gazzini

Do they just stop billing you for `await` time & keep the lambdas warm?

This looks awesome!

So, rather than proxying requests from some server, I can use their dynamic ip product to just auto-update the DNS entry for my home computers whenever my ISP changes my DNS.

I imagine there will be some small downtime every time my ISP switches (cached DNS / TTL etc...), but it should happen rarely ().

I'll also search for "dynamic DNS" and see if anything else comes up, but this looks like a really good solution, thanks for posting it.

This is a lot like the programmer tendency to want to re-write instead of refactor because to understand a system is way harder than to set out on a quest to build a new one.

And what if they did want to refactor the existing system. Can they? What does that iteration process look like? Is the system currently evolving to fit everyone’s needs?

Remember, RMS actually did want to refactor the printer. He just wasn’t allowed to. [0]

Contrast that with how fast things are moving in DeFi. It’s permissionless innovation at its best and worst. We can barely even wrap our heads around OHM, but nobody needs permission to fork it into SPELL. Systemic refactors happen faster. [1]

When somebody in the future doesn’t like a piece of decentralized infrastructure, they can _literally_ fork all of it and just make their change. With fewer black-boxes, we share more intellectual property.

[0] https://www.fsf.org/blogs/community/201cthe-printer-story201...

[1] https://thedefiant.io/olympusdao-forks/

I think the Helium project is a fascinating experiment, and I’ve had a lot of fun participating in it. I’ve been tracking my experience for a while (with pictures): [0]

It’s certainly a weird project, and watching the corporation contort itself into something coordinated enough to drive partnerships like this but decentralized enough to qualify as a cryptocurrency is a sight to behold.

The governance of this project isn’t “decentralized” in any strict sense of the word, but the separation of the dewi alliance from the main corporate entity is interesting.

The self-imposed technical constraints are equally fun to observe. The scaling issues (gossiping blockchain data across 100k+ raspberry pi devices with no broker) are insane, and the incentive mechanism for “validators” is an interesting incremental solution. Specifically, people ”stake” $200k on cheap cloud servers, which might earn money for a few hrs a week when elected to a “consensus group.” This made sense historically (on distributed raspberry pi devices) but is now really strange compared to most other blockchains where validators earn consistently.

Every one of these issues is a bottomless pit of learning opportunities as worlds collide (economics, telecom incentives, cryptocurrency price swings, hobbyist antennas, etc…).

I’m no maximalist (I’ve sold lots of coins), but there is something deeply fun about watching _anybody_ try to do something novel to compete against the old guard. That’s why I was up at 6am today to fiddle with an antenna on a friend’s roof.

0: https://gazzini.com/projects/posts/helium/

I think it’s a pretty great idea for people to “have kids, raise them right.”

I’m bullish on that strategy, even if there is no contractual guarantee that those kids will return the favor.

Credit: I first saw this on Twitter [0].

It seems like a great way to make some of the industry-standard info more approachable to newcomers. I spend a lot of time explaining to people how TC works at tech companies, and IMO employees win when that info is more palatable.

I guess it’s still hard to do an apples-to-apples comparison unless all your job offers look like this, though.

[0] https://twitter.com/annarchyy/status/1390866663373885441?s=2...

I quit my startup & picked up a boring, stable job to be a better dad a couple years ago, so I feel this too.

If the article said it’s only hard for moms, then I’d disagree with it... but it doesn’t.

It’s just critiquing the assumption that every mom wants the same thing. These assumptions form social pressures & impact career decisions, and she’s pointing out that you don’t have to struggle with that decision in the way society expects.

Sure, it would have been nice if they acknowledged dads... but given the name of the website, I don’t think we’re the stars of this show, and that’s fine.

This is the first time I’ve ever heard anyone suggest this was a coincidence.

You might actually be correct — I personally doubt it, but it’s a funny thought... to think that we all just over-indexed on this trivial detail.

Our device has 4G, Wifi and LoRa for comms...

How many of the sites are in range of a LoRa network?

I rarely hear of LoRa the US, although it seems like a great fit for many IoT use-cases.

I got a Computer Engineering degree from Mississippi State University, and I was able to take a 1-hour class on writing iOS apps.

However, it was taught by the business school, and I think that was very appropriate.

I loved reading serverless-stack a couple of years ago; it was really helpful & convinced me to use serverless for a side-project that’s still going (with almost no expenses!).

I’m surprised to hear how many separate lambda functions each service in your example had. I understand the need to deploy each service independently... but to have +10 deployments within each service seems crazy to me. Is there a reason each service needs so many lambdas (vs deploying the service code as a single lambda function with different branches)?

Fwiw, I found it possible to get quite far with a single monolithic lambda function that defined multiple “routes” within it, similar to how an Express server would define routes & middleware.

Anyways, thanks for writing that PDF, and good luck with Seed!

There’s a Typescript joke in here about treating the author as an “any” and not realizing that it was actually a Gary Berndhart.

Thank you so much for this comment. I can’t believe I didn’t know this, I’ve been inverting my colors & sometimes wearing cheap, non-polarized sunglasses at night — no more!

I had to fiddle with the settings quite a bit to get a 1x zoom w/ the filter, though.

It wasn’t covered by insurance, but the Freestyle Libre CGM was $37 at CVS & lasts for 2 weeks. The major cost factor (for most people) is getting a doctor to write you an Rx.

Levels is around $400, but all you have to do is fill out a form online & wait. They get a doctor to evaluate your profile & write an Rx, and then they fill the prescription & send it to you with patches & everything. Their app also does more data analysis than the default Libre app, and it puts the food-log directly on the glucose graph, which is surprisingly useful.

I have a friend who is a doctor. He was using it himself & offered to write me an Rx.

I’m now doing it a second time, using the levels health app [0]. If you can make it to the front of their massive waiting list (and don’t mind paying their fee), it’s the easiest way to get one.

[0] https://www.levelshealth.com

I’m not diabetic, but I’m using a CGM for a few weeks, out of curiosity.

My only surprising finding was that diet sodas don’t spike my glucose, but they do crash it, probably because my insulin response is triggered by perceived sweetness.

Bingo! Although, this system would categorically eliminate the issue from the following Soviet-era joke:

A man goes into a shop and asks: “You don’t have any meat?” “No,” replies the lady, “We don’t have any fish, it’s the shop next door that doesn’t have meat.”

Under the new system, the man wouldn’t have to bounce from store-to-store. There would be one convenient online shop with no meat AND no fish.

This brings a smile to my face. You don’t always need “F* you” in order to say “F* you”.

I nearly want to collocate a server with them, just for the heck of it (see their “services” page).

I wonder if stuff like this has an outsized effect on the anti-maskers, anti-vaxxers, or the general “anti-science” crowd.

I know many of these people, and they’re not actually against science at all — they just believe that the “science” being preached at them is simply political fodder or, at the very least, is intentionally biased to fit a narrative.

Ideally, they’d be able to mentally allow for different authorities / scientific foundations to exist, and 1 bad Apple shouldn’t ruin the whole bunch. Unfortunately, “scientists == corrupt” is an easier (and more provocative) conclusion when stuff like this happens.

This is a really interesting argument, and I’d like to understand it more clearly. For the sake of honing in on this:

Many subreddits have different sets of rules & restrictions. If these options were significantly expanded, could that ever meet the requirements?

Another interesting niche example is stackexchange.com, which isn’t a social network in the traditional sense (yet), but it has many different communities with different rules where a single identity, but variable set of privileges, follow you around.

Could these evolve into what you consider federated social networks, or are there more strict separation requirements (such as different login / identity providers among sub-communities)?

[edit: typo]

This is so nostalgic!

I personally learned to program on a TI 84+ Silver Edition in 10th grade. I got _so good_ at writing TI-Basic on the calculator, pressing 2nd->catalog & scrolling to a specific command... because you didn’t type function names or keywords, you selected them from a list.

I even put some games & programs on ticalc.org — I just checked, and they’re still around! One of my apps is “zz” at the bottom of this list... man, the good ole days:

https://www.ticalc.org/pub/83plus/basic/math/trigonometry/

Inclusion, control, and openness seem like a very interesting lens through which to examine office politics. It seems like inclusion & control are decided intentionally, while openness is more of a culture thing that depends on the people more than anything else.

To try & combine the 2 frameworks... sometimes there are complex incentives for intentional mis-alignment.

I guess that some information is assumed to be private — sensitive personal info, or tentative strategic shifts (“we’re probably laying off X department next quarter”). But this ever-present masking of underlying intent is something else entirely.

It does seem generally selfish — it’s one of these locally optimal, holistically disastrous behaviors. It seems like openness will inevitably slide away from any organization — it’s generally dangerous to be more open than your peers.

But, anecdotally, I’ve noticed that my assumptions about how “open” my team is become a self-fulfilling prophecy. I generally have better luck when I do the “risky” thing of acting more openly.

Or, to frame the last point differently: static analysis only goes so far, this is a dynamic system, and we ourselves are an important variable.

The segue from winning at board games -> winning at business is great, but then it goes into slightly too much detail about Appian.

I‘ve accepted that native advertising & PR are everywhere, but still... I wish it could just be a cool story about a billionaire board game champ (awesome!) without the embedded enterprise software ad (lame!).

[edit: typo]

Your last two works have been more amusing and exciting to write than they will ever be to read.

I’m imagining HG Wells on a modern software team, code-reviewing a large refactoring diff.

I’m interested in decoupling personal data from the application layer. Solid[0] is one such experiment.

It seems doomed though... and even if it does “work”, I imagine the nightmare scenario where, in order to use FB, you have to give them FULL ACCESS to aaaaaaalll of your data (email, contacts, messages, etc...).

There are also technical concerns around caching / rate-limiting, UX concerns around the complexity, and operational concerns around business models that don’t involve a Dragon’s cave full of data.

But, I hope something like this works. Or maybe just better tools for personal blogs again?

[0] https://solid.mit.edu

I have adopted this attitude for personal projects & small companies, and it was mostly great.

There was at least 1 really large error that slipped into production because we were mocking an untyped service incorrectly (a mistyped property name), and the test passed, but prod exploded. It was embarrassing, and I certainly would have caught it with more mypy discipline, but overall the pros outweighed the cons.

At big companies that use mypy, however, this is problematic. If teams only define types for the straightforward parts of the system, then the types become less useful, especially as the company scales. From what I’ve seen, individual teams use an all-or-nothing approach with MyPy, so it’s unusual for an individual dev to sparsely type anything.

The gist:

IP law lets software have the best-of-both-worlds. Source code is patentable & protected by copyright, but is also a trade-secret (because it's not publicly disclosed).

This is problematic when these black-box algorithms are used to assist in the processing of legal cases.

-----

My 2 cents:

- The overall concern that our lives are being governed by the invisible hand of black-box abstractions really resonates with me. Richard Stallman has been warning us about this for a while [0], and I've personally started considering my role in this as a software engineer [1], although I don't have any actionable advice yet.

- I don't feel as strongly that all of this code should be open-source. I think it might be sufficient to just strictly define a series of tests -- measure the inputs & outputs, describe the requirements... that kind of thing.

It's inherently difficult whenever there's a constantly-evolving ML model behind the curtain. It reminds me of the issues with using AI in the aviation industry, for instance [2]. As I recall from my undergraduate classes, aviation control systems have been unable to use "fuzzy" logic to pilot planes due to strict deterministic testing requirements.

[0] https://www.gnu.org/philosophy/free-software-even-more-impor...

[1] https://gazzini.com/essays/posts/tools/

[2] https://www.nlr.org/areas-of-change/introduction-non-determi...

I think these types of reflections are generally important — “the unexamined life is not worth living” and all that. Thanks for sharing!

I was surprised by this sentence:

My measure for a well tested system? One that lets you deploy at 5pm on a Friday before confidently heading to the pub for a few pints with the team that just shipped.

It feels... optimistic.

Pavlovian conditioning [0]: "a learning procedure in which a biologically potent stimulus (e.g. food) is paired with a previously neutral stimulus (e.g. a bell)."

In my case:

food: losing all my data (or it becoming incredibly difficult to access / control / afford)

bell: uploading all of my data to a new internet product.

----

Most companies have unsustainable business models (grow forever).

For startups, this is especially true (rapid growth forever).

This is a new internet company, but their whole shtick is that they have a sustainable business model. Given that data, maybe my bias is unfair.

[0] https://en.wikipedia.org/wiki/Classical_conditioning

This post is 1 month old.

But ignoring that, why is it too late? Have a lot of UX students come forward complaining about the program? I haven't been following the news around it, I'm genuinely curious.

This article strikes me as honest & transparent in the beginning, and very hand-wavy, we-hired-a-consultant-to-write-this-so-we're-provably-covering-our-bases-look-at-all-we're-doing-to-address-this-please-don't-sue towards the end.