Autogenerating a Book Series from Three Years of iMessages
benkettle.xyz
benkettle.xyz
Honestly when I read the title I thought it was going to be about using message history as a basis for generating a narrative account of the events using an LLM.
Hard copy means paper. Also microsd is a terrible for long term storage.
How come? I would have thought flash would be better than e.g. a regular hard drive.
Get a tape backup system and be happy. I rotate three tapes between offsite, my safe, and an active one in the tape drive. Don’t overthink “offsite”, your work office is good enough. Or your parents house. Or your neighbor on the other side of the neighborhood.
Or backup once and throw the tape in the back of the closet. At least you can restore the magnetic backup. I would not be happy with the error rate of transcribing data from paper :-)
Good sports taking one for the team! Thank you
[Here is the blog for those interested](http://www.honeylightsletters.com/)
Yet I suppose there’s a certain charm to that, so I hope I don’t sound like too much of a wet blanket.
Anyway, this is definitely a cool idea. Reading my chat history with friends is actually very nostalgic.
https://github.com/KnugiHK/WhatsApp-Chat-Exporter
It also links to a Telegram exporter.
Telegram has this natively already
As for extracting from the backup DB, they'll be encrypted..
See https://faq.whatsapp.com/1180414079177245/?cms_platform=andr...
The limits are supposedly due to email size limits but, as they also apply when exporting to non-email endpoints like Google Drive, I suspect they're more to do with preventing people from moving their chats to other services.
And yes, the db is encrypted.
Since some months (years?) ago, WhatsApp lets you set up your own encryption password for the DB backup. I set one up and used https://github.com/ElDavoo/wa-crypt-tools to get access to the decrypted SQLite and run some analytics over my messages :)
:-)
Printing and binding at home is probably the only option. All that's left to figure out is how to make the end result durable haptically pleasant...
I never understood why people care to keep their private conversation history in the first place. IMO private messages (as opposed to public posts, blogs, etc) are supposed to be temporary ("ephemeral") - one does not record every face-to-face conversation or phone call after all.
But I've seen that episode of Black Mirror, too. So I wrestle with the desire to perfectly remember everything that I've ever experienced vs the mental and emotional health benefits that clearly come from being able to forget things.
Are you trying to remember everything all of the time? Including all of the new memories?
Suppose you are able to record calls and face-to-face interactions. Are you going to spend hours of your life re-watching or fast-forwarding through mundane everyday things?
So, yes, I do want to remember everything all the time.
At the same time, I recognize that there are mental health benefits to forgetting things, so I will settle for being able to easily flip through all of my text messages with friends and loved ones.
It tells a story and its the zeitgeist of our generation.
People haven't thought about too much how to preserve something like this.
I personally like the idea and i can imagine exporting this with all / few messages of my mother and having a memory of that time.
Why? Privacy and permanency are orthogonal axes. You've never kept a cherished letter or re-read a thoughtful text message?
As for privacy and permanency - if data stops existing, it is definitely private now :)
One reason that's understandable without relying sentimentality, is they're a record of what you were doing or thinking at a particular time, much like a private diary.
There's been a few times where I've gone back though stuff like chat history to better understand something that I didn't realize the significance of at the time.
On the other hand, when I left the Army I moved into a much smaller place, put most of my stuff in storage, then three years later figured anything I hadn't touched or used for three years was something I didn't actually need, and let the facility have all of it. That seems to have included both all those old letters and all of my old photographs. I can't say I actually miss those things. People in here are saying they don't want to forget the past but the reality of forgetting is you don't know you forgot it so it has no perceivable effect once it happens.
To be honest, I'm nostalgic enough as is and don't think I need even more things to hold onto. I already don't watch new television or listen to new music. I'm mentally stuck in 1999 and not sure that's healthy.
The barrier, to me, is broken the moment we use technology. I work with this shit, I know how the sausage is made. I know that the phone calls are as encrypted as HTTP, that everyone can always keep record without you knowing, that even if they promise that something will go away, it may be won't, especially because it's a juicier target now just that promise alone. As soon as something is electronic, then it's a record.
However, this is just an opinion.
Thinking more carefully, I think my opinion goes more like this:
my data collected by others should be temporary, but I should be able to maintain my own data forever if I want.
We have learned a lot of history by reading famous peoples' letters long after their deaths.
Which brings me back to my point that text chats are equivalent to a spoken conversation and should be treated as such, and not be kept forever. Especially not printed out as a gift. You wouldn't give someone close a video of them sleeping or leaving for work over last 3 years (even if you saw them do it), nor a map of their movements from a GPS tracker (even if they told you where they are going), because that does not respect their boundaries nor privacy.
Although after reading some responses, no doubt some people will think of those as "cherished memories".
I was playing around with Nomic Atlas (https://docs.nomic.ai/) recently and dumped a bunch of my chat history in there, and it was pretty interesting to visualize and browse my messages as clusters around topics.
Which leads me to think that you could bring the searchability of digital to the physical format by generating embeddings for the messages and running topic modeling on them; then, you could create an index of topics at the end of the physical book with page number references to messages about that topic.
The thing I like the least though is the table of contents, it's so dry with just the months and years. Despite the skepticism I have about latest AI use and abuse, generating a one-liner from the contents of each month seems like it would be a fitting usage for it.
Also - noto emoji is great. It is also nice to use for 3d printing/laser cutting
> /.../00008120-001854410CEB401E >>> cd 3d /.../00008120-001854410CEB401E/3d >>> ls 3d0292d3fe90e1e22c247403c0e9105ea0f9ff44 3d8830b71e98aae80b6eaf8bdd5500d79ce74946 3d02fe309afa7de839822d6f1b8433aa90090d17 3d88cdc16ff2b5231e5ea4b52271ee195a6f4b96 3d072c4fca5db4a5678fa10b137435f757e98492 3d8a425d70f4049417e855d273c44d8199de30c9 3d0739c90579fa907246d5c21bd8d8ebaa2d9d6b 3d8a43a1921f504bb4393250f75b24bfc2c5cedb 3d0798b3cc4d2f5ad347ffb8bc5a0f9d8c82cfb9 3d8a7c0460aadabf1b7fc9adea9e6a2a6e7bc73b 3d07a0adc5c5c22dc525ccd3a93fb05a50ef1ac5 3d8b6ad12c7617b3d783790a457b0aa19b193b68 3d0880f091c51ddc145e17c78d8e6f9a3e7e20c8 3d8b82abe05a9d697102d8b665c9d499e07492ea 3d093e92cf03abf3650411e09a647630a1e0c478 3d8ba897240ad32580bf8dfd00db8f181658cdfd 3d095e908ff898be3b3ffd64a75db959a58ac70a 3d8bc227d67ec4944df8e75291102367034d7214 3d09d5dcd5a9bdad67a80cd83201a9e1fb75aada 3d8c722f1d92f7cd6f90c936c14f60f51aad128b 3d0abb83123be82abf43ce20118e72fea06023c5 3d8ca6eeabeb1c01fae05bb20f08dedf734cfd04 3d0b246304c42d2ab1eb1892d629fcdfde689cb7 3d8d0c6b1bf7946c6bef91d60cccb32207b7bc01 3d0bb5f49e6f0e31348ef8feb9a38d4ce71f5ec7 3d8fd2fbcaf3079a683a8e486ecde8875f0a591d 3d0c1283936c45fec533a507b78558b5aa3159fa 3d8ff93bd94b3ea14edc77d1e677cf4ee4306e4e 3d0cb8e28462780bb9af1440e297ecd8224c70ff 3d90ea8bfbf62feda080cd0ccbd12fa5c8673993 3d0ce10de5f69606c52882215b99ebab259dc194 3d932638fe8ed669725b7a143c6a8b02b8959923 3d0d7e5fb2ce288813306e4d4636395e047a3d28 3d93c92679aa9d398331e27fdeed64b5094e68d1 ...
All I could think was, "Oh no, the nam-shub of Enki!"
I want some kind of local document host that I can run a summarization or filtering script over to extract the portions that are legible to TTS, pipe it into something nice like ElevenLabs (if I was rich) or whatever, and then host a OGG for me to listen to on the go...
Isn't that backwards?
I have an Android backup version of Signal message around 2020. Of course I have decrypt key. Since I can't restore it with current version, or the 2020 version of Signal (on Github release), how can I decrypt and extract all the message? Thank you.
Not sure of the situation for the mobile backups though!
I suspect this person's project will become very popular as a service. This is a great idea.,
Just in case you missed my other comment.
Not my repo.
Also before going to the article, I thought it was about using an LLM to write a book with stories and characters inspired by the message history
A book full of years of texts would be an interesting artifact, but how often would you pick it up? How interesting could it really be? Would you even want anyone else to see it?
Now suppose each exchange comes with a haiku summary, a fresh high-level look at the conversation that condenses its vague essence into a little linguistic locket, portable and easily recalled. The interplay between the mundane raw material and the poetic take would render it much more interesting, and tend to reward repeated examination.
A human poet would undoubtedly do a better job at this than an LLM, if any human poet could be persuaded to start and complete such a project. But having an LLM do it would be a very low-cost, low-effort way to at least try for interesting results.