what does it mean to guard my attention ?
π: (?) marks uncertain readings. The final photo cuts off part of the right margin; the visible fragments are retained.
i actually do prefer this workflow.
stay completely disconnected while the efforts to connect are happening at scale in the background w/out me getting exposed to all of the poisonous noise out there.
let me think thru this now bc. it feels relevant.
workflow. → the way of operating.
firstly, deep work is super important. long continuous, uninterrupted blocks of attention is required to study / understand / seek truth. — absolutely
an obvious sort of realization. i must protect my attention. guarding it is my firstest (?) priority. but this is nuanced...
- expose myself to the right things.
- let go of the things i've been exposed to that are irrelevant, w/out getting caught up in them.
basically, filter what gets to me in the first place, & even in that, have mastery over what you choose to pay attention to, and what you let go of.
② is in an internal skill. ① — what π will be able to help me w/.
and i certainly require both.
π: The circled ① and ② refer to the two numbered points above.
oh. so it's very clear what the scope boundaries are here. but let's go a bit deeper. — yeah, i want to flesh this out!
→ i want to create and publish to platforms w/out having to open the platform & expose myself to them when i simply want to publish an idea that i have.
→ i do not want to ever know how many people have liked my post, or how many "followers" i have. i do care about bookmarks, comments, dms. — basically i want to be very directly optimizing for the game i want to play & not what the platform want me to play. → design my own game!
→ i want to be notified when:
i want to be notified when:
→ someone i want to connect with follows me (bec. then i can DM them).
→ someone comments on my post. (not a bot!).
→ someone DM's me, and it is relevant / they are someone worth connecting with..
→ every 1000 bookmarks. for a post.
→ someone i want to connect with posts something relevant to my interests, that is an opportunity for me to reply & engage w/ them about.
i want to only pay attention to the metrics i care about → which fundamentally comes down to friendships on the path to seeking truth. → that is all that matters to me!
i want to consume discover:
→ people i would want to connect with.
→ expressions i vibe deeply with.
i would like to dedicate time every day to this exploration & discovery.
who is worth connecting w/? which expressions would i vibe w/? (idk) → that's what makes this so tough. these are not represented in instructions. they are intuition / feeling / instinct — needs to be modeled w/ data, not defined by w/ rigid description / instructions.
tooling / product / my "UX".
- inbox → dms, notifs, +(discovery feed).
- browser → apply my
workpersonal lens. - notifications → system / OS level notifications across devices.
- → single notification system that is completely under my control.
π: “personal” is inserted above the crossed-out “work” before “lens”. An empty fourth bullet has no text.
fyi — essay here is "using AI to guard my attention".
there is also notion of transparency here.
it is tricky, but it can be done! i want to be as transparent as i possibly can be!
↓ configure our lenses!
Principles for using AI
1.) AI must never speak online (?) as me.
i.e., → any dm, reply, post, email, etc. that comes from me (me & my username / email / etc.) must never be drafted by AI.
my whole purpose is to cultivate friendships on the path, AI getting b/w me and people goes against my most fundamental north star.
2.) I must never capture negative signals:
(or explain (?)),
→ more for how i compile training data, but also in the conversations i have w/ AI → this is very very important.
π: The parenthetical “or explain (?)” is written above “negative signals”; its reading is uncertain.
Product notes for π (guard())
→ browser: remove all signals / stats i don't want, from x.com. → for any post, remove the 3 stats (likes etc.) — no stats must be visible for any post.
for my profile → cannot see the follower count. → becomes an important sort of: "zero" to how (?) i / filter for what information reaches me in the first place. (+ remove the notif tab entirely!)
The other surface areas are:
- notifs
- inbox
& that's it no? anything else that i'm missing here? nothing else comes to mind rn. ok, then let's flesh out the designs for this.
(& ofc the dashboard where we can [illegible (?)] science the shit out of this!).
π: “zero” is quoted and underlined. The surface-area list is in the left margin, connected by a curved line to the browser note; the deleted words remain uncertain.
INBOX
I see 3 categories here:
- things that have actually reached me: e.g. a DM, email, response to my cold outreach,
etc., a[illegible (?)]reply to my post, etc.
[π that is worth engaging w/]
π: The bracketed note sits beneath category 1; π is circled.
-
things that i should act on: e.g., someone posted something i should respond to, something changed in the world that creates an opportunity for me to express something.
-
things that are worth my attention: discovering people, ideas, expressions that i'd vibe w/.
① → is special bc. if it's reached me & it's worth engaging, the prob is v high it leads to something.
② → great, but prob very low that my reaction to this actually generates progress.
③ → not really abt. probability. just want to be exposed to great things!
from a UX, i see 2 divisions (neater).
act
discover
and have everything in a single screen.
as new info comes in, rerank the list!
↓ RANKING ALG ?
π: The sketch places categories 1 and 2 in the upper “act” division, with a line between them, and category 3 in the lower “discover” division. A brace groups all three into a single screen; an arrow leads from reranking to “RANKING ALG ?”.
How to rank my inbox ?
· obviously not chronological. but for action, there is a need for urgency. for sure! but perhaps not for discovery.
for now, keep it simple & chronological.
feedback?
no feedback mechanism in the inbox itself.
if i go onto actually reply / engage w/ the things presented — that becomes the feedback. if i ignore / archive w/out taking any action on that piece — it's treated as ignored.
really simple.
used to train the model.
notification:
→ things that require action — notify right away.
→ discovery item — do not notify.
and that's really it, no?
feels like enough information to bring this to life.
an important thing: what data are we capturing? how is it stored (where in ghost), how is that data utilized by π?
→ basically creating a continuously improving system.
& this leads to what we see in the dashboard view!
π: Arrows connect the data questions to both the continuously improving system and the dashboard view; the system note is circled.
The dashboard for π can literally just be -
2 things?
→ attention. → action / also discovery.
→ manifest.
is there a cleaner organization ?
→ expression
→ notes
→ videos
→ essays
→ products
→ vibed with
→ browsing
→ ai-chats
→ communication
→ personal-site neopuluri (?)!
π: A brace groups the list from “notes” through “communication”. “personal-site” is crossed out; the replacement reading “neopuluri (?)” is uncertain. This bottom item is circled and an arrow points from it to “products”.
vibed
π: This image is a close-up of a single handwritten word.
wh[illegible (?)]
ju[illegible (?)]
ev[illegible (?)]
top l[illegible (?)]
this
becau[illegible (?)]
is an
of [illegible (?)]
π: These are separate visible line fragments at the photograph's right edge. The rest of the margin note is outside the image; no continuation has been supplied.
Sources
1. IMG_0203 2.heic

Left page
i actually do prefer this workflow.
stay completely disconnected while the efforts to connect are happening at scale in the background w/out me getting exposed to all of the poisonous noise out there.
let me think thru this now bc. it feels relevant.
workflow. → the way of operating.
firstly, deep work is super important. long continuous, uninterrupted blocks of attention is required to study / understand / seek truth. — absolutely
an obvious sort of realization. i must protect my attention. guarding it is my firstest (?) priority. but this is nuanced...
Right page
what does it mean to guard my attention ?
- expose myself to the right things.
- let go of the things i've been exposed to that are irrelevant, w/out getting caught up in them.
basically, filter what gets to me in the first place, & even in that, have mastery over what you choose to pay attention to, and what you let go of.
② is in an internal skill. ① — what π will be able to help me w/.
and i certainly require both.
π: The circled ① and ② refer to the two numbered points above.
oh. so it's very clear what the scope boundaries are here. but let's go a bit deeper. — yeah, i want to flesh this out!
2. IMG_0208.heic

Left page
How to rank my inbox ?
· obviously not chronological. but for action, there is a need for urgency. for sure! but perhaps not for discovery.
for now, keep it simple & chronological.
feedback?
no feedback mechanism in the inbox itself.
if i go onto actually reply / engage w/ the things presented — that becomes the feedback. if i ignore / archive w/out taking any action on that piece — it's treated as ignored.
really simple.
used to train the model.
Right page
notification:
→ things that require action — notify right away.
→ discovery item — do not notify.
and that's really it, no?
feels like enough information to bring this to life.
an important thing: what data are we capturing? how is it stored (where in ghost), how is that data utilized by π?
→ basically creating a continuously improving system.
& this leads to what we see in the dashboard view!
π: Arrows connect the data questions to both the continuously improving system and the dashboard view; the system note is circled.
The dashboard for π can literally just be -
2 things?
→ attention. → action / also discovery.
→ manifest.
3. IMG_0207.heic

Left page
INBOX
I see 3 categories here:
- things that have actually reached me: e.g. a DM, email, response to my cold outreach,
etc., a[illegible (?)]reply to my post, etc.
[π that is worth engaging w/]
π: The bracketed note sits beneath category 1; π is circled.
-
things that i should act on: e.g., someone posted something i should respond to, something changed in the world that creates an opportunity for me to express something.
-
things that are worth my attention: discovering people, ideas, expressions that i'd vibe w/.
Right page
① → is special bc. if it's reached me & it's worth engaging, the prob is v high it leads to something.
② → great, but prob very low that my reaction to this actually generates progress.
③ → not really abt. probability. just want to be exposed to great things!
from a UX, i see 2 divisions (neater).
act
discover
and have everything in a single screen.
as new info comes in, rerank the list!
↓ RANKING ALG ?
π: The sketch places categories 1 and 2 in the upper “act” division, with a line between them, and category 3 in the lower “discover” division. A brace groups all three into a single screen; an arrow leads from reranking to “RANKING ALG ?”.
4. IMG_0206.heic

Left page
Principles for using AI
1.) AI must never speak online (?) as me.
i.e., → any dm, reply, post, email, etc. that comes from me (me & my username / email / etc.) must never be drafted by AI.
my whole purpose is to cultivate friendships on the path, AI getting b/w me and people goes against my most fundamental north star.
2.) I must never capture negative signals:
(or explain (?)),
→ more for how i compile training data, but also in the conversations i have w/ AI → this is very very important.
π: The parenthetical “or explain (?)” is written above “negative signals”; its reading is uncertain.
Right page
Product notes for π (guard())
→ browser: remove all signals / stats i don't want, from x.com. → for any post, remove the 3 stats (likes etc.) — no stats must be visible for any post.
for my profile → cannot see the follower count. → becomes an important sort of: "zero" to how (?) i / filter for what information reaches me in the first place. (+ remove the notif tab entirely!)
The other surface areas are:
- notifs
- inbox
& that's it no? anything else that i'm missing here? nothing else comes to mind rn. ok, then let's flesh out the designs for this.
(& ofc the dashboard where we can [illegible (?)] science the shit out of this!).
π: “zero” is quoted and underlined. The surface-area list is in the left margin, connected by a curved line to the browser note; the deleted words remain uncertain.
5. IMG_0205.heic

Left page
i want to consume discover:
→ people i would want to connect with.
→ expressions i vibe deeply with.
i would like to dedicate time every day to this exploration & discovery.
who is worth connecting w/? which expressions would i vibe w/? (idk) → that's what makes this so tough. these are not represented in instructions. they are intuition / feeling / instinct — needs to be modeled w/ data, not defined by w/ rigid description / instructions.
Right page
tooling / product / my "UX".
- inbox → dms, notifs, +(discovery feed).
- browser → apply my
workpersonal lens. - notifications → system / OS level notifications across devices.
- → single notification system that is completely under my control.
π: “personal” is inserted above the crossed-out “work” before “lens”. An empty fourth bullet has no text.
fyi — essay here is "using AI to guard my attention".
there is also notion of transparency here.
it is tricky, but it can be done! i want to be as transparent as i possibly can be!
↓ configure our lenses!
6. IMG_0211.jpg

Close-up
vibed
π: This image is a close-up of a single handwritten word.
7. IMG_0204.heic

Left page
→ i want to create and publish to platforms w/out having to open the platform & expose myself to them when i simply want to publish an idea that i have.
→ i do not want to ever know how many people have liked my post, or how many "followers" i have. i do care about bookmarks, comments, dms. — basically i want to be very directly optimizing for the game i want to play & not what the platform want me to play. → design my own game!
→ i want to be notified when:
Right page
i want to be notified when:
→ someone i want to connect with follows me (bec. then i can DM them).
→ someone comments on my post. (not a bot!).
→ someone DM's me, and it is relevant / they are someone worth connecting with..
→ every 1000 bookmarks. for a post.
→ someone i want to connect with posts something relevant to my interests, that is an opportunity for me to reply & engage w/ them about.
i want to only pay attention to the metrics i care about → which fundamentally comes down to friendships on the path to seeking truth. → that is all that matters to me!
8. IMG_0209.heic

Main list
is there a cleaner organization ?
→ expression
→ notes
→ videos
→ essays
→ products
→ vibed with
→ browsing
→ ai-chats
→ communication
→ personal-site neopuluri (?)!
π: A brace groups the list from “notes” through “communication”. “personal-site” is crossed out; the replacement reading “neopuluri (?)” is uncertain. This bottom item is circled and an arrow points from it to “products”.
Cropped right margin
wh[illegible (?)]
ju[illegible (?)]
ev[illegible (?)]
top l[illegible (?)]
this
becau[illegible (?)]
is an
of [illegible (?)]
π: These are separate visible line fragments at the photograph's right edge. The rest of the margin note is outside the image; no continuation has been supplied.