Jump to content


  • Content Count

  • Joined

  • Last visited

  • Days Won


Sir_Substance last won the day on July 29 2017

Sir_Substance had the most liked content!

Community Reputation

13 Serf

About Sir_Substance

  • Rank
  1. Fair enough. I don't consider what's currently available to be "self-driving" in any meaningful sense. The tipping point at which the interesting stuff will happen is the point where insurance companies start offering to ignore your past driving history if you buy specific models of car. That'll be the starting gun indicating that the actuaries who are running the numbers have decided that robots are better than humans at driving. Personally, I rate Teslas autopilot feature similar to how I rate their "dancing car" feature. Gimmick and maybe kinda neat under some circumstances, but ultimately not very significant.
  2. I certainly consider it to be the logical development once there's a critical mass of self-driving cars on main roads. It'll be much cheaper than re-designating some roads as self-driving only, so it'll get government support provided it can be done accurately enough. Yes, I think it'll happen to motorbikes They're not actually allowed to sneak up the middle of lanes in most places anyway, and they'll slow down all the traffic behind them at the lights because they won't be able to accelerate in a single unified block with the 50 networked self-driving cars behind them. Cyclists are harder to deal with because they're so much slower than cars, and not as well equipped. It's pretty easy for car manufacturers to program the scar tissue planner so that if a human-driven car doesn't indicate, it won't be given a slot to cross traffic, which would enforce good behavior on human drivers. You could *try* that with arm-signalling with bikes but it's a harder computer vision problem to solve and not everyone has two arms. I don't have a good prediction for that one, other then probably self-driving cars avoiding lanes with cyclists in them entirely.
  3. Sir_Substance

    After 17 years, it's time to look back...

    I'm actually still on there all the time, I think I set it to invisible 6 months ago and never unset it
  4. That's where we disagree, I don't see that practical need. Self-driving cars are going to push out human drivers, so it doesn't matter if the human drivers feel the traffic is going "too slow". Insurance companies won't want to cover humans who might run red lights at 100km/h when they could cover robots that never do, and so increasing insurance costs will marginalise humans. On top of that, the ability to time-share a car between 3+ people and have it autonomously route between them will rapidly speed the adoption of self-driving cars once they become available, with one obvious result: Self-driving cars will form "scar tissue" around human drivers on main roads to prevent them from messing with the traffic flow. When you drive in a traditional car, you'll get gently and tactfully boxed in by the three nearest self-driving cars, they will watch your movements and indicators and move with you to let you get where you indicate you want to go, but they will body-block you from running a red light at 100km/h or even getting to 100km/h, so that other cars can be more certain of your behavior. And yeah, that means those self-driving cars are going to force you to drive at 15km/h if they're not sure you can drive faster without causing an accident.
  5. Sir_Substance

    After 17 years, it's time to look back...

    Hrm, is my memory failing me, or do I remember HentaiBob eating the ultra-spicy megadeath burger at the maid and magpie without using the provided gloves, touching his face and immediately bellowing in pain, and then shoving a small tub of yogurt into his eye? Good times :D
  6. This is a beloved ethical question, but it's actually not a realistic one. Who is responsible if a driverless car runs over and kills a person? Definitely the company that made it will be found liable, for one specific reason: Driving faster than your braking distance is a strictly human problem. Self-driving cars _will not_drive faster than their confidence speed. The confidence speed will be the speed at which the car is certain that no pedestrian it can currently see can manage to jump in front of it before it can stop, and no pedestrian who might be hiding in a shrub out of LIDAR view could jump in front of the car. Critically, that means that whenin CBD's, down streets with lots of hedges, going around blind corners and in any other situation where visibility is compromised, self driving cars are just not going to exceed 15km/h, which is around the speed that a car with a reaction time of almost zero can confidently drop below injury speed from at short notice. That sounds like it'll make self-driving cars uncompetitive with human-controlled cars, but there's a few mitigating factors. 1. Insurance on human cars will go through the god damn roof 2. Self driving cars will eventually (and it'll take years from when the first self-driving cars hit the road) start sharing networked sensor information to help mitigate the problems with blind cornders 3. Self-driving cars will /really quickly/ start sharing map data about areas with shit visibility or lots of pedestrians and avoid them where possible 4. You won't care if your commute through the CBD takes 15 minutes longer if you can be eating breakfast and reading the news while it happens 5. Self-driving cars can all accelerate as one block when the traffic lights go green, and you get about four as much traffic (IIRC) through each light change if you can do that, which means what self driving cars lose while driving near pedestrian zones they make up in spades during light changes But yeah, this notion that self-driving cars might end up in a situation where they can't stop in time and have to choose someone to kill? Humans suffer from impatience, robots do not. Robots just won't ever drive fast enough to allow that situation to be possible.
  7. Sir_Substance


    Keenly awaiting the promised linux port :3
  8. Sir_Substance

    What happened to the good space games?

    I keep looking at the Steam page for this, and not buying it. :/ Screenies look really nice, but the reviews are very polarised: one side loves it, and the other... eg: http://www.quartertothree.com/fp/2016/05/18/bone-dry-sci-fi-stellaris-game-doesnt-even-work/ I'll pick it up in some insane sale. I'm not sure I really rate that guys criticisms. I've put more time into stellaris than almost any other game I own. I'll respond to his two main criticisms: 1. The sector system increases late game micromanagement instead of reducing it. Bull. Shit. I raise my eyebrow that he can assert he loves Paradox's historical games, and then trot this line out. Comparing late game EU micromanagement to this is utterly night and day. However, it's a bit of an irrelevant argument because it's invested in the notion that micromanagement must be objectively bad, but the reality is that micromanagement is a game mechanic. What matters is whether you enjoy the implementation, and I think the Stellaris system hits a happy medium that would be about right for quite a lot of people. 2. The storylines are dull and the game has no soul as a result. I know what he's getting at, but I think he hasn't delved into the game very deeply if he thinks that. The game basically has two types of events. There's little mini-events that happen as you explore the galaxy. Because galaxies are big and you're going to play lots of games in big galaxies, these events repeat a lot. It's basically the same problem as the NPC's in TES: Oblivion repeating the same dialog all the time. However, there's a second layer of longer, more detailed and way more interesting stories buried in there that you're much less likely to come across (horizon signal being one of the best, but far from the only one), and I've even seen two separate storylines unexpectedly interact. I still don't know if it was a bug or deliberate, but it sure was interesting. There's depth there, but it's not spoon-fed to you. Spend a game digging into the universe instead of conquering it, and you'll find some interesting stuff. I'd also say that it's been a great game for roleplaying for me. The species creator is extremely versatile, and I've had a lot of fun playing everything from short-lived temperamental space-nazi penguins to dull-witted, pacifistic psychic space cacti with a fetish for government paperwork. If you can invest in your own stories a little it's a fun playground. tl;dr I think his review is pretty shallow. The main reason I would suggest holding off on buying it is that paradox has been muttering for a while about making it mandatory to have a paradox account to play MP. I'm a big hater of forced accounts for things which have no technical need for it. Depending on how you feel on the matter, that may or may not be a factor for you.
  9. Sir_Substance

    Home Server advice

    I run a home server setup using a NUC and a cheap seagate nas that supports NFS. the NUC runs proxmox (an open source hypervisor) and I set up all my services as VM's on that, with the drive images storage on the NAS. I reverse proxy internet traffic to the other VM's via an NGINX VM that also terminates the SSL for my websites, with unencrypted traffic running over the internal network. There have been a few interesting ecosystem improvements since then, notably caddy is probably a better choice than nginx for this purpose now, and some of the other hypervisors are starting to grow hypervisor level docker support. I'd be happy to elaborate on details in a PM, if you're interested. If you're mostly looking at this for a content server, there are a few raspberry-pi sized devices with a bit more kick, like the banana pi or the orange pi. I don't have much experience with them, but I do know that at least the raspi 1 and 2 units are pretty marginal on having the computing power needed to run a media server.
  10. Sir_Substance

    Same-sex postal survey is a go

    The form itself, sure. I was talking about the envelope - there are no regulations about that. Also, unless they're doing special snowflake rules for this vote, that's not correct. During an election when the volunteers count the vote they are looking for clear intent. In a ballot with 4 candidates, it is sufficient for you to number 1, 2 and 4. Given that there are only 4 candidates, the counter knows who you intended 3 to be, because you marked all the others, so there's only one possibility. If there were 5 people on that ballot, it would become an informal vote, because there's no way for the counter to know who you would pick as 3, and who as 5. So when it comes to marking your ballot to hide it from being observed by third parties, by standard AEC practice there's no problem with that as long as it's clear to the counter which option you selected. Feel free to draw as many butterflies, dicks or black squares on the ballot as you like, just don't scribble over the boxes on the voting side.
  11. Sir_Substance

    What happened to the good space games?

    <sigh> That hory old chestnut again. I realise that he tells you what you want to hear, working for the competition and all, but you really should check your sources. Derek Smart is a fuckwit. Correction 1: My doubts about SC well predate my starting to work for CCP, and stem from a professional understanding of the scale of the work they've promised to their backers. It's a bit cheeky of you to imply otherwise, I'm all but certain we've had discussions on this over beer in the past. Correction 2: Derek Smart is not the source of this particular sign, and it can hardly be called an old chestnut when it's from 3 months ago. The source is a pair of documents filed with the UK government: https://beta.companieshouse.gov.uk/company/08703814/charges/AHMo7d0tVN50wGM-FC6tbyhYlss https://beta.companieshouse.gov.uk/company/08815227/charges/W3FsufjDb8gTRZZqoCvnEAJKzSk In both cases you can see that the agreement includes a "Floating charge covers all the property or undertaking of the company.". You can read more about that here: http://www.companylawclub.co.uk/fixed-and-floating-charges The short version is that SC has taken out a loan, and secured it against 100% of all their assets and IP, worldwide. Think about that for a second. Star citizen collected more than 100 million dollars from crowdfunding sources. That's obligation free cash in hand. They're now taking out a loan that comes with obligations, and to get that loan they've bet the entire farm, including the remaining money from the crowdfunding campaigns. Check sections 4.1.8 (Foundry42) and 4.2.8 (Cloud Imperium Games), they've staked "the Accounts (including all monies standing to the credit of each Account, all interest accrued on each Account and all debits represented by the foregoing)". I don't wish SC ill, I've got a small ship of my own in a digital hangar somewhere I'd like to take for a spin some day. But you've got your head in the sand if you think the signs are good. Right out the gate, they made a tonne of promises about what the game would look like and what features it would have well before they'd even done enough technical exploration to know what they were really capable of as a company, let alone how far they could stretch the engine. There was a lot of questioning about the organizational challenge they were taking on, but they had all the money in the world, and as long as they had money they had all the time in the world. People figured there was a chance they might just about be able to pull it off. But now it looks pretty damn suspiciously like the money is gone, and if true that means the time is gone too. I hope they pull through, and either way it'll be a hell of a case study to read. But SC is clearly more in peril than it has ever been. We might not know how in peril they are, maybe not that much in an absolute sense, but their company position today is quantifiably worse than their company position a year ago.
  12. Sir_Substance

    Post Your Latest Real Life Purchase!

    Actually, thus far I've mostly used my bitcoins for buying games on steam, it's had native support for a year or two now.
  13. Sir_Substance

    Post Your Latest Real Life Purchase!

    Arrived a few days ago, it's quite a nice little object and the web UI for it is pretty great. Counts as a purchase, although it hasn't arrived yet: I've totally drunk the nim coolaid. It's not quite at the point where I can use it at work, but I have high hopes for it as a compiled python or a more friendly go.
  14. Sir_Substance

    What happened to the good space games?

    Be careful about pinning too much some on SC, the signs are not exactly rosy at the moment: http://dereksmart.com/2017/06/star-citizen-final-countdown/ Beyond star citizen, I don't really know why the space genre is such an under-served niche. You should know that x3: reunion now has four spinoffs/expansions/remakes/I don't really know what the deal is there. Never the less, there's been X3: Albion Prelude, X3: terran conflict and X Rebirth (and friends?) since then. You might want to take a closer look at one of the more recent entrants in the X universe. I have absolutely no idea if they're any good or not, but if you haven't looked, you might be missing something.
  15. Sir_Substance

    Same-sex postal survey is a go

    You're all very pessimistic, I suspect it'll be ~70% yes. Correct me if I'm wrong, but because of the way this is being run, voting isn't mandatory? So it's about which side of the two extremes is less apathetic, and I'd wager the liberals are more energetic than the conservatives. This is another reason I think we'll see a clear majority. Truthfully, I don't really care that much about gay marriage as an issue. That might be hurtful to some people for whom this is their whole world, but no one can care about everything, and this is a non-issue for me. As such, I'm inclined to vote yes because it seems like the speediest way to get parliament to stop tying itself in knots over the damn thing and move on to real issues. I think we'll see a protest turnout from people who would otherwise be on the fence and maybe voting no.