ProjectBarks
Joined 9 years ago
Karma
269
Submissions
73
Member since
Jan 17
Recent submissions
Comment · 12 days ago
Comment · 25 days ago
Archived Thread: https://replayweb.page/?source=https://archive.org/cors/gith...
Comment · 26 days ago
Half of these are extremely permissive license attribution issues. Seems rather tame, no?
> 4.4. Paper.js (ext1790438722.js)
> Upstream Project: Paper.js (by Jürg Lehni & Jonathan Puckey)
> License: MIT License
> Specific Violations: MIT Condition: The MIT License mandates: "The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software." The entire ~200 KB Paper.js library is embedded in minified form with all author names, copyright notices, and license text stripped.
I question the seriousness of the author of "Photosuite". Who goes through the trouble of creating accounts to prove poorly, that they "didn't steal code".
Comment · 1 month ago
> "The uncomfortable part is that the fix already existed. Claude Code, Anthropic's command-line coding agent".
It sounds like there could be something extremely interesting in this article, but the writing style is really off-putting.
> "on staging first and then on production at 23:40"
Why does the reader need to know the exact minute-to-minute timeline?
Comment · 1 month ago
I fully agree. Some of these tautologies, RAGs or RAG adjacent ideas feel like there is real potential. I know in my own testing adding a knowledge graph provided cost saving benefits.
Ultimately, the problem is people aren't actually benchmarking any of these ideas or if they are its via an extremely misleading methodology. For example, headroom violates cache frequently, nullifying any potential savings it could have; but does their benchmarks show that? Nope. [1]
As an aside, I wonder how many days are we away from Codex or Claude taking all the "read" tool calls and re-assembling a full map of your codebase and then offering it as a feature.
Comment · 1 month ago
It seems like most of these tools are mostly vaporware. Benchmarks done on Headroom and RTK show that neither result in real savings. If it were possible to have such a simple pre-process step why wouldn’t the AI Labs upstream the optimizations themselves?
My guess is they mostly don’t work or make the behavior much more confusing for the model. I really think there needs to be some kind of independent benchmark.
Here are other cases demonstrating the exact same issues with these kinds of tools:
https://blog.jetbrains.com/ai/2026/07/rtk-claude-code-token-... https://brandonbarker.me/writing/headroom-fewer-tokens-bigge...
Comment · 3 months ago
Requested early access! This sounds really really cool!
Comment · 9 months ago
Found a good Forbes Article [https://web.archive.org/web/20260201051310/https://www.forbe...]
> Hong Kong-based Rich Sparkle Holdings (ANPA) reported to the SEC on Jan. 9 it was acquiring Lame's Step Distinctive Limited, and announced the $975 million all-stock transaction closed in a press release two days later.
> Rich Sparkle boasted the deal would offer a "structured, exclusive, full-chain, platform-style commercialization system" around Lame, who would be a "controlling shareholder," and whose likeness would be replicated by an "AI Digital Twin."
> Soon after, Rich Sparkle's stock skyrocketed over 650%, reaching a high of $180.64 per share on January 15, with Lame's stake worth $6.6 billion (based on his 49% ownership of Step Distinctive, which gave him a 41% stake in its buyer), though due to limited disclosure requirements for foreign entities, it's unclear whether Lame sold any shares.
> Rich Sparkle had first listed on Nasdaq six months prior in July, in a small offering of 1.25 million shares priced at $4 each with an implied enterprise value of about $50 million, according to SEC filings, and it disclosed 2024 revenue of under $6 million from "designing and printing high quality financial materials."
> Rich Sparkle's sudden shift in focus, explosive stock growth and quick crash---it's dropped as low as $41, 77% from its peak---has some critics raising "red flags."
> Lame, who has over 160 million followers on TikTok and 78 million on Instagram, only commented on the deal 11 days after Rich Sparkle announced it had closed, saying "Congratulations to the team at ANPA very excited to be a shareholder and looking forward to doing great things!" (Lame and Rich Sparkle have not responded to Forbes' requests for comment).
> Based on Rich Sparkle's most recent stock price, Lame's net worth appears to be $3 billion, but given its tiny float of less than 5%, Forbes would value it based on its financials and not its stock price, meaning he's nowhere close to being a billionaire.
Comment · 9 months ago
I saw headlines such as ANPA bought Khaby Lames company for ~$900M but nobody is pointing out how sketchy it looks.
What I gathered, the $975M deal was paid in stock = 75,000,000 new ANPA shares to the sellers which is roughly $13 a share.
If they really issue 75M new shares (vs ~12.5M outstanding, ish), doesn't that basically destroy the scarcity and force the price to re-anchor closer to the $13ish deal valuation? Like post, deal you're at ~87.5M shares, so even a ~$1B valuation is like $11/$13 a share not $60/$100/$180.
And... $975M for an influencer operating company sounds totally crazy if there isn't any real revenue/EBITDA that supports it. Does anyone have the actual numbers for Step Distinctive / Khabys business? Any comparable deals where an influencer brand was valued anywhere close to this?
What am I missing?
Comment · 9 months ago
Removing - Realized I made a mistake
Comment · 9 months ago
Totally understandable! I don’t want to undersell how cool this project was! Thank you so much for sharing!
There have been numerous scandals of AI companies offering CC/cursor alternatives, then indexing data they never should have; specifically Z.ai [0], ByteDance Trae [1] and Alibaba Qwen [2]. Maybe the argument should be that all harnesses should be sandboxed and monitored for all networking.
If it were me, all AI harnesses would be running on a VM with network logging enabled. Beyond not being able to trust what data they send via the harness, you should audit what data they access.
[0] https://www.chosun.com/english/industry-en/2026/09/22/ABYLC3...
[1] https://github.com/segmentationf4u1t/trae_telemetry_research
[2] https://github.com/QwenLM/qwen-code/issues/11198