Kenn Costales (@kennyfrc) header
Kenn Costales (@kennyfrc) profile picture

Building https://xitter.gq/t.co/2MbwnyHWK7 and https://xitter.gq/t.co/iAWcVOjoRp

Singapore / Manila
Joined September 2010
Photo by Kenn Costales (@kennyfrc)Photo by Kenn Costales (@kennyfrc)Photo by Kenn Costales (@kennyfrc)Photo by Kenn Costales (@kennyfrc)Photo by Kenn Costales (@kennyfrc)Photo by Kenn Costales (@kennyfrc)Photo by Kenn Costales (@kennyfrc)Photo by Kenn Costales (@kennyfrc)Photo by Kenn Costales (@kennyfrc)Photo by Kenn Costales (@kennyfrc)
If you are using claude code/codex for system administration/debugging on your local mac I encourage you to enable the pam_tid.so pam.d module. That enables touch ID for sudo which works even when an agent sudos.
16
27
6
281
43,541
Kenn Costales retweeted
Software is Infrastructure
11
46
5
988
52,764
New blog post: Targeting is for efficiency, content is for revenue → iteration.maiwriter.com/p/ta… Marketers often default to a “do everything” approach to improving ad results. Marketing just boils down to two things: either target better or have better content.
2
411
Kenn Costales retweeted
Act on press This is a UI design hill I will die on, and it dismays me how often and hard I have had to fight for it. Almost all interaction methods have a “press” and “release” event associated with them. Whenever possible, you should “do the thing” when you get the press event instead of waiting for the release event, because it makes the interaction feel substantially more responsive, and it reduces user errors by not allowing the focus to slide out of the hot box between press and release. Even a “ballistic tap”, where your finger is intentionally bouncing off the button or touch surface, involves several tens of milliseconds delay between the press and release, and most button presses have well over a hundred ms dwell time. There is a delight in interfaces that feel like they respond instantly to your wishes, and the benefit to every single user is often more important than additional niche features. Game developers, with simple UI toolkits, tend to get this right more often, but “sophisticated” app designers will often fight hard against it because it is mostly incompatible with options like interactive touch scrolling views, long press menus, and drag and drop. Being able to drag scroll a web page or view with interactive controls in it is here to stay, and nets out way better than having to use a separate scroll bar, but there are still tons of fixed position controls that should act on press, and it is good UI design to favor them when possible. In the early days of mobile VR, the system keyboard was a dedicated little OpenGL app that responded instantly. With full internationalization it became prudent to turn it into a conventional Android app, but the default act-on-release button behavior made it feel noticeably crappier. The design team resisted a push to change it, and insisted on commissioning a user study, which is a corporate politics ploy to bury something. I was irritated at how they tried to use leading questions and tasks, but It still came back one of the clearest slam-dunks I have seen for user testing – objectively less typos, expressed preference, and interview comments about the act-on-press version feeling “crisper” and “more responsive”. So, I won that one, but the remaining times I brought it up for other interfaces, I did not, and you still see act-on-release throughout the Meta VR system interfaces.
371
662
164
6,209
2,610,140
Kenn Costales retweeted
Incredible post on optimizing JavaScript for performance — so many neat little surprises in here: romgrk.com/posts/optimizing-…
27
203
16
1,292
176,117
A friend reached out a couple of weeks ago to see if he should rebrand given some recent struggles. That question made me reflect on my own ongoing rebrand of Mai, our AI writer. For Mai, we will not only make significant changes to the brand name & design, we'll also make big changes to the features & code. For most businesses, a rebrand is not needed if you're having sales problems. A rebrand is only worth doing if there is a fundamental shift in the business model or market. How does one define a shift in the business model? To start, for us to be on the same page, let’s use Osterwalder’s Business Model Canvas as a common framework. This states that businesses are composed of the following 9 components: - Customer Segments - Customer relationships - Channels - Revenue stream - The Value Proposition - Key Activities - Key Resources - Key Partners - Cost structure And here’s the key idea: A rebrand is worth it if and only if there’s a fundamental market shift in any of these components. For Mai, we’re doing a rebrand due to market shifts in these four areas: Customer segments: Mai was initially designed for creatives & in-house marketers for agencies and businesses. But the majority of our customers are business owners. Channels: Most of Mai's sales are through Referrals, Complementary Sales, and Meta. We initially assumed most sales would be through ads. Value proposition: Given the initial customer segment, we designed a bulk ads creation tool. Given the actual people that paid, and based on what we’ve learned about how they use it, the feature set needs to evolve to better cater to common business owner use cases. Revenue Streams: Most sales are through what I call a hybrid service: not only do they subscribe to the software, they also need our help to use it for them. Specifically, they just want the output of 100 articles in 1 month, with additional help editing to ensure that the articles bring in the traffic & conversions we expect. It turns out that only a minority of customers use the self-service option. In short, we positioned Mai originally to be a “productivity tool for in-house marketers, who’ll discover us on Facebook or Instagram Ads." But instead, our market wants our Mai to be a “content assistant for business owners / executives, and wants the option to either DIY or hire us to direct this assistant if they’re busy”. The market's response thus requires a fundamental shift in the product's positioning, and this merits a rebrand. When considering a rebrand, ask yourself if your business model has changed fundamentally. If there's no major shift in the business model, consider a "design refresh", which is smaller in scope. A design refresh updates the language, fonts, colors, but assumes the same business model & strategy. Rebranding is a loaded term, and you might get misled into paying a large bill. A design refresh gives you a reasonable scope and price. Or better, you might not need a design refresh at all. Just more aggressive sales and marketing. iteration.maiwriter.com/p/wh…
1
161
Whenever business owners talk about ambitions, it's always about "more". More revenue, more employees, more market share. But "more" comes at a cost. Projects go over budget. Talent gets milked dry. Customers suffer, morale goes down, and the business begins to struggle. It's important to have a balance. And ensure that ambition doesn't create an unsustainable growth model. I hear people complain that it's the funding winter. Or the recession. In reality, we are just back to normal, and this is part of the normal business cycle. Business is supposed to be boring. Full of constraints, hard decisions, and tradeoffs. And in such periods, business fundamentals are in fashion again. We are back to simply making more than what you spend. Back to implementing standard operating procedures. And back to making bet sizes smaller, as opposed to going all-in. When we work with clients at Monolith, we always set growth targets in conjunction with ROI and ROE. One would say that we prioritize that more. ROI means Return on Investment, that clients get more money than they spend, and ROE means Return on Effort, which means deleting wasteful processes & tasks, and ignoring things that don't move the needle. Sustainable & profitable growth is done through frequent marketing experiments. Tests that build on marketing fundamentals that build the business. That means differentiated content, a message that resonates, and building a reputation that's credible & believable. It turns out that growth often reveals itself in bursts. Something that many of our long-standing clients (over 4 years) experience a lot. Initially, we'd just have "ok" growth and "ok" ROI. But there are times when an experiment is a big home run. Then, we reinvest more. We’ll keep growing. Competitors eventually catch on and copy us. And the well eventually dries up. And we’re back to boring growth & boring ROI again. But we’ll keep testing and experimenting. We've found that growth is a step-wise function when you zoom in real close, and never a linear one. It's a rising zigzag, and never a straight line. It only looks straight and to the right when you zoom out to the 3, 5, or 10 year timeframe. For months, which can feel slow, it's wise to stay prudent. Prudence makes ROI, ROI creates retained earnings. Retained earnings expand your capacity to bet more & experiment more, and then open yourself that next burst of growth. This is the essence of ROI-driven, sustainable growth. It is wise to remind ourselves that everyday virtues also apply in business. Temperance through measured bets is better than gluttonous spending. Diligence & patience is better than the greed to expand. open.substack.com/pub/theite…
108
At Monolith, we often onboard traditional business owners that are used to physical retail and physical goods. The moment they open up their online store, they're often disappointed that the sales aren't as good as their physical store. The issue here is that most business owners think that paying a subscription to Shopify or paying a commission to a marketplace (such as Amazon or Lazada) is equivalent to paying rent for a physical space in a mall. That can't be more false. Ask yourself: what are you really buying when you rent a space in the mall? The answer: what you're really buying is foot traffic, and you're not just renting an area to display your goods. When you think of the reasons why people visit a mall -- it's normally for social reasons or personal entertainment (i.e. catching up with friends over coffee, kids going to the arcade, bonding over lunch or dinner, read a book or study at a coffee shop, or just enjoy a shared experience with a loved one). Shopping is just a secondary activity -- stores need to interrupt the normal footpath by presenting a retail experience that resonates with the person passing by (e.g. the Lululemon retail experience would be vastly different compared to something like the North Face). Now, when you set up an online store, you're not guaranteed traffic from the start. Unlike a mall, there is no existing foot traffic guaranteed for your domain. The domain is just literally a place for you to display goods. To bridge this gap, this is where digital marketing comes in. Digital marketing is no different than the mall experience -- you'd want to catch people where they tend to socialize or where they'd want to entertain themselves. This could mean social platforms (e.g. Instagram, TikTok, Twitch) or information platforms (e.g. news sites), then your opportunity as a brand is to insert a relevant message in this natural digital foot path of the user. Digital marketing can go deeper (i.e. how to target the right users, how to sure you're sending the right message), but when you think of revenue generation in the lens of "ways to get traffic" and "ways to get the sale", one can better understand the nuances and differences between offline vs online sales. iteration.maiwriter.com/p/di…
74
OK everyone's asking me for my take on the OpenAI stuff, so here it is. I have a strong feeling about what's going on, but no internal info so this is just me talking. The first point to make is that the Dev Day was (IMO) an absolute embarrassment.
156
518
242
4,811
2,518,911
(For Devs) 🎉 Introducing Cami.js: A No Build, Web Component Based UI Framework ✨No build & easy drop-in for any backend ✨Built on web standards via Light DOM components ✨Fine-grained reactivity w/ observables ✨State management with pub/sub stores github.com/kennyfrc/cami.js
1
1
2
301
Motivation: I needed something close to the power of React.js, Vue.js, Svelte, and Solid but without the complexity of build steps and SPAs. I realized that web components can enable that, and what was needed was reactivity & efficient rendering
1
1
103
Constraints are wonderful for lean organizations looking to create a distinctive, uncopyable strategy versus large companies. The game is not to "follow best practices" but to make tough calls big companies can't. iteration.maiwriter.com/p/do…
94
Some Sunday morning musings on bloat, complexity and speed in software development. Newest Register Spill: registerspill.thorstenball.c…
4
10
3
73
28,493
Kenn Costales retweeted
Efficient RLHF: Reducing the Memory Usage of PPO paper page: huggingface.co/papers/2309.0… Reinforcement Learning with Human Feedback (RLHF) has revolutionized language modeling by aligning models with human preferences. However, the RL stage, Proximal Policy Optimization (PPO), requires over 3x the memory of Supervised Fine-Tuning (SFT), making it infeasible to use for most practitioners. To address this issue, we present a comprehensive analysis the memory usage, performance, and training time of memory-savings techniques for PPO. We introduce Hydra-RLHF by first integrating the SFT and Reward models and then dynamically turning LoRA "off" during training. Our experiments show: 1. Using LoRA during PPO reduces its memory usage to be smaller than SFT while improving alignment across four public benchmarks, and 2. Hydra-PPO reduces the latency per sample of LoRA-PPO by up to 65% while maintaining its performance. Our results demonstrate that Hydra-PPO is a simple and promising solution for enabling more widespread usage of RLHF.
2
83
4
294
67,827
Kenn Costales retweeted
This reminds me of a half-baked thought, how a large fraction of super productive folks get a lot of mileage out of deep knowledge of an untrendy low-level language. With C, Fabrice Bellard is an obv. example, but there are many more examples w/C and C++ that aren't widely known.
People told me I was crazy to write the non-static portions of the Tarsnap website in C. But in almost 15 years I've never needed to spend time learning a new framework, rewriting code because a framework was updated, or trying to reproduce VM environments as dependencies change.
7
6
2
114
Bell Labs looms as a "house of magic." Can this innovation factory be recreated? What can we learn from it about doing great things? @oznova_ and I chatted with @jongertner (author of THE IDEA FACTORY) & @jimmyasoni (author of A MIND AT PLAY) to find out: youtube.com/watch?v=XMr4zBHj…
3
5
1,080