emenel

yes! totally. and even at small scales the design of the process and component parts is very directed and intentional. so the resulting model is a model of all of those choices.

2026-05-06 19:31:22 UTC

- reply

sort of sometimes lol. there’s no mystery in how generative “ai” models are created and work though.

2026-05-06 19:26:45 UTC

- reply

imo one of our mistakes often is saying these are black boxes. they’re not really. even if the specific weights and data are impossible to fully inspect, they are still performing the computational process as programmed. we can’t verify outputs against inputs because of the complexity, but each component of that complexity is well known.

2026-05-06 18:21:17 UTC

- reply

agreed, it is like all those other things. any form of analysis is limited and contextual.
so all i'm saying is that this specific method of analysis (machine learning and its offshoots) does a specific thing and produces a specific kind of output (a set of statistical weights etc). the way this is done was designed for specific reasons, choices and decisions were made.
when you employ those methods and tools you are also enacting those decisions on your data, and the resulting model is a representation of those decisions as much, or more, than of the data itself.
the specific model of that specific data will be unique, but it's a designed representation.

2026-05-06 17:35:01 UTC

- reply

how do you observe it? what is observed and observable? what tools are used for observation and how do they work? is observing neutral? and how do the observations get recorded and turned into a model? …

2026-05-06 17:19:52 UTC

- reply

i'm responding to when you initially said that the models "just are." i'm saying that they aren't like that because they are the direct result of an already specified and determined process that can only produce a certain kind of model... and we're best to look at the models are artefacts of that process (all that it entails) rather than something that just "exists"

2026-05-06 17:08:24 UTC

- reply

i mean that the specific types of analysis aren't neutral. someone is always deciding what and how to analyse, and then how to model those results. it's a designed process...
like how science is a specific methodology and produces specific kinds of outputs/outcomes that aren't neutral or objective (contrary to the current popular understanding of sciences).
to go back to the original point of all of this (lol), when you create a ml model of a dataset it is a designed process that produces a specific and limited kind of result, regardless of the data. that is an intentional set of decisions made by somebody.

2026-05-06 16:59:50 UTC

- reply

i'm not saying the results are designed. i'm saying that the type of results are determined by how the process is designed. it will only create models that model the data in the way that the process is designed. so the output doesn't represent any natural representation of the data, it represents that specific process.

2026-05-06 15:57:05 UTC

- reply

sure... but in this case the output, the model, is the result of a specific designed process. that process creates this artefact, and this artefact only. a different process would produce a different artefact.
a set of data doesn't have a natural model, order, or set of relations. those are all choices and decisions.
"a model" in this context specifically means a computational statistical model of datat-point relations based on frequency and context-frequency. it creates the weights of the relations from a self-refining gradient descent ... (and etc whatever else the training code does with the data).
that's a very specific way of analysing and representing the data that was chosen by people to do a specific thing.

2026-05-06 15:32:29 UTC

- reply

when you make a model with your own sound files, how are the processed into the model? what attributes are modelled? how are those relations represented? what attributes are left out? who made the decisions about how the data becomes a statistical model?
those are the design ... and in this case they've already been done by, i.e., google (tensorflow) or etc.
the type of model that the process produces is a designed artefact, even if the specific contents of the model (weights, nodes, etc) is different for each output.

2026-05-06 15:20:43 UTC

- reply

sure, it could be arbitrary data. it could be a list of random numbers ...
the question then is what do you model? how is the model derived from the data?

2026-05-06 14:57:59 UTC

- reply

tbh it’s not the quality, it’s the ontology. what is being modeled and what does the model represent? there’s no neutral or objective answer. someone (group, org, corp, etc) designed a system that models a specific kind of data in a specific way.
any and all modeling is designed…

2026-05-06 14:27:41 UTC

- reply

what's a netword? imo it's a model of relations...

2026-05-06 14:21:43 UTC

- reply

yes. because you’re using a system that already determines the type of model. the code you use for training a local model is the design of the model … it’s the algorithm that takes your input data and produces a specific type of model.

2026-05-06 14:14:36 UTC

- reply

even if the data is arbitrary you have to choose how to model it. the data itself has no natural model. the model is always designed.

2026-05-06 14:11:17 UTC

- reply

yes. what is it modeling? what relations between data are important? what does “training” mean in detail? those are all choices that change what is being modeled.
models are a human invention… the dictionary is also an llm, it models the alphabetical relation between words and associates them with definitions …

2026-05-06 13:57:08 UTC

- reply

it’s obviously built on massive amounts of data. but it is not an objective or natural outcome or representation of that data. it’s a specific, designed, and created artifact that uses the data in a specific and intentional way.

2026-05-06 13:46:36 UTC

- reply

the models themselves are intentionally created is what i'm saying. from data selection to training method to operation... there is no "natural state", it's entirely designed... even if the output is unpredictable (to a degree)

2026-05-06 13:39:56 UTC

- reply

but there's a process to derive those weights, refine/edit the data to produce a specific kind of output (done by thousands of abused and exploited workers), manual editing of the resulting weights to make the output "safe" (aka censored and also done by exploited workers), put the model into a piece of software for specific kinds of interactions that are also guided and constrained .... and querying the model (aka writing a prompt) is also a known process ... computers can never be "just the way it is" because left on its own a computer does nothing. anything it does has to be translatable to linear instructions for the cpu.

2026-05-04 16:34:36 UTC

fuck dell forever.
https://www.dell.com/en-us/blog/year-of-the-linux-laptop-omarchy-on-xps/

2026-04-07 01:11:22 UTC

- reply

also, it’s terrible.

2026-03-11 15:46:17 UTC

"gotchas" are always bad critique

2026-02-23 05:12:22 UTC

- reply

do you know the history of urbit? if not, it's worth looking into... brainchild of curtis yarvin...

2026-02-01 20:27:38 UTC

mood today: this mortal coil

2026-01-29 23:40:33 UTC

- reply

the topics they identity as “extremism” are highly questionable.

2026-01-27 01:32:59 UTC

- reply

awesome! it is competitive, that’s true. i hope they get in somewhere you’re all happy about :)
i work at UofT, happy to answer any questions :)
i also just applied there for a phd …

2026-01-27 01:27:20 UTC

- reply

consider canada …:)

2026-01-27 00:14:46 UTC

- reply

“value in schools” should not be a benchmark of any kind…

2025-12-17 16:34:33 UTC

remember that one of the the main reasons you can't easily self-host/community-host email is because of google...

2025-12-09 20:52:33 UTC

Stop only citing Christopher Alexander challenge

2025-10-09 10:37:10 UTC

rip framework unfortunately.
supporting fascists in tech is a big problem.
https://community.frame.work/t/framework-supporting-far-right-racists/75986/25

2025-03-28 22:54:28 UTC

It should be totally normal for well made software to not need updates. Ever.

2025-01-30 16:30:55 UTC

- reply

they think that everything they do is entirely self-invented... no history before "i" thought of this amazing new idea... which ultimately means that there's no growth and no opportunity for actual intellectual or ideological development... they're stuck in an arrested development of immature toxic american white maleness. everything is terra nullius. tbh, even less objectionable thinkers fall into this trap. zitron is like this too, even what i agree with what he's saying it's always without citation...

2025-01-30 15:42:39 UTC

- reply

it’s also all part if the silicon valley ideology and culture. being ahistorical, “self-made”/“free thinking” (aka anti citation)… it’s so insular that it is the intellectual equivalent of how teenagers think they’ve invented everything …

2025-01-15 22:32:10 UTC

- reply

and to be totally honest, that friction and constant battle was a big part of what made me leave design/ixd... even though i miss so many of the people.

2025-01-15 22:31:31 UTC

- reply

i'm more than happy to be irritating about important things ... and somehow i always find myself on the unpopular but ethically "correct" side of these things (correct in quotes because it's a moving target...)

2025-01-15 21:57:09 UTC

- reply

this is me in all social and professional situations now.

2024-08-29 01:31:37 UTC

- reply

i propose: baguelette

2024-08-29 00:19:59 UTC

- reply

depending on class size, i also often start class with an optional sharing of “good news”, which is just anything the students want to share that is positive and happened in the last week. It doesn’t have to have anything to do with class, and it’s totally optional. I’ll start by sharing something, and then there are always at least 3-4 other people who want to share if not more. Sometimes I have to cut it off after the first 4-5 because we need to move on. Both of these techniques I learned from one of my profs during grad school, @npub1qma…49kn, and then adapted for undergrads… since we all know it is not the same as a graduate seminar :)

2024-08-29 00:16:51 UTC

- reply

this is absolutely true in my experience as well. I also try to have highly structured discussions, using a think-pair-share type process, with the addition that they also have to propose the questions for everyone to discuss: https://en.wikipedia.org/wiki/Think-pair-share