Would you like some magic? I was coaching someone the other day and I said something that made him laugh and say

"That's helpful, in a very Robbie way."

"What do you mean?"

"Well, it was helpful, but not in the way I wanted it to be."

When I was learning NLP, I'd occasionally drive up to Marin to the office of my NLP teacher to get a one on one session. I'd come with some problem, some issue in my life, some unwanted experience, or thwarted desire, and the hope that he could help me with it.

So we'd do the session, and afterwards I'd consistently have the experience "that was really helpful, but we didn't fix the issue."

And then, a little bit later, if I remembered to check, I'd notice that the issue just wasn't an issue anymore. As in, it was no longer something that was capturing my attention, causing me distress. No longer something that I would pay someone for help with.

So I reflected on this, trying to understand how it worked.

And I realized: in addition to the issue, I was bringing something else as well: An idea about how the issue would be fixed. I didn't notice this at the time, it all felt like part of the package. But I not only had an unwanted experience, or thwarted desire, but also some pretty specific idea of what "fixing the issue" would look and feel like.

And if our idea about how it would be fixed was good, then we would have fixed the issue ourselves! In fact, this happens constantly in your life. There's some desire or some unwanted experience, and you have an idea about how to address it, and then you address it and it works. 99.99% of the time, this is how life works. "I'm cold. I'll put on a sweater," "I'd like some ice cream. I'll eat some ice cream."

But the ones that stick around, the desires that seem perpetually out of reach, the unwanted experiences that haunt us, those are the ones where our ideas about how to fix it are actually impediments.

In my own work as a client, and then as a coach, it's become clear to me that the role of the coach is very often to help the client get *unfixed*; to release the fixed idea of how the issue should be resolved, and become free to see, feel, and think about it fluidly, from many different angles. From there, the collective intelligence of your life can find a new way forward.

Do you have some issue like this, that's stuck around despite your great efforts and intentions? Something that you'd like to get unfixed about? I'm currently offering free coaching sessions, where we can do just that.

I've opened up a few spots for you fine people on the frontier of sustainable and sane algorithmic media use, where I'll waive my fee. This is not a sales conversation, it's 90 minutes of us doing real magic together on an important part of your life.

If that sounds fun, let me know! 

I don't think we have DMs here yet, so you can just tell me your interested and a way to contact you here:

https://forms.gle/1bCSAbFKSwKST58v7

#UpTrustFeedback grab bag

1. I can't write two posts at once.  It shows a current draft if I try to write a new draft in a new tab.  But also it says "can't post, no content", because somehow it hasn't sync'd properly.  I'm currently getting around this by just posting the first one knowing I can edit it later.
2. You just lost some text that I had drafted as a comment [EDIT: maybe it's still there but I can't find it].  This is a result of two distinct failures in how your app is set up:
   1. You didn't use `onbeforeunload` to keep the page from changing when there was drafted text that hadn't been used. "Are you sure you want to..."
   2. I wasn't intending to change pages! I held the Cmd key while clicking on Home, which, if you use links in an ordinary way, would have caused that click to open the link in a new tab. But you're doing some fancy extra javascript, and so it didn't, and instead it just navigated to Home in the same tab and lost my comment draft.
3. I just now tried to use cmd+[ to de-indent the list item, and this navigated to a completely different screen, which would have lost the text again if you guys weren't pretty good at saving drafts.  Please don't violate standard keyboard shortcut expectations by doing something that looks like it completely loses all of my work!
4. If I put a title, it shows up as small bold text at the top of the post.  But if I put a header in the post, I can make the post have a title, visually.  But the "title" is not it.  This is confusing and weird.
5. In general, there are too many options.  Like formatting is kinda nice but it evokes the idea that we're gonna be all fancy.  As an extreme example, why on earth am I choosing whether a link should open in a new tab?  Just have some sane default and don't present me with so many choices.  (Good sane default is "new tab for off domain, same tab for same domain")
6. The lil icon buttons below the post are too tiny.  Also it's confusing to have the reply count distinct from the button to make a reply. I keep wanting to push the reply count button instead.
7. Why are you summarizing all emoji reactions as eg ♡ 4 ? That's not a very good way to summarize 🤔🤷😤😒.  Pick a non-heart more neutral icon, or show the actual reacji! (I'd recommend the latter, as it's much richer).
8. Don't call what we write or read here "content" 🤮  The word has much of its history in advertising and marketing—it's the thing that's not the ads, around which ads can be placed.  That starts us off with a framing that what we write doesn't matter, it's just stuff that fills the space.  Instead be willing to talk about things as "perspectives" or "takes" or if you don't want to be so bold, just "posts".
9. If you can't edit a post after people have reacted to it (which is fairly sensible!) then it should tell you that when you hit the EDIT button, not the SAVE button (after making said edits)
10. Lots of random other lil things
    1. sometimes the title of the page is stuck on a different post name than the one I'm on
    2. inconsistent use of cursor:pointer when hovering over buttons (notable, it's absent from the profile icon in the top right)
11. no way to follow specific people? maybe this is intentional? you want to magic-algo things?  but it's sad—I see someone say something interesting and I want to like...  not lose track of them. and maybe the way to do it is to updoot more of their bops, but it feels somehow unsatisfying, and annoying if I'm like "cool I'm done gonna sign off for now"
12. if I hit the feed again, either show me something I haven't seen before or say "that's all for today", but don't just show me the same few posts I saw before

Retrieval-Augmented Generation: embedding tech that's standardizing. https://www.anyscale.com/blog/a-comprehensive-guide-for-building-rag-based-llm-applications-part-1

Here’s a post from a year ago by a company called AnyScale on how they built out their own custom LLM-system stack tailored to their dataset. It’s long but straightforward–it shares the code and analyses they did at each step (like 50+ steps), but what’s happening is pretty understandable to non-coders, I think. I guess reading through parts of this could give more UpTrusters more shared vocab and common knowledge of what kinds of methods are good toward what ends, and some ways to reason about them.

Some good snippets/concepts:

Vector DB

A database augmented with the ability to quickly/immediately return the records nearest in an embedding space.

Chunking data

Splitting documents into smaller parts and embedding them separately, so that you can find and use the relevant parts.

re: Fine Tuning

This can especially be impactful if we have a lot of: 1) new tokens that the default tokenization process creates subtokens out of that lose the significance of the token; 2) existing tokens that have contextually different meanings in our use case

(that particular aspect isn’t relevant for us probably, until we have lots of content that’s in rare languages or something.)

i'm thinking about getting back into climbing... what brands of shoes should i consider?

what else do i need to successfully climb?

what other questions should i be asking?