apinstein

Here is the “internal” pattern language I started on “patterns for building software projects with AI”: https://github.com/apinstein/pattern-language-building-with-...

It’s mostly AI-written patterns based on my personal prompts/observations.

And here is the skill for organizing the pattern languages for a single project:

https://github.com/apinstein/skills/tree/main/skills/pattern...

I haven’t really battle tested these so hard yet but it’s a fun concept and really should get back to organizing them more intentionally. Happy to share this early phase just cause it’s cool to see the interest.

The shared language of a software project is not English or Python but it is the common understanding of what its concepts mean, where the boundaries are, which invariants matter, who owns what, and why the system has the shape it does. This language is rarely written down in one place. It lives partly in documentation and code, but also in code review, conversations, arguments, and the experience of having to explain a change to somebody else.

This is so true. I am a big fan of Christopher Alexander’s “Pattern Language” concept, which addresses this exact problem! In fact he recommends developing your own pattern languages for your own domains (which of course led to the famous GoF Design Patterns book).

I have been experimenting with a “Pattern Language” skill which instructs the AI to maintain 3 pattern languages for every project. One in the business domain, one in the product domain, and one in the technical domain. It is working really well. It is always super cool to see it reference the pattern languages during planning and curate them during implementation and review.

I credit using it with keeping my 100% ai-coded projects well organized, aligned across domains, and easy to work on.

Exploring highly interactive instrument (piano) practice to see if AI can help students practice better. Full duplex voice agent alongside your practice session. Also exploring live AI jamming partner to practice playing with others.

Furniture isn't part of the property by default; only if included in the sale. So pics with furniture aren't a problem, unless you say it's included and you don't get what's pictured.

Codex.app is gone and merged into ChatGPT.app. The upgrade process was... messy... Codex's self-update just deleted the Codex.app w/o further instruction. And ChatGPT updater failed the first time while also bricking the prior installed ChatGPT.app.

Seems good/fine once you get through upgrading the app.

I mean it’s already fraud to do so today… you cannot misrepresent the condition of a property. I think the harm so far has just been wasted time and thus no standing for lawsuits yet.

It’s pretty much always been against the rules of the mls to use altered photos. Pre ai, people would sometimes get busted for making grass where there was none, or making walls with cracks a solid color. It’s not allowed to alter condition of property.

Staging and virtual staging (including sky replacement and removal of trash cans) are allowed so long as they don’t alter the true nature of the salable property.

I am surprised that AI hasn’t been policed more. Probably they just don’t have the ops for it.

You mentioned it at the end of your article that you should definitely think more about horizontal transfer. It basically lets any individual in a population install a runtime genetic package that can be shared between individuals. It can work extremely fast.

In the lab, we would combine in a gene of interest with an antibiotic resistance gene in a single plasmid and introduce it into a bacterial community. Within 48 hours the only individuals left alive will have the plasmid with the gene of interest.

I am playing around with building my own similar and am faced with the question you pose.

How can you tell if your prompt process works? I feel like the outputs from SDLC process are so much more high level than could be done with evals, but I am no eval expert.

How would you benchmark this?

I built https://www.chatmycal.com/

Ever get an email or handout with a massive schedule - in text, image or pdf - and you have to hand re-enter dozens of events? I built ChatMyCal to fix this.

Copy/paste the email, or take a pic and it will perfectly extract the schedule and publish it as a subscribe-able calendar. Then “transfer” it to the group admin and save everyone else from the same thing.

On the inside, it’s basically Cursor for calendars. So you can use AI to batch edit things, decorate with coordinated icons, add rules to apply to all events, etc. It can also develop full schedules like “make a monthly book club for zombie books” or plan a weekend foodie trip to Miami. Not sure all the best uses yet!

While true, this either matters for you or it doesn’t. Classic Innovator’s Dilemma.

My EV gets only 230mi range at max, and I only charge to 85% which is like 190mi. But I do it at home and never have any range anxiety.

The trajectories for battery improvements indicate it is just a matter of time before those with larger range needs are addressed satisfactorily.

If you cannot slow charge at home or work, it’s a tough story, EV’s aren’t right for you yet, and that’s ok. Roll out of slow charging is less clear that it will be solved in a scaled way. I am not one that believes that 5-10m EV charging is a good goal, it’s very high power and likely not a good price trade off for the time saved. Current 20-30m will likely be the broad solution for those that want EV and cannot charge at home, though I think that’s not a very good solution.

It’s not a hand wave…

The cost to serve a particular level of AI drops by like 10x a year. AI has gotten good enough that next year people can continue to use the current gen AI but at that point it will be profitable. Probably 70%+ gross margin.

Right now it’s a race for market share.

But once that backs off, prices will adjust to profitability. Not unlike the Uber/Lyft wars.

Here's the play. It's very simple, and it's quite good.

Stripe processes a LOT of money. The customers that get that money need to move it around. Often to banks. Stripe makes no money on that.

Over the last few years, stablecoins have become a preferred means to hold and move money (for convenience, etc).

Stablecoin providers make money on their float -- selling stablecoins means you get free deposits, and risk-free rates are presently around 4%. For every $1M in stablecoins your customers hold, you can make $40k/year. Stablecoin providers like Circle pay about half of that back out to partners that sell the tokens.

Stripe is huge, and well-trusted by customers for handling payments. By adoption stablecoin infrastructure to control financial flows into stablecoins, they can amass huge amounts of stablecoin sales.

If even ~3% of their transaction volume gets held in Stablecoins, and they make 1% a year on that, it's about $1B a year in bottom line.

~$10e9 (daily avg vol) * 365 * 3% (converted to stablecoins) * 1% (net income) = ~$1B

Oh wow, a bunch of Alzheimer’s grants at Columbia were canceled, including the Alzheimer’s Disease Research Center. Unclear if this study was affected…

https://taggs.hhs.gov/Content/Data/HHS_Grants_Terminated.pdf

There are so many things to control for… - compare to taxi / professional drivers - locations - times

Still a great nominal achievement but anytime a sponsored research study doesn’t even attempt to control for basic things it raises a lot of red flags.

hmm. I don't see why not optimize closing costs. I have done so on 4-5 mortgages myself. When sourcing your lending options, if you are a good borrower at least, you will have multiple options. I always asked for a full closing cost estimate and compared. Usually saved $2000-5000 through a combination of:

- effectively shopping around items like title insurance, appraisals, etc by pointing out differences b/c competing vendors - identifying BS items that are not even on all offers, and simply having them removed. people like to add bogus fee lines.

For sure doing this as lead-gen is great. Agree that there is a huge risk of uploading personal info -- in the future local AI's will be able to do this. In the short term, they should partner with a known brand to give credibility.

It’s fascinating how no one mentions that Trump didn’t pass comprehensive immigration legislation during his first term despite it being core to his platform.

This issue is a mess and has been kicked down the road for literal decades at this point. Maybe finally it will get passed…

I started and ran a real estate photography platform from 2004-2018. We started r&d on this in ~2016 when consumer VR first came out. At the time we used photogrammetry and it was “dreadful” to try to capture due to mirrors, glass, etc.

So I have been following GS tech for a while. I’ve not yet seen anything (open source / papers) that quite gets there yet. I do think it will.

In my opinion, there are two useful ways GS can bring to this industry.

The first is ability to use photo capture to re-render as a high production quality video similar to what people do with Luma AI today. While this is a really cool capability, it’s also not really that hard to do anymore with drones and gimbals. So, the experience of creating the same thing via GS has to be better and easier, and it’s not clear when that will likely happen due to how painful the capture side is. You really need good real time capture feedback to make sure you have good coverage. Finding out there’s a hole once you’re off location is a deal breaker.

The second is to create VR capable experiences. I think the first real useful thing for consumers will be so you can walk around in a small three or 4 foot area and get a stereo sense of what it’s like to be there. This is an amazing consumer experience. But the practicality of scaling this depends on VR hardware and adoption, and that hasn’t yet become commonplace enough to make consumer use “adjacent possible” for broad deployment.

I could see it being used on super high end to start out.

--- Founder / Technical Co-Founder / Exec Leadership / Head of Product ---

  Location: Atlanta, GA
  Remote: Yes, but prefer hybrid/on-site.
  Willing to relocate: No, but open to occasional travel.
  Technologies: 25 years of full stack work in C/C++/Obj-C/php/js/ruby and more. Data Science in R. Embedded, MacOS, iOS, Web (1.0, 2.0, SPA). AWS/GCP. Postgres.
  Résumé/CV: https://www.linkedin.com/in/alanpinstein/
  Email: please contact through LinkedIn

The only fair way to analyze this is by looking at opportunity cost, which isn’t what TFA does.

Founders often have slightly higher market value (though not always) than first employees, so they are giving up more to go the startup route.

Separately, TFA further underestimates founder risk as they are typically not taking salary during pre-seed, and no or low salary during seed. However employees 1-5 typically get mostly cash, often much closer to market.

Thirdly, there is also often a lot more stress in being the founder. It is a complex, all day job. You have the weight of keeping things going for all employees, and when cash is low it’s your paycheck that gets delayed/cut first, not your employees.

That said I am all for reasonable early stage liquidity where it makes sense, but as many other commenters have mentioned, it tends to not be life changing super early for most early employees. Most employees would rather keep the bet on the table. Also, I am strongly against large founder secondaries. I think it’s helpful for founder to remain feeling “not financially successful”, especially first time founders, so that they keep their heart in the game. I followed this practice with my companies.

This is really great. I imagined such a thing should be created, amazing to see it in reality. It would be great for those of us not limited to exclusively voice to be able to use commands as well, as I still think in some cases doing explicitly what I want for simple things is easier than figuring out how to explain it :)

I think it would be significantly less accurate. Their error rates for performing physical tasks would be different b/w they lack the sensors to accurately train decent world models. For instance, I don't think they could catch a ball at the same skill level as a sighted child no matter how hard they tried.

So the lack of that sensor will cause the brain to develop poor representations of motion in 3d space.

How lack of those representations would affect other representations is less clear; because seeing the fusion between the LLM (which similarly doesn't have an embodied world model representation) and the robot AI (which presumable does) obviously works really well.

Now, it's possible that the 2 models are just inter-communicating between their own features (apple the concept and apple the image/object) and then being able to connect that together. The point of this meaning that there could be benefits from separate training and then post-training connection to bridge any gaps in learned representations.

However, I'd think that ultimately a model that can train simultaneously on more sensory input vs less will have a better/more efficient world model with more useful & interesting cross-connections between that space and applied uses in non-physical domains.

A reminder that if you have any accounts where that company uses “voice phrase as password”, call them and have it disabled. Usually they have other options like a secret pass phrase.

I also taught my whole family a passphrase to verify that any call “from family” is actually that family member and not a shakedown scam.

Super easy precautions against really painful consequences.

I played around a lot with VR180 when it came out. The experience is incredibly, almost uncomfortably intimate for personal videos. I felt so awkward watching demos of other peoples “blow out the birthday candles” moments. However, the fact that it was uncomfortable means that the technology itself is very good, otherwise it couldn’t produce such and emotional experience.

On the tech side, I’m just guessing, but it looks like Apple has an even better version of VR189. A 6dof version of VR180 seems entirely plausible for Apple to pull off with NeRFs and would be even more incredible.

Again, I agree that it’s a bit weird for personal memories, both on the recording side (possibly awkward to wear goggles in those situations) and even watching personal memories.

However, I’d expect Apple to make recording spatial videos possible w iPhone/iPad, which at least fixes the awkward recording issue.

Even with that possibility, I think Apple hurt themselves using this “personal memory spatial video” example.

For me, the far better use cases are for entertainment. Professional, live (and recorded) spatial video will be huge. Everyone can have front row, court side, or even birds-eye views of all forms of in-person entertainment. Sports, plays, comedy, concerts, orchestras. The experience of watching it is so intimate experientially I think it will be amazing. Looks like the tech to make it happen is finally here. Imagine them owning “the App Store” for spatial video pay-per-view…

Excited to see where this goes!

Subject Matter Expert

IIRC the 8 on a bight is best for load parallel to the rope, and the butterfly best for load perpendicular to the rope. The butterfly is really fun to tie though :)

I used the microblink one for a small toy project and it worked well in all browsers.

https://demo.microblink.com/self-hosted-api/pdf417

Are you sure?

Results: Of the 223,128 citizens of Itajaí considered for the study, a total of 159,561 subjects were included in the analysis: 113,845 (71.3%) regular ivermectin users and 45,716 (23.3%) non-users.

That reads to me like 160k people participated, and 113k optionally choose to take Ivermectin as prophylaxis.

Am I missing something ?

They went to all the trouble to create a huge trial and didn’t make it double-blind, placebo controlled? That is so suspect. We know how to create good experiments, so I just have little trust in bad experiments, especially at this scale.

I don’t care what the outcome is, treatments are treatments. I have no hope for a particular outcome. It’s just bad science. And I might argue bad faith science.

Potential biases: for instance in Brazil there is a very high level of prior Covid infections. What I’d previously infected people were more likely to opt in to treatment bc they are more afraid of reinfection, and the result came from reduction due to prior immunity? What if people that took the Medecine behaved differently than non-takers?

A few thoughts.

1. For me, being physical is important. It isn’t for everyone, but if it is for you, don’t ignore it. It brings perspective, clarity, and renewed energy.

I don’t enjoy the gym rat thing, though. For me, I’ve really loved tennis & climbing. I set myself goals in each and work up to them. I use them as toys to work on mental fitness. I’ve been surprised at how much it has helped in work & life.

2. I’ve worked with a bunch of very smart people right out of school who worry about perfection. One thing I have learned that’s helpful to point out is that the school format is great for teaching a lot of information fast, and measuring a certain type of achievement, school is not life. In fact, it’s counterproductive to work/adult life. A perspective shift is needed. Think of this as a developmental phase of life. Most people experience it.

In school, you are being trained to be a perfectionist. You get one shot, and then are judged, with finality. In this game, it’s worth it to overinvest in everything to maximize your one-shot score.

In life, you can generally control the clock, and the world offers multiple rounds. So, you can play with an idea over a long time, iterating it towards more and more usefulness / perfectness. Note that the real world is much messier than school, and often the problem itself changes over time. It’s important to let go of the idea that you’ve been trained on, that there is one perfect answer to each question.

“Perfect is the enemy of done” is a great quote here.

—

For your side projects, if I were you, I’d pick any one and just finish it. Not perfectly, just done. Enjoy the process, appreciate the result. Don’t judge yourself on the outcome as much as the process you used to get there.

Good luck!