I can't be the only one who saw this and thought, well sure a 15 year old can make a better faster 3D printer. Most of the reason I haven't bought one myself has been how underwhelmed I have been by the quality and results.
The only ones where I thought there precision and accuracy were useful were the UV/Near-UV plastics that operated on DLP lithographic principles. The consumables for those had too narrow of a usage range and ridiculous cost scales. The Makerbot and similar ABS extrusion machines are just dog slow, and get slower with increased complexity, not to mention consumables in the still silly price range.
Building something better and faster is easily within reach, and most of the commercial $10-100k models overcome all of the weaknesses of hobbyist / consumer offerings. Bringing some of the commercial speed optimizations (multiple nozzles) and such is trivial given the weak position of the hobby market hardware.
The upside to a certification program, even if privatized, allows the assignment of liability which supersedes the bullshit in the EULA. The idea is to create a "seller beware" instead of "buyer beware" market and to empower the consumer in such a way as to scare off the majority heap of charlatans. Further the cost of entry of certification clears a lot of that out anyway. Certainly the spam crap apps out of China and India.
It also gives a registered address to send the legal summons and other such when someone does end up producing dangerous crap.
All of the butthurt that privatized certification is useless comes from people who don't understand how awful things would be in industries so regulated otherwise.
Only because the market has developed around boat owners who all have trucks and trailers. In places where this isn't the case, dry and wet docks are plentiful and affordable. How much better would it be to have a service come pickup the boat and drop it in the lake for you and by the time you get there Saturday morning it is fueled and ready to go.
The post WWII rugged, independently capable of all feats American has been a horrible target for the populous. Weird given that WWII was the epitome of teamwork and interdependence in all things. Really the only distinguishing feature from today's specialization to the point of incompetence 2 degrees to the left or right, is entirely the idea of having perspective enough to go grab the right person for a given job or realize one must imitate it for a bit.
I love trucks, but I f***ing hate all the guys who drive one for status or compensation. I would never daily drive a truck for commuting (long term). It's a working class tool that should look well utilized. If you like trucks for show, get one for show and drive a midsize car for commuting. You can pay for it with the gas savings.
At least the sports car guys can claim to actually drive their cars in the intended fashion on a regular basis, and put money back into society with speeding tickets. Maybe we need a fine for driving down a highway every day with an empty truck bed.
No, no battery technology has not gone significantly further. You are confused with CPU / RAM / Flash trends in the mobile market which have markedly improved battery life (e.g. Surface Pro 1 vs 2: Ivy Bridge vs Haswell - battery didn't change). We're still on Li-ion variants and will be for some time.
Tesla building there own battery plant allows some optimization of form factor and cell packaging which will come out to appreciable but nothing huge in weight savings and increased safety.
A bigger portion with aluminum is the stamping / machining / welding / inspection / prep & paint steps are all massively more expensive. Aluminum parts have higher manufacturing fail rates and you run the machinery way slower because the metal is more sensitive.
They might be able to go with aluminum skin and select non-structural parts with a steel body. Several Japanese and a few European luxury cars use this technique. The Infinity G / Q series uses aluminum hoods (and skin?) with steel bodies for example. Structural aluminum is just pricey and difficult to work. From a material science / engineering standpoint, aluminum always fatigues to failure by definition; so designing, building, and validating aluminum parts takes a lot more time and expense.
Linux is just as crashy as Windows. Sure that means about a 100x decrease in frequency from the 90's, but it's still absurdly buggy and subject to the constant patch cycle bullshit. That said, it's fine as an isolated from the ECU / BCM as an infotainment system. Heck, it can even control the A/C for all I care as long as it never hooks as software into the ECU (a hotline to tell the ECU to engage the A/C clutch is fine).
Let's keep automotive ECU systems in the stone age with assembly or occasionally QNX. I suppose I could get behind a BSD variant if the code was stripped down enough and custom tested against fixed hardware enough.
So it sounds like it really catches the cusp of income ranges. I make an upper 5 and can pull the trigger, but this seems like a toy for the low 6 figure incomes.... A teaser for the 4k denied so long by the bulls**t adoption of 1080P in the computing space.
For gaming (not text or web) if the refresh is high enough (30 hz is not), scaled resolutions look fine. We've hit high enough resolutions where certain scaling operations just look like anti-aliasing instead of blurring.
Scaling rightfully got a bad name when it was upscaling 800x600 content to a 1024x768 or 1280x1024 17" monitor. It looked blurry. Scaling 1920x1080 to 2560x1440 on a 27" monitor looks really good. I'm more interested on the gaming side if these 4K TVs will take 1920x1080 or 2560x1440 at 60 hz and maintain refresh rate (technically if it is 120 hz it should, but I have my doubts about their scaler). Doing productivity work at full resolution would mostly be fine at 30 hz, if occasionally annoying.
Not fishy at all imho. PST files are a bitch, and a surprising number of organizations, public and private use crappy email systems with no auditing, backups, or archival systems.
This should be a throw the damn meter maids and other officers who were writing tickets for that spot in jail for conspiracy. They should have been turning in work orders for the the roads department to fix the paint arrangement.
Same fate should befall any judges who were presented with pictures of the spot as defense by people ticketed there.
Most of our government problems can be quickly remedied by apply the law equally to government agents. Arrest them regularly for fraud and conspiracy when it is easy to prove and the incentive to be corrupt goes away.
Diverse might mean all men versus men and women though. Give a similar age range, experience level, and size of group; an all white males group are unlikely to come close to a group of white males and females. Competition for mating opportunities works in the workplace amazingly well. This is fine as long as the engagement term of the group is of a length where no poisonous amorous relationships or jealousies develop... so in the software world of 3-6 month long projects, it's great.
The effects can be further exaggerated on shorter projects by using more single and unwed males and females of similar backgrounds.
The bumpy surface they have is going to be horrendous for noise pollution and suspension damage. An amorphous glass surface with a friction texturing is probably also going to cause more tire wear than concrete. They only vaguely guess at lifespan of the friction texture and don't address how it handles loading up (bits of rubber and car oil getting caught in the pores)
Further the bumpy surface will likely cost commuters more in mileage due to being a rough road than it will produce in electricity. It's an economy of scale thing where everyone loses a few tenths of a mile to a gallon, but that adds up quick over thousands of cars per day.
The implementation cost issues are being completely ignored. We use asphalt and concrete because of economies of scale. This technology is orders of magnitude more expensive per mile.
How about the hallmarks for what qualifies as dementia...
It's dementia when they are stealing your medicine.
It's sweet old lady / kind delusions when they tell you how much they love the cake you brought them last week for their birthday (who is this person? --just go with it).
An interesting remedy would be a "code view" mode for spreadsheets where calculations were displayed as nested operations and such. It would require a stronger intent manager that could recognize the same sequence of code running on rows 4-53 until column N, but it could work. Sure, this mostly sounds like a database, but instead give it the modularity and ease of current spreadsheets and everything works out.
True, but inventory management and reporting don't have any need to coexist on the same network. It's easy enough to have the POS side running on one VLAN and a one-way replication of aggregate sales numbers pushed to the inventory management and reporting side. Heck, just replicate a copy of the database with all of the customer's personal information and CC#'s stripped out.
An odd angle to why Target got hit with such a huge data loss breach was the fact that they were getting too nosy about their customer habits. They used analytics to tell when individual customers were pregnant and send targeted advertising (pun not intended): http://www.forbes.com/sites/kashmirhill/2012/02/16/how-target-figured-out-a-teen-girl-was-pregnant-before-her-father-did/
These kinds of data leaks occur because they were playing the big data game and combing obsessively over personally identifiable data repeatedly to create better consumers. How much harder is it for the admin to secure the network when the folks in marketing want giant swaths of data they lack the responsibility to handle. The data breaches at Target were easy, because it had become too much of a hassle to try to secure data half the company was rooting through. I doubt eBay was any different, and if anything had an even more haphazard attitude given their model is not be responsible for the products they sell.
All of this can be simplified by architecting purpose designed networks, and for a minimum of cost. You have a firewall (and possibly switch). There are 2 VLANS. On one (let's say VLAN 100) is the free Wifi, Pandora feed to the house audio, and internet connection at the workstations the managers blow time at. On the other (let's call it VLAN 222) are the network connections for the POS equipment. On VLAN 222, the firewall allows no inbound connections with the slim exception of VPN secured traffic. Outbound connections on VLAN 222 are restricted to OS/AV/POS update hosts on SSL or similar and CC auth processors. Generic internet access is banned on VLAN 222. The back office POS software runs in a VM that only has access to VLAN 222. The manager workstation runs the VM if necessary as well as has it's own access to the internet (if necessary). The POS terminals, even if they are those hip, all the rage, iPads, do not have internet access.
This is more or less (minus VMs, DSL, and iPads, and replace VPN with dedicated password protected dial-in) the way we designed POS security in the late 90's when I was doing POS. As far as I can tell, it is mostly PCI compliant.
The issues we're seeing is people getting all manner of malware (from pr0n/etc.) on the manager back office workstation, similar from the POS terminals, and using Logmein / Teamviewer with weak passwords on the back office server. We knew better 15 years ago, so anyone who is getting hit by such garbage is a lame hack.
The uptime from various cloud vendors is pretty poor. Sure the server is up, but some networking or SAN component is sketchy a lot more than in-house managed servers. Cases in point: 1) I've worked with several virtualized storage architectures on Amazon AWS and we've had instances lock up due to brief, hard to track down SAN drops. 2) I had a customer have to force shutdown 2 VMs in CBeyond's cloud because their SAN latency went up enough that databases started dropping offline. It took CBeyond 2 days to get their SAN back to full operational status.
I do wish the cloud providers would modify their storage model a bit. When starting an instance / VM, use the SAN to copy the whole image to an available server's LOCAL storage array. This fixes a great many latency problems and does not make the servers that much more expensive to build / operate (just a tad more storage in RAID 10 per server). The only drawback to this is for big data users who need beyond a couple dozen TB for a server in the cloud. Most of those situations are already using clustering software that is resistant to failure of a few nodes.
How about we make the copyright holder responsible for providing suitable replacements as part of their copyright renewal process. It would be preferable to require a new stamping off a master every 5-10 years and provide identical media replacements - certainly to the Library of Congress and other designated archives (CD for CD, Book for book, VHS for VHS). I could see some wiggle room where digital downloads of equal or greater quality be made available to consumers.
Even if we say fuck the consumers, the copyright holder should certainly be responsible to provide replacements to archives as part of the copyright registration. I would see such as minimal evidence for copyright enforcement.
Of those examples, they are still mostly video accelerator / transcode acceleration area, and a couple have USB 3.0 / SS versions. Outside of the die hard MBP/MP users, anyone with a non-Apple laptop who works in industries where such hardware is necessary will have a dedicated render station to run those cards. You seem to forget that a MBP is going to have CPU, RAM, and I/O buses which simply can't match a regular desktop much less server-class workstation motherboards.
The other part that you are ignoring is the fact that anyone who deals with video or CG at that level is going to need serious storage. Even a 1TB SSD option isn't going to cut it. Sure you can plug storage in via ThunderBolt, but the cost just spirals up getting all these niche parts.
So your use case still boils down to Mac Pro users, which while selling alright, comprises a smaller portion of the PC sales market than desktops with Linux pre-installed.
A surprising number of middle management type jobs exist only to prop up the cost of project and process management software that would replace them. Let that sink in for a minute.
Automation is great and all, but ultimately it already exists and is easier to implement for someone just pushing / processing paperwork and acting as a "facilitator" than it is to mechanize the process of flipping burgers. Then take a nice look at the software that is really powerful and really automates process oriented work.
One of the side effects of pricing software exclusively based upon demand (because supply is infinite and creation costs incredibly scalable) is that sometimes, demand will be artificially created. I work in IT, and it's incredibly annoying to look at how terrible Cisco's CLI and GUI configuration tools are compared to their competitors, yet Cisco certs are exceptionally valuable. By making ASDM a steaming pile of shite, Cisco has made people who can configure the relatively simple product of an ASA 55xx very valuable.
Distinctively, this is a huge issue with using FOSS software. The whole "libre" and "open" philosophies breed organizations that just let everyone do their own thing. Nobody wants to grind at the mundane crap like documentation or unified interfaces and such. And that is when there is an organization like Apache pushing the project. How much worse are small dev team projects getting work on a personal free time basis or all the extensions for various FOSS projects that get crap maintenance.
Every few years we get a trickle of horror stories about working at Microsoft, but the reality is that those horrors exist at most companies, because someone has to buckle down and do the shit work.
Slashdot Mirror
I can't be the only one who saw this and thought, well sure a 15 year old can make a better faster 3D printer. Most of the reason I haven't bought one myself has been how underwhelmed I have been by the quality and results.
The only ones where I thought there precision and accuracy were useful were the UV/Near-UV plastics that operated on DLP lithographic principles. The consumables for those had too narrow of a usage range and ridiculous cost scales. The Makerbot and similar ABS extrusion machines are just dog slow, and get slower with increased complexity, not to mention consumables in the still silly price range.
Building something better and faster is easily within reach, and most of the commercial $10-100k models overcome all of the weaknesses of hobbyist / consumer offerings. Bringing some of the commercial speed optimizations (multiple nozzles) and such is trivial given the weak position of the hobby market hardware.
The upside to a certification program, even if privatized, allows the assignment of liability which supersedes the bullshit in the EULA. The idea is to create a "seller beware" instead of "buyer beware" market and to empower the consumer in such a way as to scare off the majority heap of charlatans. Further the cost of entry of certification clears a lot of that out anyway. Certainly the spam crap apps out of China and India.
It also gives a registered address to send the legal summons and other such when someone does end up producing dangerous crap.
All of the butthurt that privatized certification is useless comes from people who don't understand how awful things would be in industries so regulated otherwise.
Girl: Did I fall asleep?
Doctor: For a little while.
Girl: Shall I go now.
Doctor: If you like.
Only because the market has developed around boat owners who all have trucks and trailers. In places where this isn't the case, dry and wet docks are plentiful and affordable. How much better would it be to have a service come pickup the boat and drop it in the lake for you and by the time you get there Saturday morning it is fueled and ready to go.
The post WWII rugged, independently capable of all feats American has been a horrible target for the populous. Weird given that WWII was the epitome of teamwork and interdependence in all things. Really the only distinguishing feature from today's specialization to the point of incompetence 2 degrees to the left or right, is entirely the idea of having perspective enough to go grab the right person for a given job or realize one must imitate it for a bit.
I love trucks, but I f***ing hate all the guys who drive one for status or compensation. I would never daily drive a truck for commuting (long term). It's a working class tool that should look well utilized. If you like trucks for show, get one for show and drive a midsize car for commuting. You can pay for it with the gas savings.
At least the sports car guys can claim to actually drive their cars in the intended fashion on a regular basis, and put money back into society with speeding tickets. Maybe we need a fine for driving down a highway every day with an empty truck bed.
No, no battery technology has not gone significantly further. You are confused with CPU / RAM / Flash trends in the mobile market which have markedly improved battery life (e.g. Surface Pro 1 vs 2: Ivy Bridge vs Haswell - battery didn't change). We're still on Li-ion variants and will be for some time.
Tesla building there own battery plant allows some optimization of form factor and cell packaging which will come out to appreciable but nothing huge in weight savings and increased safety.
A bigger portion with aluminum is the stamping / machining / welding / inspection / prep & paint steps are all massively more expensive. Aluminum parts have higher manufacturing fail rates and you run the machinery way slower because the metal is more sensitive.
They might be able to go with aluminum skin and select non-structural parts with a steel body. Several Japanese and a few European luxury cars use this technique. The Infinity G / Q series uses aluminum hoods (and skin?) with steel bodies for example. Structural aluminum is just pricey and difficult to work. From a material science / engineering standpoint, aluminum always fatigues to failure by definition; so designing, building, and validating aluminum parts takes a lot more time and expense.
Linux is just as crashy as Windows. Sure that means about a 100x decrease in frequency from the 90's, but it's still absurdly buggy and subject to the constant patch cycle bullshit. That said, it's fine as an isolated from the ECU / BCM as an infotainment system. Heck, it can even control the A/C for all I care as long as it never hooks as software into the ECU (a hotline to tell the ECU to engage the A/C clutch is fine).
Let's keep automotive ECU systems in the stone age with assembly or occasionally QNX. I suppose I could get behind a BSD variant if the code was stripped down enough and custom tested against fixed hardware enough.
Actually, the US military has already been playing with drones like this that use inductive latch hooks to recharge off the local grid.
So it sounds like it really catches the cusp of income ranges. I make an upper 5 and can pull the trigger, but this seems like a toy for the low 6 figure incomes.... A teaser for the 4k denied so long by the bulls**t adoption of 1080P in the computing space.
For gaming (not text or web) if the refresh is high enough (30 hz is not), scaled resolutions look fine. We've hit high enough resolutions where certain scaling operations just look like anti-aliasing instead of blurring.
Scaling rightfully got a bad name when it was upscaling 800x600 content to a 1024x768 or 1280x1024 17" monitor. It looked blurry. Scaling 1920x1080 to 2560x1440 on a 27" monitor looks really good. I'm more interested on the gaming side if these 4K TVs will take 1920x1080 or 2560x1440 at 60 hz and maintain refresh rate (technically if it is 120 hz it should, but I have my doubts about their scaler). Doing productivity work at full resolution would mostly be fine at 30 hz, if occasionally annoying.
Not fishy at all imho. PST files are a bitch, and a surprising number of organizations, public and private use crappy email systems with no auditing, backups, or archival systems.
This should be a throw the damn meter maids and other officers who were writing tickets for that spot in jail for conspiracy. They should have been turning in work orders for the the roads department to fix the paint arrangement.
Same fate should befall any judges who were presented with pictures of the spot as defense by people ticketed there.
Most of our government problems can be quickly remedied by apply the law equally to government agents. Arrest them regularly for fraud and conspiracy when it is easy to prove and the incentive to be corrupt goes away.
Diverse might mean all men versus men and women though. Give a similar age range, experience level, and size of group; an all white males group are unlikely to come close to a group of white males and females. Competition for mating opportunities works in the workplace amazingly well. This is fine as long as the engagement term of the group is of a length where no poisonous amorous relationships or jealousies develop... so in the software world of 3-6 month long projects, it's great.
The effects can be further exaggerated on shorter projects by using more single and unwed males and females of similar backgrounds.
The bumpy surface they have is going to be horrendous for noise pollution and suspension damage. An amorphous glass surface with a friction texturing is probably also going to cause more tire wear than concrete. They only vaguely guess at lifespan of the friction texture and don't address how it handles loading up (bits of rubber and car oil getting caught in the pores)
Further the bumpy surface will likely cost commuters more in mileage due to being a rough road than it will produce in electricity. It's an economy of scale thing where everyone loses a few tenths of a mile to a gallon, but that adds up quick over thousands of cars per day.
The implementation cost issues are being completely ignored. We use asphalt and concrete because of economies of scale. This technology is orders of magnitude more expensive per mile.
How about the hallmarks for what qualifies as dementia...
It's dementia when they are stealing your medicine.
It's sweet old lady / kind delusions when they tell you how much they love the cake you brought them last week for their birthday (who is this person? --just go with it).
An interesting remedy would be a "code view" mode for spreadsheets where calculations were displayed as nested operations and such. It would require a stronger intent manager that could recognize the same sequence of code running on rows 4-53 until column N, but it could work. Sure, this mostly sounds like a database, but instead give it the modularity and ease of current spreadsheets and everything works out.
True, but inventory management and reporting don't have any need to coexist on the same network. It's easy enough to have the POS side running on one VLAN and a one-way replication of aggregate sales numbers pushed to the inventory management and reporting side. Heck, just replicate a copy of the database with all of the customer's personal information and CC#'s stripped out.
An odd angle to why Target got hit with such a huge data loss breach was the fact that they were getting too nosy about their customer habits. They used analytics to tell when individual customers were pregnant and send targeted advertising (pun not intended): http://www.forbes.com/sites/kashmirhill/2012/02/16/how-target-figured-out-a-teen-girl-was-pregnant-before-her-father-did/
These kinds of data leaks occur because they were playing the big data game and combing obsessively over personally identifiable data repeatedly to create better consumers. How much harder is it for the admin to secure the network when the folks in marketing want giant swaths of data they lack the responsibility to handle. The data breaches at Target were easy, because it had become too much of a hassle to try to secure data half the company was rooting through. I doubt eBay was any different, and if anything had an even more haphazard attitude given their model is not be responsible for the products they sell.
All of this can be simplified by architecting purpose designed networks, and for a minimum of cost. You have a firewall (and possibly switch). There are 2 VLANS. On one (let's say VLAN 100) is the free Wifi, Pandora feed to the house audio, and internet connection at the workstations the managers blow time at. On the other (let's call it VLAN 222) are the network connections for the POS equipment. On VLAN 222, the firewall allows no inbound connections with the slim exception of VPN secured traffic. Outbound connections on VLAN 222 are restricted to OS/AV/POS update hosts on SSL or similar and CC auth processors. Generic internet access is banned on VLAN 222. The back office POS software runs in a VM that only has access to VLAN 222. The manager workstation runs the VM if necessary as well as has it's own access to the internet (if necessary). The POS terminals, even if they are those hip, all the rage, iPads, do not have internet access.
This is more or less (minus VMs, DSL, and iPads, and replace VPN with dedicated password protected dial-in) the way we designed POS security in the late 90's when I was doing POS. As far as I can tell, it is mostly PCI compliant.
The issues we're seeing is people getting all manner of malware (from pr0n/etc.) on the manager back office workstation, similar from the POS terminals, and using Logmein / Teamviewer with weak passwords on the back office server. We knew better 15 years ago, so anyone who is getting hit by such garbage is a lame hack.
This is how it already works. It's a 3gpp or similar high compression digital codec running over IPv6. This is all just marketing fluff.
The uptime from various cloud vendors is pretty poor. Sure the server is up, but some networking or SAN component is sketchy a lot more than in-house managed servers. Cases in point:
1) I've worked with several virtualized storage architectures on Amazon AWS and we've had instances lock up due to brief, hard to track down SAN drops.
2) I had a customer have to force shutdown 2 VMs in CBeyond's cloud because their SAN latency went up enough that databases started dropping offline. It took CBeyond 2 days to get their SAN back to full operational status.
I do wish the cloud providers would modify their storage model a bit. When starting an instance / VM, use the SAN to copy the whole image to an available server's LOCAL storage array. This fixes a great many latency problems and does not make the servers that much more expensive to build / operate (just a tad more storage in RAID 10 per server). The only drawback to this is for big data users who need beyond a couple dozen TB for a server in the cloud. Most of those situations are already using clustering software that is resistant to failure of a few nodes.
How about we make the copyright holder responsible for providing suitable replacements as part of their copyright renewal process. It would be preferable to require a new stamping off a master every 5-10 years and provide identical media replacements - certainly to the Library of Congress and other designated archives (CD for CD, Book for book, VHS for VHS). I could see some wiggle room where digital downloads of equal or greater quality be made available to consumers.
Even if we say fuck the consumers, the copyright holder should certainly be responsible to provide replacements to archives as part of the copyright registration. I would see such as minimal evidence for copyright enforcement.
Of those examples, they are still mostly video accelerator / transcode acceleration area, and a couple have USB 3.0 / SS versions. Outside of the die hard MBP/MP users, anyone with a non-Apple laptop who works in industries where such hardware is necessary will have a dedicated render station to run those cards. You seem to forget that a MBP is going to have CPU, RAM, and I/O buses which simply can't match a regular desktop much less server-class workstation motherboards.
The other part that you are ignoring is the fact that anyone who deals with video or CG at that level is going to need serious storage. Even a 1TB SSD option isn't going to cut it. Sure you can plug storage in via ThunderBolt, but the cost just spirals up getting all these niche parts.
So your use case still boils down to Mac Pro users, which while selling alright, comprises a smaller portion of the PC sales market than desktops with Linux pre-installed.
A surprising number of middle management type jobs exist only to prop up the cost of project and process management software that would replace them. Let that sink in for a minute.
Automation is great and all, but ultimately it already exists and is easier to implement for someone just pushing / processing paperwork and acting as a "facilitator" than it is to mechanize the process of flipping burgers. Then take a nice look at the software that is really powerful and really automates process oriented work.
One of the side effects of pricing software exclusively based upon demand (because supply is infinite and creation costs incredibly scalable) is that sometimes, demand will be artificially created. I work in IT, and it's incredibly annoying to look at how terrible Cisco's CLI and GUI configuration tools are compared to their competitors, yet Cisco certs are exceptionally valuable. By making ASDM a steaming pile of shite, Cisco has made people who can configure the relatively simple product of an ASA 55xx very valuable.
Distinctively, this is a huge issue with using FOSS software. The whole "libre" and "open" philosophies breed organizations that just let everyone do their own thing. Nobody wants to grind at the mundane crap like documentation or unified interfaces and such. And that is when there is an organization like Apache pushing the project. How much worse are small dev team projects getting work on a personal free time basis or all the extensions for various FOSS projects that get crap maintenance.
Every few years we get a trickle of horror stories about working at Microsoft, but the reality is that those horrors exist at most companies, because someone has to buckle down and do the shit work.