edraferi

Find a system or process that uses a series of rules to categorize, label, or action things, especially one that is occasionally incorrect. Model these rules with a ML algorithm using the rule outputs & user corrections as labels. See if you can build an ML system that out-performs the rules. (If you can, it'll probably be by looking at data that the rules didn't consider.)

Note: your ML system will likely be less explainable than the existing rules. This won't matter as much if the current rule collection is already more complex than a human can deal with. It will matter a LOT if your decisions are subject to regulation.

Do any of the TV review sites test these? That would be very helpful since you can't showroom them.

And Pomegranate

https://github.com/jmschrei/pomegranate

> Every layer understands the command it was run in the Dockerfile to create itself

How reliable is this? Can it be modified after creation by a malicious party?

That is, if I get a wild docker image, can I trust the results of `docker history`?

I like having a live demo available, but not on the landing page. It should be one click away to avoid this kind of confusion.

Growth is coming to the region, regardless of what Amazon does. HQ2 is just an accelerant. Embrace growth or resign your community to irrelevance.

I think Metro hit bottom a couple years ago and is on the way back up. Dedicated funding (finally!) will enable steady improvement. If Amazon comes to town and uses their clout to push for better transportation, that will on accelerate things.

Ezra Kline interviewed Anand Giridharadas for the Ezra Kline Show podcast back in September[0]. Very interesting.

Anand basically argued that the global elite spends a lot of time looking for ways to help the global impoverished, but fails to make an impact because they never consider solutions that would erode their own wealth/status.

[0] https://www.vox.com/2018/9/5/17821522/anand-giridharadas-win...

This thread was very helpful:

> Explain this to me then: What's the difference, conceptually, between (A) a VM host running a unikernel capable of solving one particular kind of problem, and (B) an OS running on bare metal running a process capable of solving one particular kind of problem?

They are very similar. The main difference is the shape of the OS APIs. In the former scenario, the host provides things that look like CPUs, NICs and things that look like block devices, with no high-level abstractions (e.g. threads, sockets, and filesystems). If these are needed, then the unikernel will link in device libraries that provide them. In contrast, in the later situation the OS will be providing these even when they are not needed and they will add overhead to those code paths.

Additionally, the amount of process state owned by the OS is far greater in the latter case. Things like file descriptors, thread priorities and so on make it far harder to snapshot or migrate an OS process. Hypervisor interfaces tend to be as close to stateless as possible.

Even !g does some masking for you, although you still wind up on google.com

Because there’s too much overhead, it’s too hard to share, and the benefits don’t show up until the problem is more complex than most people ever need.

You have no right to privacy when crossing an international border. The solution is simple - don't carry any data you don't want to give to the country you're entering.

How does this address the unique parts that LEGO created for this specific build? For example, they created interesting screw-adjustable beams to mount the curving outer skin.

> supporting the weight of a person and a car in the air across the distance between the front and read axles would have seemed pretty impossible to me

Yeah, there is a core steel chassis for that. The main LEGO stuff is the body and engine.

Any chance you can share that spreadsheet?

I'm not too worried about acquisition. Per the article:

> ...there was little need for the company to raise more money. DuckDuckGo, which had a last known raise of $3 million in 2011, has been profitable since 2014.

Anyone interested in this should read the Atlantic feature about it:

https://www.theatlantic.com/technology/archive/2017/08/insid...

> Only tier 1 dotcoms can run stuff on GPUs economically. For everybody else, just getting many times, possibly few magnitudes, more commodity x86 machine time still makes sense.

Disagree. Anybody running appreciable ML workloads buys and racks their own GPUs.

> centralized computing in clouds [could] make having hardware that's optimized for some specific workload much more tenable.

I think this is the key. GPUs are pretty much the only consumer-level specialized co-processors, and those have been selling like gang busters. Plot twist, it's the data centers that are buying them up... to the point that the manufacturers tried to prevent you from buying more than one at a time.

Of course specialized processors need to have proper networking and storage proximity to really deliver value. May be more challenging to offer a specialized service like this than you'd hope. Perhaps the big cloud vendors will dominate again, so chip manufacturers need to just hope for a few big contracts to Amazon / Microsoft / Google.

I think part of it is the secure element. Apple is moving that direction with TouchID on MacBooks.

They have a couple Data Models:

Calendars [0] is a simple Event + Attendees model Contacts [1] Re-uses vCard Mail [2] wants RFC 2822 compliant strings Photos [3] have just basic metadata with a URL to fetch pixels Tasks [4] is a simple to-do model

Overall the data model seems like a significant amount of wheel-reinvention. These data models are all just JSON records... they should be JSON-LD with @context pointing to a shared schema, probably defined at schema.org.

[0] https://github.com/google/data-transfer-project/tree/master/...

[1] https://github.com/google/data-transfer-project/blob/master/...

[2] https://github.com/google/data-transfer-project/tree/master/...

[3] https://github.com/google/data-transfer-project/tree/master/...

[4] https://github.com/google/data-transfer-project/tree/master/...

> That's something quite difficult to set up and scale.

Especially if the admins lock themselves out of the kind of data you need for fraud prevention.

Very interesting that their main reason for rejecting ActivityPub is GDPR. Basically, you can't delete information after its been federated out.

This seems to go to far. Give the users control over their data in your system AND give them control over how their data is sent to other systems.

Edit: In the linked issue, OpenBook CEO said that you can't delete mastodon accounts, but Mastodon got that feature a couple weeks after his post:

https://github.com/tootsuite/mastodon/pull/3728

Yes. VSCode is good. Skype for Business is a garbage fire.

Hmm in my community it’s very odd to get married before your late twenties. Children and house come several years later (if at all). Looking at the parent and grandparent comments I think there are two main reasons for this:

(1) If you grow up around educated people, college is the obvious way to kick off a financially stable life. That means school until 22+ followed by working for as much money as possible to dig out of student debt. By the time you can breathe a bit, you’re 30.

(2) Marriage is moving from a foundation to a capstone. In my social circle, virtually nobody got married in their early 20s. They didn’t feel they could before figuring out their careers, choosing a city, having some independent adventures. Sure there was romance, but not the long term commitment and financial integration.

I think The common thread between these two issues is “hunger.” If you’re like GP and grew up in a poor area, you’re willing to take risks to get out. It makes sense to marry the best partner you can find locally and climb out together. You’re not “discovering myself before settling down” you’re pair-climbing a mountain, recognizing it’s safer with help. Hopefully when you reach the summit the shared experience will keep you together against the pressure of expansive options available there.

One key question: does marrying reduce or increase your geographic flexibility? For college-first people, you have to move around while single to build your career, then marry wherever you wind up. If you marry first, it’s harder to find a place that optimizes earning potential for both partners.

I think a lot about what this means for my own children. I want them to take more risks, but I don’t want them to fall short of what I have achieved. I want them to support themselves as early as possible, but I know education is generally a more profitable use of time. (Should they be working minimum wage jobs or studying or school nights?) There are many reasons why the youth unemployment rate is so high.

Gnucash, but I haven't gotten the hang of it's budget features yet.

Most budgeting is done in a spreadsheet. Cash in --> Cash out, line by line.

This is common in the US.

Apparently pregnancy is a protected status, but parenthood isn't: https://www.nolo.com/legal-encyclopedia/federal-antidiscrimi...

Does “organic” necessarily mean non-GMO? Personally I think GMO is great but dislike massive use of fertilizer and pesticides. And really, that’s mostly for downstream effects. I want to encourage sustainable land use and reduced environmental harm. Fine with me if we need to hack some plants to enable that.

I recognize that genetic modification is sometimes used to enable greater use of these chemicals (eg roundup ready grains).

There's an economic theory that job satisfaction is a kind of compensation. That is, you have to pay people more to do unsatisfying jobs.