nr378

5 ms·

1.
Is Anthropic the New Netscape?(deadneurons.substack.com)
2 points by nr378 12h ago|1 comments
2.
The Quiet Unwinding of Microsoft and OpenAI(deadneurons.substack.com)
6 points by nr378 8d ago|0 comments
3.
The AI Capex Bubble and the Parallels to 5G(deadneurons.substack.com)
1 points by nr378 9d ago|1 comments
4.
Software Has Never Been Capital Light(deadneurons.substack.com)
2 points by nr378 10d ago|0 comments
5.
by nr378 14d ago
The competition for the least reliable developer service continues between GitHub.com and Claude.com...
6.
by nr378 21d ago
Thanks, this option is good to know. We're currently on "GitHub Enterprise Cloud" on github.com and are affected by this outage (even though we use self-hosted runners!), but we're not on "GitHub Enterprise Cloud wi
7.
by nr378 22d ago
GitHub needs to completely bifurcate their enterprise/paid services from their free services at the infra level.
8.
by nr378 1mo ago
> Uber losses were $12B over 5 years. Uber burned through roughly $32 billion in cumulative losses before reaching sustained profitability. The rough timeline: - Founded 2009, and lost money every year for about 14 years - Biggest single
9.
by nr378 1mo ago
High growth companies often have significant negative cashflow during the early high growth era, followed by positive cashflow in the years later down the line. This phenomenon is known as the J-curve[1], and Uber is a good example of how t
10.
by nr378 2mo ago
> The frontier LLM labs run on a huge fixed cost and very low marginal cost. > Imagine that you want to buy a few B300s to run GLM 5.2 and rent the service out to other people. How could this business be viable and sustainable in the
11.
by nr378 3mo ago
Arguably Google is both (with GCP and Gemini).
12.
by nr378 4mo ago
Claude Teams and Claude Enterprise are 2 distinct plans. Simon is right that Enterprise seats have no included usage (and so all usage is charged at API billing rates), whereas Teams seats do.
13.
Why Integrated Copilots Suck(deadneurons.substack.com)
3 points by nr378 5mo ago|0 comments
14.
Where AI Will and Won't Replace Us(deadneurons.substack.com)
2 points by nr378 5mo ago|0 comments
15.
Why the most valuable things you know are things you cannot say(deadneurons.substack.com)
162 points by nr378 6mo ago|63 comments
16.
Thermodynamics, Organisations and Governments(deadneurons.substack.com)
2 points by nr378 6mo ago|0 comments
17.
An Open Letter to Mr Zuckerberg(deadneurons.substack.com)
3 points by nr378 6mo ago|0 comments
18.
Redux for Enterprise Context(deadneurons.substack.com)
2 points by nr378 6mo ago|0 comments
19.
by nr378 6mo ago
Yep I think you can reasonably argue that immutability + strong conventions are the most important dimensions (as opposed to FP vs. OOP, as much as I like FP and dislike OOP): Immutable by convention + Strong conventions: 91.3% - Elixir 97.
20.
by nr378 6mo ago
The data doesn't well support the claim that FP is best. Elixir tops the table at 97.5%, but C# (88.4%) is OOP and scores almost identically to Racket (88.9%), and Ruby (81.0%) and Java (80.9%) both outscore Scala (78.4%), which is exp
21.
Agentic Context Management: Why the Model Should Manage Its Own Context(deadneurons.substack.com)
7 points by nr378 6mo ago|0 comments
22.
by nr378 6mo ago
> 3. Storing it the way this article presents makes it usable for agents, but not humans. Whereas the point of knowledge graph, ontology, etc is to create the same layer for both humans and AI to interact with If storing it this way make
23.
The Great Rotation from Bits to Atoms(deadneurons.substack.com)
2 points by nr378 6mo ago|0 comments
24.
Housing: The Greatest Policy Failure in the Western World(deadneurons.substack.com)
4 points by nr378 6mo ago|0 comments
25.
by nr378 6mo ago
Dario has made a specific cohort argument here. His numbers (from various interviews) are: you train a model in 2023 for $100M, deploy it, and it earns $200M over its lifetime. Meanwhile you train the 2024 model for $1B, which goes on to ea
26.
by nr378 6mo ago
Based on the docs and API surface, I think the filesystem abstraction is probably copy-on-mount backed by object storage. I suspect it works as follows: when a task starts, filesystem contents sync down from S3/R2/GCS to a local d
27.
AI Is Not Going to Kill Software Engineering(deadneurons.substack.com)
3 points by nr378 7mo ago|0 comments
28.
Software Engineering Isn't Dead Yet(deadneurons.substack.com)
1 points by nr378 7mo ago|0 comments
29.
by nr378 7mo ago
Oh that's quite a nice idea - agentic context management (riffing on agentic memory management). There's some challenges around the LLM having enough output tokens to easily specify what it wants its next input tokens to be, but &
30.
by nr378 7mo ago
Nice work. It strikes me there's more low hanging fruit to pluck re. context window management. Backtracking strikes me as another promising direction to avoid context bloat and compaction (i.e. when a model takes a few attempts to do

More ›