whatever drones a year all with a need for ML-powered planning, routing, and terminal guidance. Elons space data centres seem kinda dumb, but in a world where Palantir needs to be tightly in the loop for orbital or near-orbital operations, or autonomous orbital defence... The worse things get in those ways the stronger the long-term positioning of the cloud giants to build or capture critical defence operations and associated spending.

We’ve learned Skynet as AGI won’t come from the tech, so the bubble gasses out. But Skynet as Skynets military is here, now, and the AI/cloud providers own key logistical elements, so the bubble loses gas slowly.

Unlike the dotcom era, infrastructure is being built by mature highly profitable companies with broad product portfolios to exploit future trends and needs... Data-centres or even their power/water contracts seem like something with a lot of value even if we take a cynical view of LLMs profitability.

Am I overstating the case? My understanding is that a data center tends to be ‘purpose built’, so they may need a gutting for repurposing, but assuming a lot of ‘sobering up’ I’m envisioning several giant cloud providers with excess capacity and a scaling potential.

Be it Jevon’s paradox sparked by cheap compute, another huge tech fad, or a ML breakthrough that brings another kind of model to the forefront, we’re likely to want a lot of compute at some point. It doesn’t seem like bad long-term positioning for the tech giants or investors.

Ok, but if they had instead bootstrapped industrial nuclear technology into our known design-space we’d have actionable scalable solutions to transportation fuel generation (OPECs geopolitical sway, Chinese economic stability), mass desalination, environmentally friendly electrification of industrial processes, and industrial heat applications that could save incredible amounts of greenhouse emissions on a yearly basis. Instead we are reliving the oil dynamics of the 70s because of our poor industrial accounting, doubling down on emissions heavy tech in energy intensive operations, and maintaining harmful environmental chemical pollutants in our processes farmed out to the developing world…

Renewables in isolation have an entrenching effect on other aspects of our global emissions footprint & some likely causes of WW3. Humanity has an energy and greenhouse gas crisis, not an electrical generation crisis. Transportation fuels are central. Delaying our high energy future results in, among many other things, intensely polluting data centres instead of the obvious pairing of 24/7 large scale steady energy with no emissions to our increasingly digitally dictated economies.

Wind/wave/solar is nice, cheap enough anything pays for itself, but there is a bigger picture. Tribalism and anti-nuclear sentiment are not in line with our short to medium term environmental goals and are only marginally relevant to the long term and existential ones. We have already cursed generations by our trepidation, reluctance, and submission to monied interests. We cannot get an iPhone 20 without building and iterating on 19 models first.

I love it. Where else can I give a jumbled mostly-ok explanation of the standard model and cosmology until the other persons brain turns off from overstuffed-boredom and not get fact checked?

I do so wish everyone espousing 1:1 casualty ratios as necessary & fair, and that anything else is immoral, would take stock of where they live and specifically how those countries avoided falling to the axis powers in WW2…

And, courage of their convictions, if their own people are so inherently evil for making “the other bastard die for his” they should be renouncing citizenship, yielding their ill-gotten wealth, and self-immolating for their nations eternal shame.

Categorical reasoning as applied to war does not yield a tit for tat, its category versus category. All of column B to preserve any column A. “Not worth the bones of a single British Grenadier”.