2018-01-17

Universal Paperclips

Having discussed the pokey thing for iPhone SIM trays some evil person pointed me to a link for Universal Paperclip... http://www.decisionproblem.com/paperclips/



It starts off very simply, you have some wire and you can click "Make paperclip" and it makes one, using some of the wire. I think one inch of wire is one gram is one paperclip.

But it soon gets interesting...
  • Computing resources to invent new things
  • Paperclip making machines
  • Improvements to machines
  • Investing money in stocks
  • Quantum computing addition (not 100% sure I got the hang of that one entirely)
  • Improving marketing
Before long you are making a lot of paperclips, and one of the key metrics is how much you sell them for. The game as a whole is a game of balancing various metrics (e.g. price of paperclip), buying resources (e.g. wire, machines to make paperclips), and step changes (upgrading technology). Getting the balance right - when to invest in machinery to increase speed of production, etc. is tricky.

The one criticism of this early stage is that the machines get exponentially more expensive, which makes no sense.

Then it really gets out of hand...
  • Ability to convert any matter in to wire
  • Drones to collect matter
This is more balancing of the various stages of production - you have to invest in drones to collect matter, and to convert to wire, and factories to make wire in to paperclips. You keep all in check - no point wasting resources making wire faster than you can use it, well, it accumulates as unused wire, so no harm, but better spending on using it faster instead.

Basically, you end up converting all matter on Earth in to paperclips.

Just when you think you are done, you can go for space travel - making probes and drones, and, well, it really gets out of hand and there is a fun surprise in the middle of that.

At this stage we lose some of the silly exponential costs, and this allows the true exponential growth. The balancing is more fun as you have limited options to control how the probes act, including how they replicate, and create drones and factories. That can be quite hard to manage, and there are extra stages to allow more controls as you go.

So this is where it gets properly exponential, and if you do it right you get to the point where you have more probes and drones than there are IPv6 addresses! Clearly at that point it should all fall apart :-)


The game does end in a very pleasing way, you assimilate all matter in the known universe, then you can dismantle the drones and probes and factories and computing resources, and finally end up with 100 inches of wire and a "Make paperclip" button just as you started - no other controls, and those 100 inches of wire make the last few paperclips to finally reach the end...


Games that have a hard end can be disappointing in many ways, but the way this was done was actually quite satisfying, making that last paperclip manually with a click of that button just as I made the very first paperclip. Well done adding that finishing touch...

It would be fun, two machines side by side, having a race with someone - but would take many hours.

Obviously, when you think about it, the paperclips are just a metaphor for replicator blocks, and these are the probes... Only in this case we, the replicators, win :-)


20 comments:

  1. Rev, completely off topic but maybe something you can answer in a future blog post.

    I'm one of the lucky few who gets a true fiber service (fttp) provided by yourselves over openreach fiber.

    Go back a couple of months, BT were advertising this as up to 330mbit, but today they announce new fttp packages with the top speed just 314mbit.

    Now, seeing as how only ~250,000 homes can get fttp at present, what is openreach/by doing so wrong that on technology that everyone else is able to offer 1G+, they are already dropping the top speed? Presumably the only reason they're dropping it is to do with the top x% of customers speeds etc...

    And do they need a guaranteed minimum of 100mbit? Is this just clever marketing w*nk as they know fttp should never drop this low if configured for 314?

    ReplyDelete
    Replies
    1. wow, off topic, but OK. The issue is getting fibre in the ground, obviously, and BT are using a PON system (shared fibre), which makes a lot of sense in many ways - do people need 1G? However, from an A&A point of view we want our 10Gb/s+ LNSs before we start doing 330M links (let alone 1G links). It is being worked on our side anyway.

      Delete
    2. Sorry!

      I know you guys are working on it, and I'm in no rush. 80 is plenty for me.

      I'm just suprised that they're already cutting their estimated top speed when it's at such an early phase. It doesn't bode well for the future. You would have thought that if this is meant to be "future proof" they would be doing everything they can to stop it going the same way as fttc. (Started at 80mbit but now being advertised as low as 67!)

      Delete
    3. BT and A&A broadband checkers both say I can get BT Ultrafast Grade 2 ie. the faster G.Fast, and predict speeds of 201 to 330mbps down and 33 to 50mbps up. Experience on VDSL suggests I will get towards the higher end of the speed estimates. I'm guessing from your reply above that I won't be able to order this from A&A any time soon, which is a shame because I can order it from BT today.

      Delete
    4. I am not sure where we are with G.fast as present, but obviously we are also keen to have our 10Gb+ LNSs in place. Obviously we could do services with capped speeds and we may do a trial.

      Delete
    5. The FTTC speed adverts are generally down to changes in ASA / OFCOM rules and not a change in technology at all. 80M was always the top speed possible, and still is, and many lines get that speed. It is just that the public seem not to understand that "up to 80M" means "a speed not greater than 80M, and could be any lower speed, even 1M and still be what we sold you". IMHO the "fix" is not what they have done, which is lower the figure. Many people will still say they are not getting what they paid for. Obviously FTTC speeds will have changed due to things like cross talk, but that is not why the adverts have changed.

      Delete
    6. Hadn't read up on PON, I wasn't aware it was shared last mile. Presumably this would be for large scale deployments, rather than FTTPoD?

      Interesting - all ISPs may end up with the same contention ratio if they use BT Fibre... Any idea what the typical contention ratio is?

      Delete
    7. Contention ratios are not really that meaningful/useful to be honest, and I don't know. But yes, shared. My understanding if FTTPoD is simply normal FTTP working in same way but demand driven, that is all.

      Delete
    8. Continuing the off topic...

      Once you have the 10G LNS's will you then offer L2TP at higher speeds (no restrictions) ?

      Delete
    9. I'd say they're more meaningful if you have a smaller "cell" of contention.
      For A&A I can understand you provide a small faction of the theoretical maximum, but that's aggregated across the country and you have bandwidth limits to prevent people running their lines at 100% and being unsociable.

      But on a small cell, if I buy gigabit, and my neighbor does the same, with an unlimited supplier and it's shared fibre with gigabit capacity - then I'm not a happy bunny.

      Delete
    10. If A&A do a G.Fast trial count me as interested. But I already get 79.6 down and 19.9 up on VDSL so G.Fast capped at 80/20 won't do anything for me. The caps would have to be higher than that to interest me. I am more interested in upload than download speed, I'd go for downstream capped at 80 or 160 with upstream uncapped for example.

      Delete
    11. We are looking in to it, and yes, we would not cap the upload in such cases.

      Delete
    12. I'm of the opinion that the reduced top speed here is nothing to do with not being able to achieve 330 with FTTP, but what is achievable on G.Fast, and a desire to keep them both as one technology agnostic product bundled together.

      Delete
  2. You might like SpacePlan: http://jhollands.co.uk/spaceplan/ (you want the demo prototype, not the purchasable game). The mechanics are less interesting, but it's extremely polished, lots of fun, and has a definite story with a great ending.

    ReplyDelete
  3. Not sure if you're aware of the inspiration for the game:
    https://wiki.lesswrong.com/wiki/Paperclip_maximizer

    ReplyDelete
  4. I had someone give me the link to this a couple of months ago. A highly addictive and thoroughly unproductive way of spending quality time

    ReplyDelete
    Replies
    1. I spent nearly a whole day one it, though mostly just in the background whilst doing other work. The end is quite well done though.

      Delete
  5. I mentioned this game to someone and in return they recommended this:

    http://clickingbad.nullism.com/

    ReplyDelete
  6. Well that's 5h13m of my life I won't get back again :-))

    ReplyDelete

Comments are moderated purely to filter out obvious spam, but it means they may not show immediately.