I think this is a general and important point (and sadly not at all discussed in his post). When an expert publicly says something that seems wrong, my default explanation is that they have a vested interest that consciously or unconsciously forces them to consider the implications of what they say and alter the message accordingly. Recent case in point: public health authorities telling the public that masks are useless and even harmful during the pandemic (presumably to avoid shortages) and then reversing their stance when masks became abundant.
This is especially true for public statements. If I were a friend of pg and we would go to a pub and he would not stop talking about how awesome one of the startups he invested in are, that would be a strong signal for me. But if he shills for one of his investments on twitter and on his blog just like some influencer-investor would do, I don't find this especially strong evidence that said startup is revolutionary even though he is an undisputed expert on startups.
I agree with your general point, but my interpretation of those events has additional axis:
1. Masks are in shortage, and may not protect you, and public doesn't know how to use them and will probably do more harm than help by reusing and handling and touching their masks
Then as time passed and we learned more
2. This is getting bad; even if masks don't save the person wearing them, if it helps others, we are at a point where we need all the help we can get. Please wear masks to help slow the overall spread.
While there's definitely part of the complex factoring of the recommendation that masks became more abundant, I feel initial message was "Crappy masks won't save you" and later message "Crappy masks won't save you, but may save others from yourself"
98% of people I talk to don't understand that surgical mask / cloth mask will do extremely poor job of protecting them; but may protect others. (it also adds an axis of complexity for those who don't want to wear masks because they feel they have the right not to protect themselves, because that's not what mainstream masks are for; it's not about wearing a helmet or not to protect your own head; it's about protecting heads of those around you)
This was the line for a while, but since we now know that it's an airborne infection that basically accumulates when you're in close proximity to someone infected until it reaches a critical point where it can grow faster than your body can fight it off, the mask is the only thing that's slowing it down.
Masks (and good ventilation) are about the only thing that protects you - social distancing means nothing because coronavirus isn't confined to the larger droplets we thought it was. Mask-wearing seems to result in less-serious infections even when you catch it because your initial infection was likely by less virus.
edit: I think there was an official reluctance to admit that it was completely airborne because the constant attempts to reopen businesses (like restaurants) would have been completely thwarted if social distancing (and constant surface sterilization) were meaningless.
I can't track down papers atm, but, on a more anecdotal level, there have been quite a few case studies of superspreader events where the people who were wearing masks were much less likely to contract the illness than people who weren't. Given the specific details at play, it's hard to explain how that could happen if cloth or surgical masks don't protect oneself as well.
The big problem here was that, early on, nobody knew exactly how the virus spread. So, in the interest of caution, they picked the worst case scenario, aerosol transmission, and speculated based on that assumption. And a cloth or surgical mask probably won't protect the wearer very well in that case. But it turns out that droplet transmission seems to be the better model.
So close but so far.
At a minimum please give https://www.thelancet.com/journals/lancet/article/PIIS0140-6... a read. The real story is that SARS-2 does spread through aerosols whereas droplet transmission is pure unproven dogma. The paper I linked goes into some reasons why that is the case (for ex the fact that transmission is more likely close to someone doesn’t actually provide strong evidence for droplet xmission)
EDIT: I should add that even if it were primarily droplet transmission - which I very much doubt - masking would likely still fail for the stated goal (source control) in a community setting due to improper usage. And improper usage doesn't just mean "the mask is below your nose", it means "you're not changing out the mask the moment it gets damp", "you're touching your mask with unwashed hands", "you're touching your hands with unwashed mask", "you're standing closer to your conversational partner to compensate for the fact that masks muffle hearing" (<--this last one isn't "improper usage" so much as an inevitable result but I digress). And this is all without discussing any of the various negatives of mask-wearing that it's become trendy to pretend literally don't exist.
But to your point, you already correctly hinted at the fact that if aerosol transmission is the dominant transmission mode, then masks don't work even in theory, let alone in practice.
note: "Evolving understanding/new facts" or "new/differing priorities" are not the same as "deception" to me. e.g.:
* Thinking A and Saying A; then changing mind and Thinking B and Saying B; are not a deception
* Saying "Because of X, A"; then later "Because of Y, B", because Y became more important than X, is also not deception to me
* Thinking A but Saying B is a deception to me (for reasons and intents that I may or may not agree with)
Once the mask supply was able to meet demand and after they were reasonably certain it would protect against the spread of the virus, they adjusted their recommendations accordingly.
None of that is deceptive.
Yeah, lots of time has passed since people started to use masks to prevent a spread of infection. Like, 100 years?
My impression, in my locality, is that at beginning of pandemic people focused on notion of using masks to protect themselves (and many though not all governments indicated that's not efficient/recommended).
Then "as time passed and we learned more" focus moved to using masks to protect others (though many individuals, in my circle, still aren't clear on that).
My impression is that we have evolving, and still not necessarily 100% certain evidence/understanding, on how it spreads and what are the most effective measures. It's made more complex because
a) there's no silver bullet; most measures increase your chances to some percentage. This makes discussion between experts and public more difficult as public tends to think in binary terms.
b) While yes there are many public health measures we've known for 100 years (wash hands, have clean water, cook/boil/heat things to sterilize them, sneeze in elbow/Kleenex, wear mask, remove waste, etc etc), not all are equally effective against all vectors. What seems "Common sense" / "Logical" to a layperson like myself, may be more nuanced to an expert with experience.
I mean, for what it's worth, I'm 100% certain my dad, a year in, is still worse off for using a mask because of how he uses it. Many and especially older people around me reuse their masks for days and weeks , touch them constantly, put them under their nose for prolonged periods, don't squeeze/tighten them sufficiently, etc. Even if all that touching doesn't hurt, their belief that they're protected coupled with incorrect usage coupled with likely increase in risky behaviour is a net negative.
People can scream liberty and freedoms and personal responsibility, but I feel public health officials have to look at cold hard facts, of both disease but also people's actual behaviour (as opposed to some ideal non-existent form) and how it actually affects spread rather than how it logically intuitively should.
You may argue whether it is okay for a technocratic establishment to lie to people and manipulate them; but to pretend that they didn’t feed the public with half-truths is just ridiculous.
Masks right behind washing hands - honestly, until Covid, I clearly lived in ignorance and had NO idea that a) Surgical masks are tested and designed to protect others against exhale and b) N95 masks are tested and designed to protect occupant via inhale. Just that basic, yet absolutely crucial concept was unknown to me. I had very limited exposure to them across two continents and half a dozen countries, and it certainly wasn't part of my own education about most important hygienic achievement or practice across four different educational / political / health systems. That doesn't necessarily mean masks aren't crucial, but it perhaps doesn't make it obviously so either in context of daily public health.
And then we have the mucky complex detail of aerosol vs droplet vs surface, sizing and staying power, etc.
You are presenting / clearly strongly believe that (all?) masks are a binary, always-good, always-clearly-protective thing.
My understanding/interpretation is that it's more complex than that; I am OK with scientists, when faced with a net new illness, inherently evolving the opinion and recommendations. I guess this is where you see a "technocractic establishment lying and manipulating". I look at wildly different countries with wildly different values, framework, goals, assumptions, methods, and yes "political wiggle-waggling", and most of those culturally politically different scientists following broadly similar path of learning and understanding.
It is likely our frameworks and life paradigms are too different to come to an agreement - at some point we all, you and I included, fit inbound facts and perceptions into our root approach and beliefs.
So any argument I could realistically make against $sillyAppThatHasVCBacking will not matter if enough Paul Grahams back it and $sillyApp makes itself a moat.
WeWork would be the most prominent example of this; even after their 2019/2020 crash, they'll probably come out of the pandemic okay compared to its competitors simply because it just got big enough.
https://techcrunch.com/2021/02/05/a-look-at-how-proptech-sta...
I don't mind Mighty as a product, I don't mind their team, their pricing or their slick marketing website. But please, call it what it is: A nice and slick Remote Browsing product, one of multiple ones. Cloudflare recently launched an RBI product with much more humble and honest marketing about where it will be useful.
The only way to solve web app performance is to put 5-7 engineers in a room and say "we want it fast. As fast as possible. Get to work."
But no, they have the front end team, who has to run everything by the compliance committee, and the architecture committee, and the API committee, and the back-end team, that has to run everything by all the committees and the front-end team, and they also have to wait on the infrastructure team before they can test any of their stuff, and before you know it 12 weeks have gone by before an engineer has written an actual line of code.
And none of this corporate structure can be fixed because the CEO doesn't understand it, and all the management he hires to fix it are financially incentivized to continue holding their little fiefdoms.
But for other people? Can you imagine Adobe saying "Here is our product, now please purchase this third-party cloud service to be able to use it." They either improve their software or launch their own server-driven app to capture those $30. That's my take.
And I mean that in a bad way.
Privacy and users rights are already pretty bad, but we still own our computers.
I can see how it would start in the corporate world.
I thought we have moved past thin clients and moved on to centrally managed software, either in the form of javascript web apps or as apps from an app store.
Thus they insist on making the nodes (the servers) smart and the clients (leaves) dumb.
The problem with that, of course, is that the clients inevitably gets faster, and cheaper, and software gets pushed to the leaves again.
We're carrying around what used to be super computers.
This is also why I would be very cautious about investing in something that bets on offloading computation from clients: If you time it right, it could be huge for a while, but you're betting against advances in the speed and cost of computation. The same advances you'll rely on to scale and drive down costs.
As such it feels to me like they're playing a game of musical chairs.
1. Software is too heavy, let's push it on the server 2. Dumb client gets faster 3. Servers cost money, let's use the unused CPU cycles on the client 4. Bloatware happens, go back to 1.
No surprise that this means that there is no single solution. Complexity sometimes makes sense at the edges and sometimes make sense in the core. It just depends.
I think we have a lot to learn about building systems that give us the flexibility to move things around as needed. Plan 9 was interesting from that point of view because it gave a way for edge resources and core resources to be composed via 9P at any place in the network.
Consistency is less of a challenge than it might seem. It's a challenge if you're frequently disconnected from the network for extensive periods of time and might access and modify data from multiple disconnected devices in that period (note modify, not augment and create something new, which is easy to accommodate). But supporting that is very different from supporting mostly-local computation.
Years ago, I pushed as much work onto the client as possible to reduce the workload on my servers. Now, people complain if you push too much work onto their phones. They want the server to do the heavy lifting so the app can be more responsive on their low-powered mobile client.
"The internet" doesn't define how much weight each end of a connection should bear. It doesn't really even dictate that there are only two ends.
Should a webapp have a braindead REST api with a select, insert, update, delete for each table of the underlying db model and have the client make 1000 nested calls to the server to render the simplest thing? Or should the server api be much more sophisticated so that a single api call can provide all the information required for the render?
There's no one right answer just because "internet".
I think about a central server with multiple terminals on a network. There the node is the beefy boi and the terminals are just I/O devices. Kind of similar to what MightyApp is doing - except with fewer privacy concerns :)
It is not an axiom though, but rather an architectural guideline that comes with its own set of trade-offs and must be applied judiciously.
But from a business perspective, there is huge value in dumb clients (not for the end-user)
And that it's mostly people who don't care about this on deeper level are why we have such a bloated ecosystem. Layers of bad code stacked on layers of bad code.
If there were more jblows and muratori's in the world, Mighty would have no reason to exist. The web would already be performant. That PaulG is bullish on Mighty strikes me as a pessimistic view and a bet that the underlying problem won't be solved.
Casey and Jon and from the engineering tribe. The cool stuff tribe is what I call the business oriented tribe.
But I think you're missing the point.
The technical merits of Mighty is a secondary problem.
I am most likely to phrase the general position (which I think Jon and Casey would support) put forth as follows: the problem a service like Mighty is trying to solve ONLY exists because the standards and practices of modern software development is fundamentally broken.
You even touch on this in your GP post about devs taking the shortcuts to ‘making something cool’. As the parent post calls it, the cool stuff tribe are those developers who will use available cheat code because they expect that doing so is acceptable.
Your GP post seems to try and counter the Jon and Casey position by saying that because developers just want to make things that they will and there is no concern for any impacts these accumulating decisions may have. I think the ‘you’re missing the point’ comment derives from here. You seem to be saying that Jon and Casey’s position is not palatable to devs because those devs are not concerned with performance and don’t like to be told to consider such aspects of the things they make. But, J&C’s point is that if the standards for developing software were not so broken then the position of the cheat code using devs would be uniformly decried as substandard and unacceptable. In a world where development standards were in line with J&C’s views Mighty would not be in a position to be a viable product because using the web via a native browser accessing properly developed web content would be a pain free and perform any experience.
I don't think this is accurate. I watched a talk with Jonathan Blow recently and he went on for quite a while railing against premature optimization.
The optimizations make the code more brittle and hard to change so that when you reach a position to measure where the bottlenecks are it is too complicated to do the proper optimizations.
I find it strange that for someone that is a relatively immune from any real scrutiny, and constantly claiming to be a bold thinker, PG is always so coy in his writing.
I would find PG's recent stream of ego driven rants much more enjoyable and potentially insightful if he would just say what he's really trying to say.
He might be trying to add strength to his arguments by making them somehow more general, but since these pieces always seem very clearly about a specific bone PG has to pick, the result is they read as some of the most cowardly essays I've ever encountered.
He is trying to build a stronger case by using an elaborate straw man argument.
Turns out to be quite transparent and poor.
"The browser uses a split architecture where some of the processing is performed on Amazon's servers to improve webpage loading performance."
I abandoned it right away.
I'm sorry, but 1 sec is not fast...
Well if you fund 1000s of ideas, some of them are bound to be successfull. There is no surprise there. That's basic maths.
funny how they all turned out to be mixed bags. the best part? people telling me web apps are "lighter". even back then I knew that was a hot load of bullshit.
good to know the other shoe has dropped and we're really going full-circle to thin clients/mainframes, but shittier.
The $50 subscription is to avoid flooding their servers while testing the software.
This product will likely starts as a corporate malware but once the beta is over and their tech is really working at scale (more difficult than you might think) they'll probably give it for free or very cheap.
They wont give away this for free:
> Each browser instance gets 16 vCPUs using state-of-the-art Intel CPUs running at up to 4.0 GHz.
People would mine bitcoins with it. Not to mention they'd be footing more than all of youtubes bandwidth costs if this gets popular.
So you think it is a dragnet-surveillance get-acquired-for-our-data play, longer term? That's even worse, if so.
This is the kind of asymmetric dismissal he’s talking about, and it’s not very good.
Dangerous? We run everything via cloud services and encrypted communication. “Sounds technically ridiculous” - so did probably every modern technological idea when it was new “you put your database in some other company’s servers?!”.
You’re mostly proving PG’s point.
Delegating power/freedom is dangerous, we're already delegating too much.
That said, the potential for mighty is real and dismissing it for these reasons is dumb. The same logic would have also dismissed nearly all modern wildly successful tech companies.
This is the subtext I came here for, thanks.
I'm undecided on Mighty, I don't have a strong opinion. I wouldn't consider it "revolutionary", unless perhaps they have a vision way beyond the product they've talked about publicly, but it would seem a bit unfair to use that as a defence against criticism.
One thing I did notice is that a few people involved or indirectly involved with Mighty seem to have a problem with the criticism on HN, which seems really strange to me if you consider yourself an open individual who supports freedom of discussion. You don't need to engage with that criticism directly, but presumably it's useful to be criticised - you'll need answers to those arguments. And if the criticism isn't constructive, well, who cares? There's always some noise.
It feels to me like "oh yes I want people to freely debate ideas and criticise them but just not the ideas I like".
An idea isn't bad if it's valuable and built on ridiculous / dumb / silly / simple / old technology. A product should be measured on output, not input... in fact I'd go as far as to say that we (hackers) should celebrate ideas that deliver incredible value with such a simple implementation.
Many things are considered ridiculous or impossible before they work.
The main point is that yes it could work and be valuable, but not for the end-user. It would be mostly valuable because of the transfer of power from the user to the server.
We should know better, we should learn from the lessons of the recent past, we should be more careful before giving away our freedom (power is highly correlated to individual freedom)
The Jealous Nerd:
> One reason they do it is envy
The Hipster:
> it's an easy way to seem sophisticated
The Dark Ages Inquisitor:
> Darwin's harshest critics were churchmen
The Luddites and Sheep:
> the sheer pervasiveness of the current paradigm
This replies to none of the legitimate criticism, and manages to be more dismissive by filing everything under "crabs in a bucket, extra salty".
They are arguing that technology shouldn't need Mighty. The mere fact they are complaining about the state of the world means that there is a problem that needs solving.
I'm rocking a 8 year old laptop. I also use NoScript, which means I'm pretty aware of what code webpages are actually running on my computer.
What I've seen is that the things that make it really chug have very little to do with how much actual functionality the website has. Beautiful CSS animations usually aren't too bad, either, even on my old computer. The real performance hogs tend to be things like scrolljacking, telemetry, and dynamic ad placement.
The interesting outlier here is gmail. Gmail fascinates me, because it keeps getting slower and slower, without, as far as I can tell, actually gaining any new capabilities.
Mobile Gmail's not just heavy—it's broken. Whatever stupid, misguided bullshit they're doing with scrolling makes it register clicks where they weren't intended if I'm not super careful.
I want to visualize 1TB of single cell RNA seq data in a browser tab, then open a new tab, change some params, and share the link to a colleague. I want it to be instant.
In the non-multiomics world I also make browser based data visualization software (most recently worked on this, for example: https://ourworldindata.org/coronavirus-data-explorer). I want to load up 500GB and facet on multiple dimensions and then share the link and have it all run instantly.
I'm not talking about Gmail and Text editors here. M1s solved that.
But why? Wouldn't you be better suited writing a native application that would do this better? This seems like you're building a problem for this solution. Most people use a browser for Gmail and to read news. Your example already works really well on the browser on my phone
Because cancer is incredibly complicated and requires incredible amounts of data, analysis, and *collaboration*.
Maybe I'm missing something in your problem statement but are you downloading all that 1TB in your browser window and changing parameters? If yes Mighty has to download that data too and considering that each user is sequestered, someone you're sharing this with will have to download all that too. If you're not and the visualization is running somewhere else and the result streamed to your system how does Mighty solve that problem?
The idea that you're going to shove 1TB of data down to the client strikes me as slightly unhinged. Even if we assume you can achieve a sustained transfer rate of one gigabit per second, it's going to take over 2 hours to get a terabyte shoved down to the client. I'm guessing, though, that the actual vizualization is nowhere near one terabyte. Data's going to have to be aggregated somehow, because no computer monitor can display a terabyte worth of information all at once. Even a 4K monitor would fall short by many orders of magnitude.
It would be much faster, and, I think, simpler, to keep the data on a central server, have it generate the visualizations, and push them down to the client.
Yeah which is why I asked for clarification from them because it seemed really weird to assert that as a positive for the app?
> It would be much faster, and, I think, simpler, to keep the data on a central server, have it generate the visualizations, and push them down to the client.
Yeah I completely agree but in this case, Mighty would have next to no advantage over just loading the visualization on your own machine. Hence, again the confusion about asserting Mighty as a means to an end
But if you're the actual app developer, I'm just not seeing a good reason why you would want to deliberately incorporate this technology into your design. Why farm the sever-side rendering out to a middleman when you could just... do server-side rendering?
Distribution. There's only 1 magic platform that allows you to ship new versions at your own cadence to 100% of people.
Developing for just M1s is great, when all your users are on M1s.
How do browsers make something distributed? How are browsers the only "magic platform" that allow you this? How is Mighty contributing to solving this problem? How does M1 matter here
- mighty creates arbitrary APIs that allow more speed so now we have yet another build target
- it manages to become successful, so now competitors jump in and now we have multiple targets to build for.
- god forbid, competitors offer “free” versions that become even more invasive to your browsing
People already have a lot of power in their machines for their non-web apps, and it is enough to run web apps as well.
That sounds like an odd use of "skin in the game". "Skin in the game" would be if that thing failed and then he gets skinned, basically. Being invested in something with potentially lots of upside and worst case a bit of lost investment isn't like that.
- Also IMO we should let mighty app (or for that matter other crazy ones) play out. We all know little about the capabilities of individuals and what the future holds. So why to ridicule and prematurely declare certain death.
Delegating power/freedom is always dangerous, especially if you're doing it for free.
> It's much easier to kill snakes just as they've hatched.
Aren't a ton of itty bitty snakes going to be harder to kill than one big snake? I've never tried to kill snakes of any age myself, so I could be way off.
- I would agree with your second point about most tech, but this is special, I'd say this is political.
And in this particular case, I don't think this is a valid defense.
First, he clearly has too much skin in the game to be credibly neutral about it.
Second, he avoids addressing the main critique about this "new tech".
People are not claiming that it is a bad idea because it is infeasible or not valuable, but because it is dangerous and also because it sounds technically ridiculous. (thin client inside thin client)
https://www.mightyapp.com/ https://twitter.com/Jonathan_Blow/status/1387101172230672389 https://twitter.com/cmuratori/status/1387645578067124224