Oracle Analysis - Math and Comparisons

The Oracle is a meant as a multi-purpose unit – a unit which can harass, provide detection, and provide scouting. For the reason that it is multi-purpose, you have to kind of choose which role you want it to fill at any given point in time, and it must also be slightly weaker than an unit of whose sole purpose is only one of those roles. A lot of people think Oracles ought to be as effective at harassment as Mutas or zealot drops – but this is the wrong comparison because Mutas and zealot drops cannot provide detection or scout a location safely for an extended period of time.

Having said that, it is my purpose in this post to show that the Oracle is so much worse at each of these roles that it offers very little for its cost. I can only hope that Blizzard will do something about this unit, because its current iteration is truly lackluster. Let’s do some analysis to support this.

Entomb Math

Entomb is 100 energy for mineral patches to be locked down for 30 seconds. These “mineral shields” are attackable, and a full line of workers kills them off in anywhere between 6 and 10 seconds from the time they are told to do so. A warning sound is displayed when harvested mineral patches are entombed.

This means that every time Entomb hits a line of 16 workers, it denies somewhere between 112 minerals ( avg. 10 seconds per patch) and 336 minerals (the full 30 seconds per patch). Every time Entomb hits a line of 24 workers (possibly irrelevant numbers), it denies somewhere between 136 minerals ( avg. 10 seconds per patch) and 408 minerals (the full 30 seconds per patch). These numbers, of course are averages and assume the workers resume mining upon the completion of the harassment. Due to the energy cost of this spell, that means this spell can only be cast every 3 minutes – unless a Mothership core helps out.

This means that if your opponents reflexes are good/passable, it will take your Oracle (assuming 112 per Entomb) a full 10.5 minutes from the time of construction to deny more resources than it costs. This assumes, of course, that you can keep it alive for 12 minutes and make 4 clean runs into the mineral patches. Notable fact: at 100 health, the Oracle has the lowest health per gas cost of any unit in the game. It has only 20 health more than a High Templar.

Now, if your opponents reflexes are bad – it only takes you 4.5 minutes to make your Oracle cost effective. I want to make this absolutely clear. This scenario (which no doubt is what people are hoping for) is only possible if:

1. Your opponent does not attack his mineral patches. He makes no reaction to them at all. He just sits there and lets his workers not mine or attack for the duration of the spell.
2. You are able to get to and from the opponents worker line twice, and not die in between.
3. You use none of the Oracle’s other abilities during this time. No detection needed, no preordain needed.

Even under this scenario, if you build the Oracle at 5:30 (which is pretty early), you will always have less resources than your opponent until the 10 minute mark. This assumes no Mothership Core involvement (for reasons we’ll show later).

If you used the Stargate solely for the Oracle, add another 12 minutes in order for the Oracle to pay for itself (for a total of 24 minutes needed to be cost effective) in the case that your opponent reacts instantly. The case that your opponent doesn’t react is unchanged. These numbers also assume that gas and minerals are interchangeable – I feel that for a protoss early game gas is worth much more than minerals, but for simplicity (and the fact that it doesn’t hurt my assertion that the Oracle lacks luster), I will leave this undisputed.

Mothership Core Involvement

But what if our friend the Mothership Core helps out. Surely then it’s worth it, right?

Let’s consider the opportunity cost of using Mothership Core energy for the Oracle. What does it cost to use Energize on the Oracle rather than a Nexus, for instance? One full Nexus of energy is 8 Chrono boosts (and you can possibly squeeze in one per Nexus as its charging). We’ll say this equates to 90 seconds of Nexus build time (i.e. you got one off while it was charging to full), because Chrono boost does 30 seconds building in 20 seconds time (please don’t dispute this, we’ve had enough threads of people debating whether 50% speed meant any variety of creatively bad math, let’s leave that out this time). This is about the same as 5.3 probes – about a third of a mineral line (or 223 minerals/minute) for the rest of the game. In exchange for this, if you use Energize on an Oracle, you get two Entombs (ranging from 224 minerals to 672 minerals).

This means that even if your opponent does not react at all to your Oracle after 2 consecutive Entombs – the game had better end within 3 minutes, or you’d have had more of a mineral lead if you just built probes. Realistically, as long as you’re building probes, you should never use Mothership Core energy on an Oracle. In essence, you’re trading 224 to 672 resources for him now, in exchange for 223 minerals/minute for you both now and later. It’s not a good trade.

If you’re not building probes, Energize can provide 4 Forcefields, 4 Phoenix lifts, or 2.66 Storms. Ironically (if they’re reacting to Entomb), 4 Phoenix lifts provides more minerals denied (at 168 fewer minerals/minute) in a very short time span, and affects mineral collection rate for the entire duration of the game. Or, if you prefer, it simply makes them spend 200 more on workers (either way, it’s better – and phoenixes are hard to kill, too).

Of course, when referring to different collection rates, it should be noted that the rates used will only be different until a player reaches the maximum number of workers he wishes to produce. Realistically, this is only generally attained quite late in the game.

Usually, though, I’d rather have FFs or storms anyway. That’s not the point, though. The point is that I see no point in the game where I’d want to spend Energize on an Oracle – except that if you don’t, it will take a long time to deny your opponent more resources than it denied you during its construction – even if your opponent pretends your Oracle is doing nothing at all. Depressing, ain’t it?

Reactions to Harassment

Perhaps the worst part about the Oracle is that it doesn't need to be stopped. When you see mutas or banshees, you need anti-air. Therefore, you need to scout for mutas or banshees and you're in trouble if you don't see them until they arrive. This inspires alterations to "safe" play, it inspires scouting, it inspires player interaction. If you don't see an Oracle until it arrives, you um... should make sure you still have workers to attack the Entombed patches? The Entombed patches aren't flying - they don't go invisible - and so the harassment itself requires nothing more than "box + A + click" to deflect. It doesn't inspire scouting or player interaction. It inspires your opponent to just wait and see - and when you see an Oracle, you think "well, I guess I could go kill that - eventually". If there's no pressure, it's not really harassment - it's just checking to make sure you're paying attention.

[image loading]

The ideal opponent for Entomb.

There's nothing inherently wrong with that, but given the amount of resources you're spending on a unit which can't contribute to attacking or defense, I find it hard to imagine your opponent will have much hesitancy to attack you at his next opportunity (you'll be at your weakest, since Stargate units are notoriously ineffective at defending against anything which can shoot up), or (if experience shows that ineffective) to just expand really, really aggressively and effectively turn your decision to harass into an economic advantage. Honestly, what does your opponent have to worry about (and therefore, what keeps him from mass expanding or pulling SCVs and all-inning) if he knows you spent 150/150 on a stargate and 150/200 on an Oracle?

Revelation

Given that its harassment potential is… lackluster, perhaps its detection ability will prove useful and cause players to want to get it anyway (in this way, the Oracle would be something that wouldn’t need to pay for itself – any mineral denial you get would be a bonus). Compare the price of an Oracle (150/200/3) to the price of an observer (25/75/1). Not a good comparison, obviously. And in exchange for that, you get an ability which can provide detection in an area (radius 6) for 30 seconds for 50 energy, which is generated over 90 seconds. This means that if you are in constant need of detection for extended periods of time (in the case of fending off swarm hosts from a distance with Tempests, for example), you are only able to detect in one area of radius 6 for 30 seconds, and only every 90 seconds (i.e. it’s a minute between Revelations). It takes 3 Oracles to keep detection up constantly. So perhaps the draw is that you are able to center the radius of the detection away from the unit itself. But really, when you consider that the site range of the Observer is 11, that the spell range of revelation is 9, and that the radius of the spell is 6 – this only provides an additional distance of 4 from the Observer’s site range, and the detection range is dodgeable – you need your attacking units to be on top of the target being detected anyway, or they’re just going to leave the area. Sometimes they can’t kill it before it moves on anyway. I watched Babyknight grapple with this exact phenomenon on his stream a few nights ago.

About the only thing the Oracle has going for it on the detection front is that it can run away if an army of anti-air/stealth units attempts to “snipe the detector” and let their stealth units rampage. How often does this happen? Perhaps Corruptor/Swarm Host/Overseer will do just that in HotS though. Who knows?

Preordain

The last and least useful of the Oracle’s abilities, preordain shows you what’s around a building and what it’s building for 60 seconds. It costs 50 energy (and therefore, can be used every 90 seconds). Doesn’t this provide valuable scouting intel? Yes and no. Let’s compare Preordain with Hallucination (phoenix). Both last 60 seconds. Both provide potential vision of the opponent’s base/army. Hallucination is mobile and killable, but leaves no risk to the initial sentry (which can chill at home). Preordain is immobile and unkillable, and requires the Oracle to come within 8 of the selected building. Under which scenarios is preordain better?

1. If the HalluPhoenix dies, it may not scout the building it wanted to scout. On the other hand, if there is anti-air leading up to the building you want to see, is the Oracle going to get there? A phoenix taking double damage only has 10 fewer health than the Oracle. And if you lose the Oracle but Hallu provides no risk to the sentry, it’s certainly not as good. But, perhaps if you are able to get close enough to a building close enough to the building you wish to scout, but not close enough to see the building itself (i.e. you get lucky). I cannot think of many situations under which this could occur. Maybe a turret which doesn’t let you see what’s being built over in a corner of his base, so you preordain the turret and you see his dark shrine? This type of thing will almost never happen – doubly so because he can see what building you’re preordaining.
2. You have a building which, upon scouting different units being made, your reactions are different AND, the unit builds slowly enough that the HalluPhoenix is not able to wait around to see what pops out before getting killed.
3. If there is an area of critical importance that you must view for a long period of time – and it happens to be close to an opponent’s building (i.e. a ramp near the initial barracks, an expansion plopping down near a building, units walking by a building with preordain on it because your opponent is bad, etc.)
4. If there is a building you need to check up on more often than your sentry energy would support.
5. Your Cyber Core is busy.

When is Hallicunation better?
1. If you have sentries, but no Stargates/Oracles. This will occur often.
2. If the scouting you wish to do has to do with spotting units and their location, not just buildings.
3. If you wish to send a summoned unit to its death to see a particular area of a base, the way you can’t do with an Oracle.
4. If you ever use Energize on your sentries for any reason (FFs?) and can therefore spit one out and still charge to full.

I doubt, looking at those two lists, that anyone could conclude preordain has anything more than highly situational use in comparison to Hallucination. And you’re going to have a sentry – at least one. If you have a standing army, you just will.

And let’s not forget, through all this comparison, that most people don’t get Hallucination for scouting in most games, and the reason why – once again, usually the observer is just better.

But doesn’t one Oracle serve all these purposes?

You may have noticed that 200/100+25/75+100/100 > cost of an Oracle (but oddly, less than the cost of two). But this is really beside the point, since scouting and detection are not things you can simply not have a plan to get when you need it – and since one Oracle can only fulfill one of these roles for a short duration, and only periodically. Let’s say you make an Oracle, and you fly into his base and find that he’s making Banshees. You want to find out when he stops making banshees (or maybe if he’s making tanks to go for a 1-1-1, so you use preordain. But now you need detection, so you make another Oracle. Your original Oracle will have no energy for 90 seconds. Your new Oracle pops out with energy for a revelation, but the banshee gets away stealthed but with very little life left, heading towards your expo. Your initial Oracle now returns with 50 energy and is able to use revelation to allow the stalkers to finish the banshee. You now have two Oracles, but when a second Banshee pops up again in your main, you need to have made a third Oracle because the second Oracle isn’t full up on energy. You use revelation and have enough stalkers that it dies this time, but now you have 3 Oracles, your preordain has long expired, you have no energy to harass and you’re unconfident you have enough Oracles to keep detection going if another banshee pops out. In addition, 450/600 of your spending is now charging up, whereas the normal robo opening against banshees would put out 2 observers (50/150), easily repels the initial banshees, and just barely have enough gateway units to repel a 1-1-1. By comparison our stargate player has done no harassment, knows a tank-based 1-1-1 is coming, but has so few resources that he is going to be hard-pressed to stop it. But if he made 1 fewer Oracle, if he did any harassment (keep in mind that’s 100 energy, so we wouldn’t have had that second scan we referred to).

In short, you can’t build one Oracle and get detection, scouting and harassment out of the deal. You can pick one, and that one expenditure is all the detection/scouting you get for 90 seconds – even though it lasts much less than that. And if you go for harassment, that’s not all you get for 90 seconds – it’s all you get for 180 seconds. If you could get scouting, detection and harassment, the unit would be awesome – but you can’t by a very, very large margin. Don’t think what I described is likely? It’s almost exactly what happened to Babyknight in a game I saw him stream. The details were slightly different, he harassed with his opening Oracle, the Banshees came much later, and so his second Oracle had more energy with which to detect – but the point is the same. If you need detection for an extended period of time, and you’re relying on Oracles to get you there, it’s going to be expensive – very expensive. Harassment is going to be hard to make cost effective. Scouting… honestly I really wonder whether anyone will use preordain. I suspect they’ll just use the Oracle as a kind of glass phoenix to scout.

Why use the Oracle?

200/100 + 50/150 for detection at base, plus scouting at your opponent’s base until it gets sniped (requires detection). Compare that to 150/150 + 150/200 * how many Oracles? One for detection/risky flybys? Two for harassment? More? Perhaps if you’re opening stargate in PvP, and you scout DTs, you might make one – though after the detection bit I guess I’m not sure how it’d be valuable again.

And combine all this with the fact that the unit is really, really fragile, and doesn’t require detection to kill and I think, at the end of the day, I’d get more from 3 observers (similar costs/supply) – so I’ll leave this as an open question.

This is posted on Blizzard's forums at: http://us.battle.net/sc2/en/forum/topic/6551205756

Please, let Blizzard know if you've been watching this unit like I have and think that it's just not worth it right now.

Oracle Analysis - Math and Comparisons

Completed

Ongoing

Upcoming