A wonderful post. And this section is the key to the main issue with the modern Democratic Party:
“Instead of asking why a lifelong Democrat reached this point, some leaders rushed to condemn me, question my motives, and push me out.
That response reveals something deeply troubling about what is missing in today’s Democratic Party.”
29
I've had swarms of agents running nonstop for weeks, and OpenAI is so much more forgiving on usage than Anthropic. Really want to keep cross-model use cases alive but I'd much rather buy multiple Codex Max than having a Claude max at this point. Claude slowly becoming unusable.
1
2
84
There’s only one way to find out what your tolerance level for this type of work is—try it. I would say an easier proxy is how comfortable one is with risk. If you’re risk averse, you’re not going to like entrepreneurship. High risk, high reward. Pick something you care about and see it through.
Bro to bro- do not try entrepreneurship if you don't have a high tolerance for uncertainty. For long periods of time, you genuinely won’t know if you’re early, wrong, unlucky, or just stupid. You need to be cheap before you can afford not to be. Because raising money and burning it is easy. Building something people are willing to pay for is hard. You need high self esteem and a low ego and you've to be comfortable being wrong, looking stupid and changing your mind. Most importantly, learn how to lose without becoming a loser. Because you’re going to lose a lot- money, people, deals & time. Just don’t lose your ability to think clearly. And btw, don’t go looking for motivation. It is useless after a while. What you really need is stamina. Both physical & mental.
4
148
Kris retweeted
Btw this is why computer science matters. Yes, you can self teach it if you really want to (I personally wouldn't have had the discipline, so college was a way better force). However you learn it, the theoretical fundamentals matter and get applied to work every single day.
If you're interested in "how" or "why": the major culprit is that each row in Alacritty has 32 bytes of metadata and each cell is 24 bytes. In Ghostty, every row and cell is represented by exactly 8 byte each. How do we do this?
The first major culprit is styles. Alacritty stores the full cell style alongside each cell (foreground, background, underline, etc.). Ghostty stores a 16-bit style ID and de-dupes all styles into a look-aside custom reference-counted hash table.
MOST cells are unstyled, and when there are styles MOST styles are shared, and when styles are shared MOST are repeated in a run (multiple cells with the same style in a row). Put this all together, and the tradeoff on compute to access it doesn't even end up being slower.
Next, codepoints. Alacritty stores multi-codepoint graphemes (like, Emoji) by having an 8-byte nullable pointer to a `Vec<char>`. This hurts doubly: (1) its almost always null (because multi-codepoint is rare) yet you pay an 8 byte cost on every cell and (2) every multi-codepoint grapheme triggers a heap allocation to make that Vec.
Ghostty stores single codepoints inline, but multiple codepoints in a look-aside table. The memory for this table uses a custom bitmap-tracked chunk-allocator (since grapheme frequency follows a measurable curve we calculated by scanning various online texts). The presence of graphemes is marked by a 2-bit content tag in our packed 64-bit cell. To keep the key small in the hash table, its limited to a 16-bit unsigned int that is an offset from a base pointer.
Okay, the astute systems programmer will quickly notice there are a lot of 16-bit integers and ask: so this is all limited to a max of ~65K values?
Nay. We maintain our grid using a linked list of contiguous ~400KB memory chunks (which themselves are in a memory pool using a custom allocator to speed up alloc/free). Each memory chunk is limited to 2^16. If/when we reach a limit, we move to the next page. In practice, this really doesn't happen except under pathological cases... the important point is we handle it.
Lots, lots, lots more details, but thats a 10,000 foot view.
These things alone account for ~95% of the difference of our uncompressed vs. Alacritty's uncompressed memory usage. (Theres also a reason why Alacritty's data structures aren't trivially compressable but thats a whole other topic)
Kris retweeted
This is what entrepreneurialism feels like. teapawt.net/tannokasa9/status/2079…
The incentive structure for congress is mis-aligned such that both parties jockey for power instead of actually fixing anything. If the executive branch didn’t have overweight power, nothing would ever happen.
REPUBLICANS HAVE NO INTENTION OF PASSING THE SAVE ACT.
Voter ID polls well, so GOP is using Democrat intransigence to score political points. That’s it.
This week we vote on a CR to fund government thru DECEMBER which means the plan to attach SAVE to an approps bill is a bust.
2
75













