greenail

you can set the "temperature" which is a lever on how stochastic the prediction is. If you are doing your own inference this is clear and easy. If you are consuming tokens this is outsourced.

maybe it doesn't turn them into assholes. maybe it just frees them to act naturally.

if "the ai" was delivering value you'd think it would be lowering both cost to develop these dev tools as well as the cost to support them. In that light the move is odd unless they don't know how to use AI to lower those costs or it cant.

maybe now isn't the time but at some point it needs to be better understood which models are best for which types of programming or styles or languages. These models are not all the same for every language. The harness is also a factor. Python seems to be somewhat an exception today but that may not last. Another question might be: is there a pattern where you prototype in language X and implement in language Y. The models seem to be very good at porting code. I've used this pattern with python -> c++ SDL to squeeze out performance after I had a working gui. Has anyone measured this in terms of speed (wall clock) and in terms of token efficiency?

my experience with elixir is that if you tell the LLM "let it crash" it is productive. If you don't tell it this it hides errors and writes crappy tests. So, maybe it is go and not concurrency programming.

I think GB10 is a bit of a counter point. There are tons of features that are not implemented for GB10 which was released 8/2025 It isn't all roses on the cuda side.

you can buy dual coaxial shaft steppers ( X40 ) for car instrument panels open them and remove the hard stops. A very small magnet and 2 hall sensors gets you end stops.

This is a bit of a simplification of how enterprise sales works. A few extra dimensions are sometimes required. For "decision makers" their own personal view of reality may not in fact be accurate. You sometimes must vet that what they tell you (and may in fact believe) actually reflects the actual power dynamics of a big enterprise. There are extra dimensions to consider; The first is: are they in a cost center or are they generating revenue. The second is: are you trying to get new technology adopted or is this considered "standardized". The third is: who might perceive your product as a risk, or a threat to their power? These data points inform your sales strategy both in who you sell to, who you avoid, and when you might choose to engage a particular persona, when to go small and when to go big.

this doesn't add anything to the conversation.

This doesn't work when you are in a relationship with someone with a personality disorder like BPD. They use it to hurt not to heal.

It is not who you know, it is who knows you! Seems like a small nuance but realizing this changes how you think about building relationships.

+1: the globally unique identifier that is reused across multiple accounts is an antipattern in my opinion. This is just as bad as reusing passwords. There should be no need for anyone to have to reuse email user id's or phone numbers or credit card numbers. These should be unique per online "relationship", the way your CC chip creates per transaction identifiers. The fake account problem needs to be solved in another way. You simply can't take these GUIDs back once they are leaked and that leaves your only choice to burn them or just accept the risks of having that data leaked all over the net.

IaaS lockin should be minimal if you make the right investments in automation and think hard about dependencies you can't unwind quickly. An IaaS deployment without automation misses the point of cloud. The easy test is "can you turn it off and get it back on without manual effort?" If the answer is no you are likely locked in. This would also be true in a colo or datacenter. The biggest benefit to IaaS is that you can make changes with reduced risk. You move away from 3 year cycles and big upfront commitments in exchange for the ability to make big changes quickly. If you move to the cloud and don't get that benefit you have not gained the core value. If you don't need that benefit you may not need IaaS cloud.

I have seen many cloud moves that don't leverage autoscaling and I've been wondering why anyone bothered. Is this just an inability to justify investing in modernization? Is lift/shift first a good strategy, it seems to me you miss the opportunity to fix long term tech debt if you just move.

if you flip the script you'd never ask an exec you don't know for a meeting without giving context. This is common courtesy if you do it without seeming demanding.

"Looking forward to our discussion. Can you give me a bit more context for the meeting and an agenda? This way I can be sure to have enough time to properly prepare. Thanks..."

would you say no to that?