Arcanaville

Token code generators are currently used in many business banking situations where a single login can authorize huge monetary transfers, such as payroll transfers. If the generators were cracked in any way but the trivial (i.e. stealing the token, disabling the requirement for a token at the server-side, or stealing the token seed records) I would know about it.

"Everything is breakable" is just as silly a position as "my security is unbreakable." There's breakable in theory, and there's breakable in practice. At the moment, I have no problem stating that token code generators and challenge-response tokens are, if implemented cryptographically correctly, impervious to password harvesting attacks. The level and nature of the skill required to break the tokens through monitoring alone is so high its literally of national security significance.

To be honest, I'm actually less concerned about the fact that the maps are predictable, than I am that 95% of the time the maps make no sense.

And you can't say that the reason why they make no sense is because the point is to force players to explore. In old school PnP games there were two kinds of maps. The random, twisty, explorey maps where the point was to explore and there were actual things to find. And regular buildings, like castles, which at least attempted to look like functional structures. Most of the maps are random twisty maps just to require players to navigate random twisty maps. They usually don't have a spectacular pay off room that rewards explorers, and they usually imply that the Rikti killed all the architects in the first Rikti War.

I would occasionally like to enter an office building that doesn't look like a scene from Being John Malkovich. And to be honest, when the building layouts are actually logical, it actually makes players *think* more: if this is the lobby, does that mean the elevators are that way; if I want to get to the storage rooms, would they be in the back of the building? Random layouts just make me think "run real fast and kill everything." I rarely even have the sense of being in a building so much as a concrete hedge maze with elevators.

And it isn't that I'm unaware of *all* the game implementation issues here when I say this. But I will feel slightly disappointed if Going Rogue shows me amazingly detailed architecture rendered in stunning Ultra Mode, and the inside of every single building looks like the corporate edition of the Winchester Mystery House.

Try to have it done in time for the GR launch, D_R.

There are three things to consider in this circumstance:

1. Motherboard memory controllers do tend to reduce their speed to the slowest RAM you have. However, its not true that smaller RAM modules are necessarily slower. What matters is the clocks speed of the modules (and to a different extent, the timing on the modules).

2. Some motherboards have memory designs in which memory banks are interleaved for higher performance. Basically, memory is read from two different banks of memory in alternating fashion, like reading two pages of a book simultaneously. If you do not install memory in pairs, interleaving is turned off and memory can be slower. In some more recent motherboards, you have to install memory in threes because the motherboard has three memory channels (if you install less than triples, it'll still work but slower).

3. No matter how much your memory configuration speeds up or slows down relative to the optimal performance of the memory, its all small percentages relative to swapping. Hard disks are a thousand times slower than the slowest memory you can put into your system, so if you are running out of memory when running the apps you want to run, more memory is better period. It doesn't matter how fast or slow it is. If the difference between the best memory configuration and the worst memory configuration is like the difference between having your file cabinets on the side of your desk or behind your chair, swapping is like having to drive back home to get it.


I would put all four sticks in. For Windows, 3gigs is almost always better than 2gigs. I would only start caring about the memory configuration when I reached 4 gigs of RAM (in the sense of wondering if I should add the RAM or not).

I still think my solution to this problem is the best one. Its I suppose a variant of your fourth option, and I've been suggesting it for about as long as the veteran reward system has existed. Make veteran rewards (at least all the ones besides the generic ones, like earning merits or a costume token) into choice tokens where you can choose a reward from a specific pool of rewards. Then add tokens at regular intervals, and add additional choices to the pools at the same time.

So perhaps at 15 month intervals everyone would get a melee attack veteran reward token which they could use to get a veteran melee attack. But at the beginning of time, perhaps there were only two such attacks, and you could pick one of them. Down the road, when the 30th month reward came along, instead of making a whole new reward unavailable to new players, you make it *another* melee attack reward token and add one or two new attacks to the veteran melee attack pool. So now the pool has four attacks and 30 month veterans get two of them. But the 15 month veterans that reach 15 months at that time will still have all four powers available to them; they just only get one choice to start.

If we do the same thing to veteran ranged attacks, costume parts, and all the other "unique" vet rewards, then it will always be true that the longer vets have *more* rewards, but all new players will have *access* to any of them in relatively short amounts of time. You are never introducing a reward that only six year veterans have any chance of getting (usually: this doesn't preclude that option if you really want to), but you can still introduce new rewards and you can still reward longer term veterans.

I don't like puzzles that are too easy to simply hint-guide away in the general case, although I recognize that its non-trivial to make puzzles that can't be.

I do believe that MMOs provide an escape hatch to puzzles that an actual puzzle game doesn't. MMOs are balanced based on reward *rates*, not actual rewards. Developers don't *like* "riskless rewards" but in actuality if you achieve a "riskless reward" in an MMO and the time it takes to achieve it is actually much higher than the average time to earn that level of reward any other way, you're actually underperforming.

This means its actually possible to *codify* the amount of time a puzzle is allowed to "stump" a player in an MMO before its reasonable to allow them to brute-force a solution by some method. In fact, brute force itself should probably be the actual escape hatch. Puzzles in MMOs can be used to offer "reward bonuses" to players that solve them quickly, and "reward penalties" (really, a slower earning rate) to players that don't, in a way that doesn't require guides.

But actually, if you adopt this sort of gaming theory, its actually more imperative that you make puzzles very difficult to strategy-guide away, because if its trivially easy to hand someone the solution, that limits the maximum level of reward that puzzle can award.

The two strategies that I think work best here are first: puzzles designed to make the optimum strategy for solving them require more time to explain than it takes to execute. Branching puzzles potentially work here, where the player can be forced to use so much work to *find* a solution, that it takes longer than the typical amount of time it takes for a reasonably good player to *formulate* a solution. Also, puzzles designed around complexity features of the game can also explode any possibility of creating a comprehensive documented solution.

The second strategy is to make puzzles in which there are no generalized solutions at all. This is not easy, but there are known "puzzles" for which there are no general solutions and which can be partially randomized by a computer. Some mathematical problems that are usable for zero knowledge proofs can potentially be turned into puzzles (hamiltonian cycles comes to mind, where you have a set of dots connected by lines and you have to find a way to traverse all the lines while visiting the dots only once each).

The tricky part is integrating these into a game in a non-forced way. For example, its theoretically possible to make an instanced map that is comparable to a hamiltonian graph (junctions connected by paths) and create an objective within each pathway the team must reach and acquire. If the junctions have particularly difficult challenges that slow the team, then the team that can achieve all the objectives while passing through the junctions the least number of times will get rewarded with the fastest completion time and by extension the best reward per unit time. Its possible to make just a couple of maps and have the computer generate random variations of those all guaranteed to have similar solutions, but no amount of running the mission would ever give you a reusable solution to the map.

The nice thing about integrating puzzles in this way is that they are much more seamless than literally placing a puzzle in front of the players in a gameplay-disruptive way. And the solution isn't binary: there is a best solution, and a worst solution, and a lot of solutions in between, but there's no way to literally be unable to find *any* solution, unless you literally cannot navigate maps without assistance.

Anomaly alarm. Interrogatory: malfunction? Computation: non-reconciled. Activated: verification procedure 1721-alpha.

Located: criteria match seven zed three.
Target description: "sturdy."
Attribute modification: damage resistance.
Effective denormalized value: ten percent.
Veteran reward value: ten point five percent.

Response: T'He O'Cho; diagnostic submission. Specification: numerical quantification evaluation.

Notation: Recommendation. Individual T'He O'Cho. Combat effectiveness: questionable. Non-combat role options: Rikti retirement plan oversight: explicit assignment avoidance protocols asserted.

It is cheating, but if we're talking strictly about AE buffing allies, its a form of cheating the devs should really account for. Technically speaking, that buffing ally should create an XP penalty as essentially a team mate. I haven't checked recently to verify what that penalty is and if its still enforced, but I am curious to know if the penalty is strong enough to make this at best only a slight improvement over most soloing tactics.

I don't know about "easily" but it should be able to run it at least fairly well. Positron said this about video card performance and Ultra Mode:

So just one of your 260s should get you to middle of the road Ultra Mode performance. However, its a bit more complex question as to whether an SLI rig will get you a lot more performance in Ultra Mode or not very much: this is at least partially an architectural question about Ultra Mode itself. I'm guessing you'll do significantly better with the SLI configuration, but I've seen SLI be somewhat hit and miss at times.

RAM and CPU should be fine for City of Heroes itself. You'd only want more RAM if you intend to be doing a lot of other things in the background while CoX is running.

Its 5% higher, not 5 percentage points higher. I.e. small luck should increase from 12.5% to 13.12% (that should be 13.13% by the way Castle).

Break frees don't improve with this veteran reward. But awakens do: specifically the part that heals you as you rez should increase by 5%.

Two-factor tokens are the most straight-forward, simple, and cost-effective way of addressing the specific security issues they have been having. No password policy can protect against password harvesters by definition, and they are specifically being targeted by password harvesters.

As one of those generalizations everyone tosses around at cocktail parties, this is true. But when it comes to brass tacks, two-factor authentication using token generators almost always is, for the simple reason that (and not all two-factor authentication systems have this property) they prevent replay attacks.

No safety precaution available to end users can protect them against replay attacks; in other words nothing can stop something from simply recording their password and then reusing it. Token generated codes cannot be replayed (not in the general case) which makes them immune to any data-collection attack on their systems.

Theoretically, good security practices can mitigate some of these attacks, but nothing has a 100% success rate here. And because increasingly these forms of malware attacks are being generated by well-funded criminal organizations rather than the general goofballs of the past, they are becoming far more sophisticated than we used to see even just ten years ago.

That's actually one of the things that bugs me a little also. Its clear Nova is balanced around being enhanced, but that means you get it at 32, then you get to actually *use* it at level 33. It doesn't actually mature until level 34.

That doesn't mean I agree with everyone's assessment of what is worthless and what is worthwhile. But that's one of the reasons behind my thinking towards general vs specific enhancement. If you *want* to enhance a particular attack's damage, you can, but the *option* exists to enhance your across the board damage, which will affect any attack you acquire in the future.

I'm not sure exactly *how* I would do that. I'm not sure "player-wide enhancements" are quite the right way. Its just a hypothetical thought at the moment.

That's why I was careful to say "options" rather than "powers." I don't expect people to have all 24 powers in the tutorial. I only expect them to have a choice of powers that encapsulates most of the viable combat options in the game. So ranged damage-dealing would be represented, but that doesn't mean you'll have every single ranged attack from the start. You'd just have enough of them to make the choice of being a ranged gunslinger a viable option. Conversely, if you want to be a blapper, there'd be enough melee options to make that choice viable. But that doesn't mean you'd necessarily have thunderstrike at level 1.

I don't think you can pin this on Roper. I think the dev team was struggling with the consequences of the pseudo-open system they were expousing long before he got there. The discussion that swirled around "Frameworks" seems instructive. First, it didn't seem clear the devs themselves had a solid idea of what they were thinking there - or had conflicting opinions between themselves. Second, whatever it was originally it changed shape as arguments built around some of the concepts behind them. Finally, we ended up getting basically *nothing* like what was originally described: instead "frameworks" became the dewey decimal system of categorizing powers, with 2% prerequisite tiers added on top.

The current CO system, as I perceive it, is the CoH system with both archetypes *and* enhancements removed. You can pick whatever powersets you want, and whatever powers you want in them, but you have minimal control over your slotting of those powers. And for me personally, that is its biggest failing in terms of being a Champions inspired game. Champions allowed two different kinds of orthogonally balanced customization: you could take a different mix of powers, and you could customize those powers in interesting ways. I think the former is at least possible to some degree in CO, but the latter is not to a high enough degree to be interesting.


Since we're playing fantasy game design, one thing I would do differently than either CO or CoX is I would focus more attention on character development. Both games treat character development as if we were a bag of powers, and development involves filling the bag. I think a better way to do this would be to consider this a three-stage process. In stage one, you decide what you can do. In stage two, you decide how well you can do it overall. In stage three, you decide to specialize in certain things to add focus to the character.

A big knock on both games to some degree is that it takes too long to "realize" your character concept. I think one way to address that is to give all the conceptual choices to the players very early on. In fact, I would present all of them by the time they left the tutorial (if the tutorial was long enough). So if you want the ability to supercharge offense at the expense of defense, you get to choose that ability early. If you want to fly, to buff others, to become temporarily invulnerable - all those types of general options would be made available to the character early on.

But they would be relatively weak versions of those abilities. As you level, the opportunity would arise to improve those general abilities in either general or specific ways. In other words, rather than the slotting progression we have now, I'd allow players to just plain increase their damage across the board if they wanted to, but at lower amounts than boosting individual attacks. Rather than build up from being able to block melee, then block ranged, then block AoE, SR would be able to block all three, but weakly, and then build up their general defense to all, or divert more attention to building up melee defense.

In the end game, I would allow players to take certain abilities and specialize in them in ways unavailable to generalists, no matter how much they buff those powers during the mid game. Maybe you could increase the firing rate (in cast time) of your basic attacks, or you could use certain powers free of its normal restrictions. Perhaps while personal force field was up force bolts were reduced in strength normally, but in the end game you could specialize in attacking while those shields were up. That sort of thing.

In a sense, you'd be picking your archetype in the beginning through fundamental character ability choices, then you'd be leveling sort of the way we do now in CoX but with more general options for enhancement, and then in the end game you'd start to get power advantages vaguely like in CO.

There are some differences; in particular the reported date conversion seems to match the general trends of the access numbers somewhat better than the average value conversion method for the Q4 '07 - Q2 '08 period. It also appears to better correlate with the substantial Q3 '08 drop in access numbers.


The OANDA historical currency exchange calculator can be given a date range, and it will generate a table of the exchange rates for each day along with an automatically computed high/low/average for that time period.

Visually to a degree, but that's somewhat illusory. For example, in adjusted dollar terms (using the method I use, which is exchange rate at reporting date, not averaged exchange rate during the period, although there are similar discorrelations with the alternate method) there is a sizable drop in revenue in Q2 '06 with no corresponding drop in access numbers. Then, while there is a correlation between the drop in access in Q4 '06 and revenue, revenue bounces back in Q1 '07 but access doesn't start bouncing back until Q2 '07. Revenue drops much faster than access from Q3 '07 to Q1 '08. There is some correlation, but there are significant offsets in some quarters and in others there is correlation in direction but disproportionate changes overall.

*If* we believe the numbers are at least fairly representative, then the answer to the question some people have asked: is it possible we bounced back after the drop in subscriptions (really, access) in Q3 '08, is yes, we did: to practically Q2 '08 numbers. We've since dropped back down to nearly the same levels as Q3 '08.

Could the increase in revenue in Q4 '08 - Q2 '09 be due to booster packs? Well, its a total increase for the three quarters of over two million dollars in revenue. If we assume $10 per microtransaction on average, and we assume we have somewhere in the neighborhood of 100,000 subscribers plus or minus, then that means on average every single player bought at least two booster packs in those three quarters. I suppose that's possible, but it seems unlikely to me (From October 2008 to June 2009 I think only the Magic pack was released).


In any case, this is my version of the revenue vs access numbers (click for larger version):

Not really, because you keep misunderstanding the point of the challenge. I'm not challenging anyone to counter my calculations. The calculations are irrelevant. I'm asking people to level fast. If what you're doing isn't leveling, or discussing leveling directly, you aren't talking about my challenge at all.

If you think my calculations are flawed, fine, I concede they are flawed. If you think they are worthless to the discussion, that's a matter of opinion I will not attempt to dissuade you from. I'm asking if anyone can level faster than a set of numbers I picked based on my own judgement on an upper limit of leveling that I believe to be primarily influenced by endurance. That's all. If you don't have such a method, and aren't interested in finding such a method, you can continue to challenge the assumptions behind the limits I've picked but since I've already stated that the individual assumptions that are factored into the limit are just there for rough estimate, and based entirely on my own judgement and not a strict numerical proof of sufficiency, you'd basically be arguing that the numbers I say are mostly my guesswork are nothing more than my guesswork. Gratz, you win.


And on the subject of "muddled." Since I'm asking for volunteers to test a limit, I asserted *a* limit. Not a range of limits, or a fuzzy set of ambiguous references. A limit. Its either breakable, or not breakable. If its breakable, I would like to know why. If its not, then its not.


I don't think you understand what the purpose of "challenges" are. You seem to think the point of the challenge is to be unwinnable or unbreakable. That's never the point of any of my challenges to the forums. If I *know* its unbreakable, I don't bother to challenge anyone to beat it. That's one of the reasons why I allowed so many potential loopholes, such as invention slotting, or AoE efficiency. They are there to create the *possibility* of breaking it, so that someone might actually try to do so.

Its trivially easy to create a limit that is essentially *impossible* to beat. The impossible to beat limit is the calculated limit of defeating the maximum reward value target spawnable at your level plus the optimal level scaling value, at the maximum AoE limit, with the maximal mission completion bonus per kill normalized with the best possible zoning time. It comes out to leveling to 50 in about an hour.

Its also trivially easy to spend a month making a very complex set of calculations projecting the best case leveling estimate given a set of about a hundred different assumptions, which generates a number you can *defend* but is also worthless to discussion.

This is my attempt to strike the interesting middle ground. But I don't expect it to be an interesting point of discussion for everyone, because I'm not currently interested in refining an estimate I've *deliberately* made extremely coarse specifically for discussion purposes. This is not about math or numerical methodology. This is about fast leveling.

Because I can't say this is false often enough, this is false. I can deduce a lot more about the state of the game from the revenue numbers than I can from the access numbers. The combination of the two is better, but if I could only have one, it would be the revenue numbers.

The reason is that the access numbers can prove dramatic increasing interest in the game. The revenue numbers can disprove dramatic decreasing interest in the game. I would rather have the information to prove the latter than the former.

Also, access numbers are more of a trailing indicator, revenue is a bit more of a leading indicator, factoring out oscillations.

Finally, revenue forms the direct basis for NCSoft continuing to develop the game.


If I had something to hide, I would hide the breakout in revenue numbers, not the access numbers. So if this is a conspiracy, its a conspiracy of the stupid.

Sales *are* primarily subscription revenue in the reported numbers, unless you believe we've been selling a couple hundred thousand boxes a quarter, in which case new sales *would* be trumping our subscription numbers.

For some reason more than one person has misinterpreted those numbers to mean "number of units" like "5000 box sales" when they are actually millions of Won, or very roughly thousands of dollars. Those numebrs are about five million dollars in revenue a quarter, which is probably not a quarter million copies of Good vs Evil going out the door.


Actually, its not. It would be, if interest in the game was so high that the vast majority of players logged in frequently. Ironically, anyone attempting to claim that interest in the game in general is dropping can't make that claim, and thus cannot make the claim that the access numbers accurately represent number of subscriptions. In fact, if interest is dropping in general, the access numbers should drop independently of dropping subscription numbers and there would be no way of knowing whether any drop in access represented a drop in person-hours logged in, or drop in subscriptions, or both.

Those numbers are only good to compare two relative periods where the average login-frequency of the average player is similar. And its clear that the numbers encapsulate times when it has not been.

I can. I put that specific set of circumstances to the test last year in six different builds. However, that's purely anecdotal: you could argue that the reason why its true is not because endurance itself is a limit, but because I'm inefficient.

However, being able to consistently break the limit I posted is less vulnerable to that charge, because I've flipped the sufficiency requirement around. Every counter-example to the conjecture is significant because its asserting possibility, not generality.

You seem to be suggesting that the exercise is pointless because the leveling enveloping I'm discussing is far outside the realm of possibility, so its irrelevant if its true. However, that's the entire point of the exercise. You don't actually know that to be a fact, and neither do I. And spiritfox is asserting its actually breakable.

What you're doing is making the mistake people made before the invention of the airplane, and asserting powered flight is impossible because no locomotive fast enough would be light enough. Ultra-high performance leveling is a sufficiently esoteric art that I'm spending some time trying to avoid that error.

That doesn't match the calculations I maintain. Hmm, lets see.

Ah, its the exchange rate. You're assuming the average exchange rate for the period in your conversion. I don't believe that is correct. I believe its much more likely that all of the revenue is accounted for in the US in dollars, and then is converted to Won for the purposes of reporting as of the conversion rate at (or very near) the reporting date. In that case, what I have, for at least the quarters from Q1 '05 and later, is: