There are three types of loss: destruction, discoverability, and synthesis.
Destruction is the one most people think of. The last tape is destroyed, or someone nukes everything in S3. Policies normally save this from happening, bit not always. (I note that lawyers are pretty good at saving things because of discovery, so perhaps corporations should sue themselves more often /s)
Discoverability is when the data is there, but noone knows where it is. This is your classic corpo Sharepoint/teams/slack problem. There was a chat, once, with some shared files, from their onedrive, but the last person who remembered where was laid off a year ago and...well, that data is effectively lost. I dont think we have great solutions here except writing even more things down and hoping that everyone follows rules like keeping all files in one place
The last case is where the data is located, but noone can vouch for it any more. Internal wiki pages are notorious here. People post about, say, some sharp edge they uncovered, and the page sits there for all eternity. You discover it years later. Can you trust it? Is there anyone who knows why the page still exists? It blocks the synthesis of what you know with what the author knew.
I talk to corporate normies a lot, and what I see is this:
- copilot as a "internal super google" is incredibly valuable. However, no company is so completely on the Microsoft train that copilot alone cam do what they want. Eg if they have Salesforce, then tough luck as copilot cant "see into" that. Moreover copilot today has strange limitations even inside the 365 stack, eg with OneNote. If microsoft can build a corporate-focused RAG that genuinely mops up all these kinds of sources, they will have a winner on their hands.
- copilot as coder is a mixed bag. Good for jumpstarting tasks, but not reliable enough to be left unguarded, ie experienced devs still need to be paid to shepherd it.
- everything else: not much of anything achieving real, deep value. Lots of anecdotal use cases like drafting emails.
- scary as hell: the rise of shadow AI IT where corp users use chatGPT with company info, ignoring the risk that what they ask gets folded into training data. Proprietary info leakage is going to be a problem I think.
None of the above makes any of the tools worth $30/mo per user. A super google copilot that genuinely captured everything across all silos and presented a unified interface (none of this cruddy plugin agent nonsense) would be worth north of $50/mo at any large corporation.
>scary as hell: the rise of shadow AI IT where corp users use chatGPT with company info, ignoring the risk that what they ask gets folded into training data. Proprietary info leakage is going to be a problem I think.
Microsoft, for all their faults, seems to have honed in on this issue explicitly. Enterprise Data Protection seems like its designed to massage this fear, whether it works or not who knows.
Except ChatGPT dot com somehow gives much better results--even without signing into an account!-- than does Copilot. I thought it was pretty much the same model under the hood but the results speak for themselves.
> None of the above makes any of the tools worth $30/mo per user. A super google copilot that genuinely captured everything across all silos and presented a unified interface (none of this cruddy plugin agent nonsense) would be worth north of $50/mo at any large corporation.
I feel like you pulled those numbers out of thin air. $30 is something like 30 minutes of work or less in a typical office. Can a super-google that covers most of what you need (but not everything in every silo) save you 30 minutes per month? Yes for sure.
> everything else: not much of anything achieving real, deep value. Lots of anecdotal use cases like drafting emails.
Well, this is what people are missing IMO. Yeah, general purpose chatbots can’t solve advanced problems. However, models and pipelines (using similar technology as chabots) specifically built for certain tasks can achieve really incredible results — like in the area of protein folding (https://youtu.be/P_fHJIYENdI)
I think there is a lot of potential, but most companies are just slapping a chatbot (which uses some other company’s models) on their app and calling it a day. Very few are taking a deep look at how ML could actually solve real problems.
Most salaried staff aren’t fully productive every minute of the workday. Studies show a typical work week’s output can be achieved in 30 hours without losing productivity.
I’ve seen this firsthand throughout my career. (That’s why I moved to working with remote-first, globally distributed teams over five years ago.)
> - scary as hell: the rise of shadow AI IT where corp users use chatGPT with company info, ignoring the risk that what they ask gets folded into training data. Proprietary info leakage is going to be a problem I think.
IT Sec has these things blocked for this reason. Probably made easier by only allowing company devices on the company network.
Looks like Day 2 snuck up on Amazon while they weren't looking.
The retail site is drowning under a wave of off brand no name muck. Even buying a brand name (say, Oxo kitchen tools) gives one no comfort that what arrives wont be a knockoff.
The purchase experience, once world beating, remains excellent, but not especially different from other stores. That leaves delivery, which with Prime is still a differentiator, but then again when it asks me to defer shipping for a few days to save boxes, which I do, I start to wonder why I'm paying for fast shipping only to pick slower shipping when I check out.
As for the TV, it has an uncanny valley feel for me. The documentaries, for example, seem like the kind of low grade filler you get on inflight TV. A few talking heads, voiceovers over still images, etc.
Webapps, and the generation of people who built them (who very likely grew up in the era of desktop apps with activation keys), killed the conversation off. You share your identity with the webapp provider to get access to all but the simplest apps, and there's a tiered access model going from freemium to premium. In particular, freemium attempts to remove that "I just need it for one quick thing" friction.
That said, the reasons for piracy remain the same: friction and price. It goes both ways really. People pay for 365 because free Office suites dont quite do what they want and that introduces friction. People pirate Photoshop because its outrageously expensive relative to the value that it provides to the pirating user.
Seconding your ending statement there: I make good money, and I pay for a lot of software to make my little life spin along, and the little bit of theoretical piracy I consider doing, satirically of course, is far more to do with how onerous the procedure is to acquire the product legally than anything about it's costs.
Movies are a great example. If I can stream a given movie on any of the services I already purchase, obviously that's great and I'll do that. Failing that, I'd like to buy it on iTunes as I trust Apple a little more than most others in the space (YMMV). Failing that, if I can buy temporary access on Amazon Prime/YouTube/Whatever, sure. If however I'm essentially expected to fork out yet another $10-$20 on yet another subscription with yet another mediocre app that will slow my Apple TVs to a crawl, or worse, demand the right to shove ads in my face... no. Pass.
I also ditched Photoshop's $10/mo subscription after years because the quality of it just kept going down. I remember being so excited when the first update for the M1 Macs dropped, and Photoshop suddenly felt to use like it did in the 2010's; snappy, performant, efficient. And it took barely 3 years for Adobe to manage to bloat it up so it ran like shit again. So I dropped that and spent $50 on Affinity Photo and I couldn't be happier. Granted some things take a few more steps to do in Affinity, but those steps are all quick, snappy and responsive unlike seemingly every action in Photoshop.
It blows my mind how the software and larger entertainment industries refuse to learn the lessons of Steam. Steam isn't perfect, of course, it has it's issues here and there but it made great strides to reduce piracy not by locking the products down so hard and with such onerous software/rootkits/DRM but instead by just making the idea of buying and getting the game so obtusely convenient that pirating it is more work and why spend hours downloading questionable software and carefully going through an installation procedure when you could just wishlist it there, wait for it to go on sale, grab it for $10 and have it in the time it takes you to make a coffee?
Software has this (very convenient) problem of only selling tool chests, not just tools.
So if you just need one or two features of a program, then you still need to purchase...err, rent...the entire thing. Sometimes you even need to rent it for a full year.
Thankfully it seems LLM's are on track to solve this problem at least partially by being able to effectively code very narrow scope solutions. You can't prompt Claude to write Photoshop, but you can prompt it to write a program that scales images and applies a basic filter. (Yes, I know there are a gazillion free photo tools, this is just an accessible example).
Not sure but would say that people pirate Photoshop even more since it went full subscription model. From my experience, once I learned they're going this way I bought the last Lightroom available and have it on my DVD. Of course this has been quite a time ago and it will not work with new cameras' raw formats but I guess I will just buy older camera, not the bleeding edge one, if my camera breaks beyond repeair
My use case is: do some photos once every two months and make them better using lightroom. Paying monthly for it makes really no sense for me.
I've tried Affinity. Unfortunately a literal lifetime of photoshop has made it basically impossible to switch. It's like getting into a car and all the buttons are rearranged and behave different.
The jokes on me. I pirated photoshop way back in the day. Now I have no choice but to shell out a subscription.
I like Affinity Photo a lot, and it's what I use, but Photoshop is definitely the more powerful program. Affinity still doesn't even have an auto-select.
> the days when you could trust a company to be fair
Those days never really existed. It was simply that their misbehavior affected groups of people who didnt have access to the media and power structures. For the US, e.g.: central Americans (banana company inspired coups), native tribes (water pollution, deforestation), poor whites (coal ash pollution), etc.
I can see that companies treated their employees better, but that might also be correlated with strong unions, less regulatory capture, more competition, or some other factor, rather than intrinsic goodness.
> I really don't know what the modern day equivalent is
This comes up on HN quite a bit, without a universal answer. I suspect that each respondent answers through the lens of what they remembered, or liked, about the platforms that were formative for them.
Example: 1980s kids seem to have had formative computer experience where the computer was a self contained thing: no understanding of electronics needed, instant on, built in programming environment. Also there was a rich ecosystem of accompanying offline media, eg books on BASIC, magazines with program listings, etc. They gave their full attention to the machine.
I also see people who grew up on Raspberry Pi and they have a different view, much more about hardware and pins and voltage. Equally valid but much more electronics-y. Also has a rich offline ecosystem.
Of course modern hardware and OSes dont get close to either and so the closest experience is probably python in a browser, which really feels more like Learning for a Job than Fun with implicit Learning. The browser itself is an attention suck too - all too easy to switch away as soon as a little boredom sets in, and, as we know, boredom spurs creativity.
I dont know if this exists, but a Switch-like platform with HDMI to the TV plus built in screen, BT keyboard, and a minimal, barely there OS might be one path. But it might still fail without the offline media. Maybe you sell it with an accompanying book of games you have to type in.
I dont think it's particularly complicated. Bezos wants to makes money, and preserve his companies. He judges that the best way to do this is to make some signs of fealty and hope that the storm passes. Additionally it must worry all the tech bros that Musk is in Trump's innermost circle since he is both smart and a wildly loose cannon.
If Trump really wanted to hurt Bezos he could, for example, decide that Amazon could not sell anything from China, could not bid on space contracts against SpaceX/Starlink with Kuiper and Blue Origin, and must offer untrammeled access to the FBI to anything on AWS. So tossing the Trumps a few million and performing a little dance is a cheap insurance policy.
Everyone in the tech bro orbit is doing a similar calculus. Microsoft must be quietly thanking their stars that they never succeeded at social networks and therefore avoided the whole moderation/censorship thing, but even they have political vulnerability.
You can admire SpaceX's innovation and dislike Verizon while still being concerned that this is a land grab by Musk.
We are talking about the FENS contract here that was awarded after a competitive bid process in 2023. Verizon and their partners (eg Saab aerospace) beat out HarrisL3, Raytheon, and some others.
If Musk is asserting that the contract was improperly awarded, then there are very well established processes to appeal. (And such appeals would also stop work on the project pending resolution.)
However what I see instead is that he is simply instructing the FAA to cancel the deal and use Starlink/SpaceX instead. That is at best murky and at worst corrupt.
As for Musk's companies' ability to execute on the details of the contract, knowing how to make cellphone calls via a satellite is a million miles away from knowing how to migrate a massive, and safety critical, telecommunications infrastructure from 2000s-era tech to 2024.
Chestertons fence and strict budgetary discipline are both needed, and one cannot ignore the first, claim the second, but hope to get by on hubris.
Another domain where I see this is in mechanical engineering. Especially in the old stuff from the early 1800s. At that time every local blacksmith would have been familiar with simple machines like the lever and the pulley. But to go from that to steam engines, jacquard looms, and machine tools (machines building machines!) must have seems like magic.
It wasn't of course. Just creativity and synthesis. James Burke's Connections series is an entertaining way to pick the synthesis apart. The one on the machine loom is especially interesting since that is where punched cards came from.
OC remains superior for power users. For example, if you fly a lot, offline mode barely works on New Outlook, but is rock solid on Classic. In day to day use, the density of emails and other information on the screen is far higher in OC than New, and no amount of tweaking the view density can get New close to OC.
Classic also supports really powerful programmability, such as forms and custom views, and although I accept that the golden age of MAPI has long passed, its a shame that nothing in New comes close.
However, that speaks to Microsoft and their vision for what they call Modern Work. In this vision, it is Teams that is the center of the universe. Teams is the thing that gets the platform treatment with plug-ins and apps and polish. Outlook is relegated to something the old business people use, and as the generations shift from writing long form messages to thinking anything over a paragraph is tl;dr, I can see Outlook dying off in a few years.
Destruction is the one most people think of. The last tape is destroyed, or someone nukes everything in S3. Policies normally save this from happening, bit not always. (I note that lawyers are pretty good at saving things because of discovery, so perhaps corporations should sue themselves more often /s)
Discoverability is when the data is there, but noone knows where it is. This is your classic corpo Sharepoint/teams/slack problem. There was a chat, once, with some shared files, from their onedrive, but the last person who remembered where was laid off a year ago and...well, that data is effectively lost. I dont think we have great solutions here except writing even more things down and hoping that everyone follows rules like keeping all files in one place
The last case is where the data is located, but noone can vouch for it any more. Internal wiki pages are notorious here. People post about, say, some sharp edge they uncovered, and the page sits there for all eternity. You discover it years later. Can you trust it? Is there anyone who knows why the page still exists? It blocks the synthesis of what you know with what the author knew.
reply