vinay_ys

Cost of labor might be significant in some countries, but in a lot of countries, cost of materials is still the significant cost barrier to installation of solar. If the cost of materials were to come down significantly, even if it is at the expense of some efficiency, and also make the installation much more flexible, then definitely more installation will happen. Capital invested will be recovered in a shorter duration making the investment a lot less risky. This makes lending programs much more accessible, making the whole thing a more self-reinforcing cycle.

This business model is bound to get attacked and suffer a painful exit soon. Here's why:

First, the whole systems of chips architecture that everyone is talking about will solve for increasing overall SRAM available to keep more model state on super fast memory and avoid going to slow memory.

Secondly, anyone serious about their data (enterprises) won't be okay with making API calls to Groq. Anyone serious about their data and have a lot scale (consumer internet) won't also be okay with making expensive API calls to Groq at scale.

Their cloud is attractive only if I can use their API for experimentation toy apps to continue developing in this direction while the rest of the major industry players systems of chip architecture catches up and solves for SRAM size bottleneck and manufacturing process bottleneck, and once that's solved, I get more powerful compute for cheaper $$ to deploy on-prem.

So, this cloud strategy is short-lived. I see another pivot on the horizon.

Passwords can be compromised on client side or server side and since users tend to reuse them, compromise on fun.com can compromise your bank.com account.

Passwords+OTP plus a whole bunch of heuristics based account lockdown mechanisms improve this quite a bit. But still OTP is vulnerable for sophisticated targeted attacks.

Passkeys improve on this while actually improving the usability as well.

Passkeys come in two flavors – independent-hardware-bound (example: yubikey) and platform-authenticator-bound (example: Passkeys by Apple).

The former is bound to a single hardware token but can be used on many devices (your phones, laptops etc) to login to your accounts.

The latter is primarily tied to your Apple ID account (where Apple is the platform provider, for example) and also to your Apple devices like iPhones and MacBooks where your Apple ID is already logged in and iCloud Keychain syncing is turned on. Apple automatically and securely syncs your passkeys to those devices.

Obviously hardware-bound is a bit more secure than platform-authenticator bound, because platform authenticator can have a SPOF within the platform (a compromise anywhere within that platform stack - client-side, server-side etc. large surface area) where such a compromise is also completely hidden from the user. In contrast, the hardware-bound has a much smaller surface area. Only possibility of compromise are the hardware, the client (browser or the app), and the particular app backend server-side. Like all software stack, it is very likely this stack also has bugs that can be exploited. But this stack is relatively lot simpler to audit and secure.

Very large companies are definitely at the mercy of governments. Just look at how they are bending over backwards to comply with DMA etc. So, it is not at all inconceivable that they are forced to put backdoors into their product by the governments.

state-sponsored organizations in this space could be usually military-like organizational structure that has people with dedication and motivation unlike corporate workers. command structure can mean less bureaucracy (can also mean more bureaucracy too in some ways) when it is directly aligned with their mission. national patriotism motivations mean they are more dedicated and focused than a typical corporate worker. so yeah, there would be quality differences.

I travel a lot for work and corporate travel booking is still in the stranglehold of Amadeus/Sabre/Travelport intermediated booking agents and they suck – there's always something or the other that's broken – seat selection is not available, price is higher than direct or offers available elsewhere, loyalty points don't get accrued, etc – and the corporate travel agent insists they have to use their clunky tool. The real problem is airlines and hotels reservation systems need to modernize and adopt modern web protocol standards that enable direct discovery and booking via any aggregator including search engines like Google etc. But unfortunately, airlines and hotels IT systems are lagging.

Yeah, for case against human individuals that standard is absolutely essential. But for case against corporate entities that are already under periodic regulatory audits, that shouldn't be required. Then, within that non-compliant corporate, if there are individuals who committed the crime (criminal negligence, intentional malpractice etc) that lead to non-compliance, then to prosecute such individuals, they would need further evidence etc. But at least, the bad airplanes won't make it to airlines, just with anonymous tips.

Why does a whistleblower need to be identified? Why can't it be an anonymous tip to the regulator, who then uses the tip during audits to dig deeper and get to the truth? If they can't get to the truth even after being provided a tip, then their audit process is too broken? This whole business of non-anonymous whistleblower has too much power-imbalance and always seems to end badly for the individual.

That continent wants their citizens to be safe. So, their citizens are going to pay the price of not having access to these developments as they are happening. I really doubt any of these big players will willingly launch in EU given how big the fines are from EU.

You are not going to take the expensive human out of the loop where downside risk is high. You are likely to take the human out of the loop only in low risk low cost operations to begin with. For those use cases, these models are quite expensive.

Currently Apple wants to charge core technology fee per install (irrespective of App Store or not, after first 1 million install) and App Store commission of 14.1666% of total revenue. This seems egregious because it is charging it on customer transaction and ultimately it is coming out of customer pockets when they have already paid Apple for the hardware.

Instead, Apple should have disaggregated their fees. IMO, they can charge the following fees fairly:

1. Developer Ecosystem Tools – compilers, libraries, entitlements – can be one time or annual fees, not coupled with number of installs or revenue. (mandatory, entitlement fees could be per entitlement purchased by the app).

2. App Store registration fees – can be one time or annual fees, not coupled with number of installs or revenue. (optional – can choose to not use App Store)

2.1 App Store technical fees for app binary release certification – charged per app release/update.

2.2 App Store bandwidth fees – per app install delivered via App Store – charged per total bytes delivered.

2.3 App Store in-app purchase payment fees – percent of payment processed (currently 0.25%).

2.4 App Store ads fees – developers can choose to pay for Apple Ads for boosted discovery of their app within App Store.

They could provide free quotas for each of the above. That way free apps can still exist.

Apple can also provide iOS level capability via System Preferences or managed device profiles to lock App stores (or alternate app stores).

IMHO, this is likely the path that would be well-received and will be followed by alternate app stores as well.

Only thing that's both safe and convenient to use is FIDO2 physical keys. For 2FA I hate it when services provide more secure options like FIDO2 keys and least secure options like SMS OTP and no way to disable them. Even worse is when some sites still do security questions.

Looks like a bogus PR propaganda wikipedia link.

Ah, this page again! it's been more than two decades? since I saw this page last? It was fun to learn about coroutines from the author of PuttY the ssh client of choice on Windows those days.

clearly, you are a junior developer :-)

That's because we as humans like predictability. And hence, stock market rewards predictability. So, a large company's finance department allocates, say, travel budget, to all divisions based on their past year's travel spend +/- some margin. So, now, if you are division head, you are going to make sure your travel spend for the current period is at least as high as last cycle to ensure your travel budget for next cycle doesn't shrink. This means you may encourage your employees to use that travel budget by traveling even if such travel wasn't absolutely necessary. This happens all the time with all sorts of budgets in all sorts of organizations.

That's BS because it clearly understands what is meant and is able describe it with words. but just refuses to generate the image. Even more funny is it starts to respond and then stops itself and gives the more "grounded" answer that it is sorry and it cannot generate the image.

Since when did incident postmortems became this watered down non-technical BS. If I'm a paying corporate customer, I would expect a much greater detailed RCA and action plan to prevent similar occurrence in the future. Publishing these postmortems is about holding yourself accountable in public in a manner that shows how thorough and seriously you take it and this does not accomplish that.

At minimum, I want a why-5 analysis. Let me start with the first question:

1. Why did ChatGPT generate gibberish?

A: the model chose slightly wrong numbers.

2. Why did the model choose slightly wrong numbers?

A: ??

Well, as hardware becomes more and more powerful, what's possible in a small footprint becomes bigger and bigger. And distributed software for disaggregated storage is becoming more accessible. You put these two together, running on-prem footprints at the scale of 50-100M$ capex makes a lot of sense. In my personal experience, at this scale, (if your cloud bill for compute+storage+local-network is $50M+/year), you can get 2-4x more on-prem private-cloud capacity for the same money. Of course, this only makes sense if you have an in-house software engineering team already and the marginal cost of adding another 50-100 engineers to build and operate this is strategically valuable to your business.

In big data AI space, this is exactly what's happening with the top 20th to 100th companies in the world right now.

Firstly, they had unlimited energy and replicators - which means they could make whatever hardware they wanted.

And they also had bio-neural circuits. And photonic chips.

So, hardware was already way ahead of software.

All this goes to show that in real world, the actual science (and fiction) around material sciences was already quite advanced compared to software.

What we really know for sure is this – sleep is good, anxiety is bad. So, if you aren't getting good sleep, find ways to fix it. And if you are anxious all the time, find ways to fix it.

By one tradition that's thousands of years old – Yoga+Pranayama+Meditation (linking physical/mental_movement+breadth) and fasting + eating moderately (eat only if you are hungry and eat enough to quench hunger, and eat 'satvik' food) and cultivating positive thought patterns (practicing gratitude, humility, curiosity) is also known to solve for sleep and anxiety.

Coming to coffee, all coffee is not made the same way and everyone don't drink the same quantity per day either. And it doesn't have the same impact on everyone's body. So, it is best to avoid jumping to broad generalized conclusions based on a few anecdotal stories.

But, it is good for people to share their experiences and for others to consider them.

Your definition of 'safer': Is it about financial security, flexibility, or control over decision-making?

Your risk tolerance: Be honest about how much uncertainty you can handle, both financially and emotionally.

Support system: Do you have people who can offer support (even practical help) on either path?

Before you decide to go full-time: Analyze your current startup's potential, your financial runway, and have a fallback plan for the worst-case scenario.

Before you decide to go part-time: Be realistic about limitations as a part-time entrepreneur.

A clear mission statement can help to guide decision-making and keep the company focused on its goals. It can provide a framework for evaluating new initiatives and ensuring that they are aligned with the company's overall vision. Also a compelling mission statement can attract top talent who are aligned with the company's values and goals. This can be a significant advantage for startups competing for a limited pool of skilled workers.

This was clearly true for Google with its mission. There are so many other successful examples.

Looks like author is burned by bad startups with inauthentic mission statements.

> Due to the quorum requirements, probability of failure increases with larger quorum sizes

Well, that's not true due to two key design points:

- Size of the read or write quorum can be chosen on a per transaction basis depending on what level of availability and consistency you want. The design goal is to lower the likelyhood of the minimum number of quorum members you need for your transaction will not be available.

- Given the transaction processing nodes are ephemeral and independent from storage nodes, if a transaction processing node which is designated as a member of a quorum dies, it can be quickly and instantly replaced with a new one and all in-flight operations will be retried by the smart clients.

Well, there's no one comprehensive link I can suggest. But you can find plenty of links on arxiv.org for research publications on these topics. You can look at the ones from big companies like Google, Microsoft and Meta/Facebook. You can also search in Youtube for various conference talks on the same topics.

At the hardware level, one thing that has changed over the years is the exponential growth in network speeds and feeds – within a datacenter and across datacenters.

Layer-3 IP Clos networks are pretty much a standard within a datacenter, making multiple redundant paths available. Across datacenters, optical circuit switched networks with lots of redundant paths are very common.

Modern network fabrics also have end-to-end mechanisms for bandwidth engineering based on per message QoS/priority set by applications as well as very accurate time-keeping.

Observability mechanisms and operational practices have also tremendously improved. A single node becoming unreachable from all other nodes is quite possible. But a single node being reachable to some subset of other nodes in the cluster and not others is nearly impossible. Only way a network partition can happen is due to misconfiguration. Purely layer-3 IP clos networks with BGP advertised multiple ECMP routes are a lot simpler than layer-2 networks of the past. With modern switch operating systems, modern devops practices, Infrastructure-as-Code etc. misconfiguration is nearly impossible.

At a higher layer, disaggregated storage and compute architectures are standard for any distributed database system. In this model, storage fabric is quite simple and much more distributed and redundant. Today, it is a lot easier to guarantee a consistent and highly available transactional write to a quorum of simple storage nodes spread across multiple datacenters in a region. On top of that, building higher-level isolation guaranteeing database transactions in a tier of ephemeral compute nodes is relatively much simpler.

This kind of system is extremely resilient to failures at all levels – node failures are detected and lost processing capacity recovered in seconds; failed transactions are retried in milliseconds; loss of redundancy due to failed storage nodes are rebuilt in seconds; loss of network capacity due to failed circuits or intermediate switches are routed around in microseconds.

So, in today's context, we build databases that provide flexible transaction isolation guarantees that is selected appropriately on a per transaction basis by the application. Application also specifies the latency budget it has available per transaction. With all these improvements, thankfully, application developers don't have worry about CAP theorem like they had to in the heydays of NoSQL databases in 2012.

Recently when I plugged my iPhone for charging it correctly warned me that it detected liquid and it might damage the phone to charge at the moment. I blasted my phone with car AC for a couple of minutes and it dried out and then I was able to charge fine. It's a nice feature to have even on Macs but I doubt the scenarios in which water goes into a Mac are as simple to deal with as they are with iPhone.

Physical memory is not fungible like that across VMs. So, you can expect stuff loaded into memory to stay there unless your kernel inside the VM decides it not to.

Thoroughly enjoyed reading DJB's writing. That style of arguing is so effective and yet so easily understandable by anyone having even basic understanding of cryptography. I hope there are students of his who are as good as him (or even better) and continue this kind of work for decades to come.

At every company I worked at, each interviewer would have gone through training to do interviews, write feedback, participate in debriefs and actually do those things in a timely fashion. Senior people would have participated/chaired in semi-formal or more-formal hiring committees and helped make hire/no-hire decisions. And the hiring tools we used showed if the candidate has applied earlier for same/different roles, and did interviews and what were the feedback earlier. There would also internal confidential ref-checks based on overlap in candidate's work history with anyone at the company currently. If the candidate was rejected at any stage, the recruiter is informed and they have access to almost all of these data in the same tool. And they are expected to communicate that back to the candidate. And there is a feedback tool for the candidate to rate the whole experience and this feedback is analyzed carefully, and processes tweaked accordingly, especially if the company is growing and expected to continue to grow for a while.

In companies that had to do aggressive cost cutting, the recruiters were the first to be impacted. These roles have had high churn and sufficient training and experience quality monitoring may have suffered during this period. That could be the reason why the certain steps in the process involving recruiters (like communicating back to the rejected candidates) may have suffered w.r.t quality of interactions.