<aside>
🧟 Put your two questions here. Please include your [name.](<http://name.Is>)
</aside>
This question feels kind of broad but what is a good approach to defining guidelines to build a markov chain to generate text?-Paridhi
I think the question is a little broad, so i'm not quite sure what answer you are looking for.
You might want to split it up into more focused questions?
If you are thinking about "how do you generate text with a markov chain" there is a simple markov chain code example on the chapter.
Regarding machine creativity, I’ve always wondered up scenarios of machine auctions where machines would collectively decide on what’s worthwhile to them. Would a scenario like this be an indicator of creativity? - Franco
interesting that "collectively decide on what's worthwhile" partially defines aucitons and ellections.
> Would a scenario like this be an indicator of creativity?
I'm not sure. You could cetainly program machines to participate in auctions following rules and hueristics (and people do, and they make a lot of money at it, see hft)
Kind of related, only slightly, this question reminds me of proposals I've seen to have self driving cars constantly auction behavior to each other. Like during a traffic jam, cars might pay each other pennies to change lanes so that one is slower and the other faster.
If we want to make a sketch using interchangeable blocks of text to be read as a sequence, should we still use something like markov chains or should those only replace words? -Aamina
markov chains can represent any 1d sequence of tokens. Tokens might be letters, words, phrases, sentences, musical notes, pixel colors, functions. A token is really just any 'thing' can be put in an order.
Could you use images in place of text if they are svg files? -Aamina
you could use images. they don't have to svg. its a little abstract so here is a bridge: imagine making a markov chain trained on a string of emojis
Can markov chain be thought of as noise? And can noise be used as a generative tool for text? -Franco
> Can markov chain be thought of as noise?
no, not really.
i'm sure you could find some common ground between markov chains and noise functions, but they are different enough that I don't think it would be productive
> And can noise be used as a generative tool for text?
you certainly could use noise when generating text
Could you share examples of generated media via text? SVGs, OBJs. - Franco
in a sense almost all SVGs and OBJs are "generated" media. Its very uncommon for someone to create one of these files manually in a text editor (but you could!).
but those are probaby better described as machine "exported" than machine "generated", usually the content is created by a person in the GUI
For examples of generated SVGs you can look at our vector week! Some of those sketches used paper.js to generate drawings and export them as svg.
For examples of 3d stuf (OBJ) look forward to the OpenSCAD week.
Neither of these are quite the process suggested in chapter (generating the SVG/OBJ directly with text generation) but you could try that this week.
I'll see if i can think of any examples.