Hacker News Reader

Comment by DanBC | Hacker News Reader

Running seti / etc on the school computers does result in thousands of dollars spend on extra electricity. He was number one on SETI for years, running the client without approval on 5,000 machines. It's not surprising that people get fired for increasing costs by that much without getting approval first.

View on HN

Yes, running 5000 computers on 100% CPU instead of idle is a lot of money and a lot of electricity spent:

If these were typical desktop computers, the CPU working at 100% can use some 50 W more than the one staying idle. 5000 computers at 100 % CPU produce then the load of 250 KW, resulting in 6 MWh per day which at 10 cents per KWh costs 600 USD per day, or 220000 USD per year. Finally, 9 years that he did it make almost 2 million dollars (!) for electricity that he managed to spend.

You can roughly double that. Thats the cost of the extra electricity needed by the ACs to offset the heat generated by the pegged PCs.

Thousands can be wasted on dumb stuff like bad office-supply purchasing decisions. Those folks don't get fired. Sounds overblown - folks afraid of anything they don't understand.

Would you still consider it "overblown" if he had run, say... a bitcoin miner on all those machines?

Sure, he didn't make profit from the seti@home instances, but he did it without approval, did ultimately cost the district money, and likely caused other issues along the way (like additional wear-n-tear on the machines)

I'd think administrators would have a very different approach to disciplining a "bad office-supply purchasing" decision versus a well-thought out choice to use school equipment & electricity for non-work related causes.

> electricity for non-work related causes

I wish we could fire teachers for using their computers for Facebook and Amazon shopping during the day.

And should private sector employers fire their employees for using their computers to do the same? Obviously I don't want to see teachers goofing off on social media instead of teaching kids, but if they aren't actively teaching, why not? The idea that every second of your time on the job needs to be spent furiously working is an attitude from the industrial revolution - I would have thought these days we were more interested in results.

in the school where I most recently taught, there was an explicit expectation that teachers could use district computers for personal business when on break. Being on facebook when they had kids in the room would be an entirely different matter.

The internet was also heavily monitored, with a proxy server that required a login to get to forbidden sites. Our IT guy had a pretty good idea of when teachers were getting on facebook etc.

This question is not intended at all to be snarky. Seriously.

And then you'll hire them?

Gets complicated fast, dunnit?

If they have shown they can get the job done, why wouldn't I? (Other than not being in HR and all).

Also, hard not to feel a little snark in the emphasis and "dunnit", really...

I would not have thought of that :) I just thought it was a contraction of "doesn't it".

Perhaps they are like farmers we paid to not grow corn in the past.

But that's just for the time-servers, not the dedicated professionals, who are probably the majority.

Do you want to be fired for using your computer for non-work activities?

Having lived in the deep south most of my life I think I understand what's going on here.

It's not that they're afraid of things they don't understand. It's that there is a certain type of person they generally don't like. In this case it might have been a man that had some sort of intellectual curiosity, maybe a bit eccentric. And they will use their ignorance as a weapon against those people.

The police department knows full well they can determine whether this was a bomb or not. It's within their ability to interview every person he interacted with to find if he claimed it was a bomb at any point. Instead of doing that they will use their lack of knowledge to allow falsehoods to move the investigation.

It was within the ability of the staff to determine the monetary cost of this man's actions. In the very least they could compare the electricity bill during a month he was running SETI versus the next month when he was not. They could calculate the power consumption of the machines.

It's not that they are afraid of knowing. But that knowing or admitting the truthfulness of the facts makes them less powerful as people.

I think it's both: ignorance is the source of their power, and they are afraid of losing it. (I also grew up in the South.)

> ignorance is the source of their power

Ignorance is a source of power only within a like-minded group, and only as long as that group is powerful.

Now get out of there and see where ignorance gets you.

Or wait until that group washes out into the irrelevance of history.

If you're stuck within such a group, you can wait your whole life for history to prove you right and still be disappointed - better to start planning your exit strategy.

"The market can remain irrational far longer than you can remain solvent."

Precisely - I actually started out with the same quote but trying to articulate a notion of epistemological solvency became unwieldy :-)

It's actually not the ignorance per se, it's the self-confidence that comes from a mindset that allows you to believe things without evidence, or even in the face of contrary evidence. That sort of self-confidence can be very powerful (c.f. Donald Trump) but, of course, it can also be very dangerous.

> Or wait until that group washes out into the irrelevance of history.

That's all fine and good on a geological scale, but we've been waiting 200+ years for the irrelevance of this particular group, and there's not a hope in sight of it happening any time soon.

Saying "don't worry, this group that currently holds power over large parts of the country only holds power currently, and only over large parts of the country" is not as reassuring as you probably meant it to be.

Irving, Texas is not typical of what most consider the deep south. It's part of the DFW metroplex which is in the nation's fourth largest population center[1], and it's in a county that went 57% for Obama in the 2012 election[2]. So let's just be careful about painting wide swaths of the country with an singular view.

[1] - https://en.wikipedia.org/wiki/List_of_Metropolitan_Statistic...

[2] - http://elections.nytimes.com/2012/results/president

I guess you are not familiar with Irving's Mayor, Beth Van Duyne... Sorry to blow your theory right out of the water.

Are you saying the Mayor defines this area more than the numerous other demographics that show it is not simply a backwoods hillbilly town?

Yes... I believe the Mayor of Irving is an elected position by the people. It's a better indicator then being the 4th biggest metropolitan in the country like that matters.

And going 57% for Obama also says something about the people. We're not saying they're bleeding-heart-liberals. We're saying it's a little more complicated than, "Oh well it's the deep South..." Too many people are dismissing this like it's just the norm. It shouldn't be. More people should be asking if Islamophobia is a bigger part of mainstream America than we're willing to admit.

It doesn't say as much as you think. Texas has done gerrymandered itself up pretty good.

That, or they just didn't like that someone illegally installed money-wasting software on school property without so much as an approval request.

PS: Money wasting from their point of view, not making any judgement about SETI here

Right. I really should have separated the two incidents from my single point.

I don't get how this point is debatable.

I did this when I worked as a network administrator for a new school campus. It was folding@home vs. SETI but same concept. Once the administration realized all the computers were running folding@home they compared the bills after turning it off and it was tens of thousands of dollars difference (granted this was 2001 and full-size desktops).

Now, one could say they overreacted, sure, in my case I just got reprimanded and told to uninstall it off all the computers. But firing for costing the district significant amounts of money isn't some testament to the ignorance of the South, unlike the "bomb" clock.

> tens of thousands of dollars difference

I still don't believe that - unless this was 10k computers+. I'll bet that if you used watts up on an idle computer, then used it on a computer running seti@home I'll bet the difference would only be 100 watt difference at the max (probably more like 50 depending on what is running on the computer).

"compared bills after turning it off" is a direct measure of costs.

5k is easily within an order of magnitude of what you would find believable, and he's only expressing an order of magnitude. I think your dispute is rather academic.

See my other post here for details, 5000 desktop computers using 100% CPU all the time can cost some 18000 USD per month (1) more versus keeping them idle.

1) if 1 KWh costs 10 cents, which is reasonable to expect, and the load increases just 50 W.

And for this guy here it's even much more:

http://setiathome.berkeley.edu/forum_thread.php?id=54609&pos...

His single computer uses 490 Watts when turned on and calculating Seti, and he's keeping it on for 18 hour more per day just for Seti. Only 500 of such computers and such usage patterns make 14000 dollars more per month.

dumb stuff

My wife was on the advisory committee for a local middle school that's struggling. One winner is that they bought SmartBoards for a bunch of classrooms but didn't get stands for them, so they just mothballed them for a year until they could buy stands with the next year's budget.

What would you have them do? If they're over budget, they're over budget and non-emergencies will have to wait.

The obvious solution would be to buy fewer smart boards this year, along with the stands, and then buy the rest of the smart boards and stands the next year.

Sure, obvious. Until you're negotiating price points and you can save X% aggregate by purchasing Y boards at a time, and then figure you're only losing <<< X%*total_cost in depreciation... then it's just as obvious, but you're making a different decision.

It's much easier to justify wasting money on normal business things every day people understand, like buying office paper. Most people can relate to that, even when things go wrong and money is wasted.

Most people, be they in the deep south or the northeast have no fucking clue what SETI is and may even have very specific metaphysical opinions on how we should deal with the possibility of alien life that doing this violates. Most people don't care enough about paper to do more than write it off as a normal expense.

I personally think that alien life is everywhere and I can't wait until we discover more evidence of it. But I think SETI is a huge waste of resources that could be spent better elsewhere. Even though I'm pro science, discovery, and utilizing wasted cpu cycles, I still think searching for life this way is a huge waste of time and resources and I'd be pissed if some teacher decided he was going to use the resources I take a part in paying for to do this.

The benefits of this are external while the cost is internal. If you're going to use resources, at least use them to benefits the kids who are in your budge.

They pretty surely do if their costs were measured by someone.

I wouldnt call someone externalizing costs for whatever noble endeavor responsible or forward thinking.

That's ridiculous. Now you're going to say I'm "breaking the law" for installing a bitcoin miner on all the computers I support and donating the money to charity (well, some of it;rent doesn't pay itself)!

Yep, government works have been fired for mining on government computers. Unauthorized computer use.

http://www.businessinsider.com/researcher-bitcoin-supercompu...

Maybe not breaking the law, but if you're not the person ultimately responsible for the costs and they're not actually your computers, that's still a dick move. I'd fire you if you worked for me and caused a massive increase in my power bill for something that had no relation to what I asked you to do.

> does result in thousands of dollars spend on extra electricity

Thousands maybe - definitely not millions as they are claiming. From what I read it was setup to calculate through the screensaver. For most of the day - the computer will be in use and I would imagine/hope that they turn them off when they leave.

Advice used to be to not turn machines off but let themgo into sleep. This avoided power cycling which some people though caused problems. A sleeping machine still costs money, but much less than a machine churning SETI.

5,000 machines churning SETI for nine years? I'd be interested in estimates of cost.

Quick and inaccurate estimate:

Let's say 100W per computer running at full blast, and assume 12 cents per kWh for electricity. Running for 9 years straight, that's about 5 million dollars. If they're running for 16 hours/day then it's more like 3 million dollars. Adjust other factors to taste, but "millions" sounds pretty reasonable.

If the machines were awake for 12 extra hours a day, and drew 100 watts, that's 0.1kW * 5000machines * 12hours = 6000kWh per day.

Where I live, electricity costs about 20 cents per kWh. 0.2dollars * 6000kWh * 365days * 9years = $3.94 million.

Well, far from being definitive, but it looks like a regular pc will cost around $227.52 (350 watt psu, $18.96 per month) in electricity per year to run 24/7.[1]

(calculated using $0.06 per kWh, which is rather low for most people in the US)

So, 5,000 machines running 24/7/365 would cost around $1,137,600 annually (5,000 machines * $227.52 per year) in electricity usage alone.

Of that 24 hours per day, we can assume 8 hours of each day the district expected to pay for them to run, so that leaves ~16 hours the district expected the machines to be asleep or off.

24 - 8 = 16 (about 66% of the 24 hour period)

So, the admin was consuming about $12.51 per machine per month the district did not expect.

This comes out to around $750,600 the sys admin potentially cost the district, for a single year. He ran this scheme for 9 years -- so...

$750,600 per year * 9 years = $6,755,400 in additional electricity costs

Keep in mind this is likely rather low, since most places in the US have a significantly higher per kWh fee than the $0.06 per kWh used in these calculations.

I'm certain that this sum alone was enough cause to terminate the sys admin.

Think about all the things the school district could have spent an extra ~$7MM+ on...

[1] http://www.dslreports.com/faq/2404

Something is wrong with your math. $227.52/year is $0.62/day

Also, bear in mind that PSUs do not use their full rating unless the hardware they are powering requires it. [0] Your number could easily be double what the machines actually used, if the PSU was somewhat oversized, as they typically are.

[0]: https://superuser.com/questions/106792/does-a-power-supply-d...

Sorry, type-o, math is still correct though. Meant it to say "per month" not "per day".

> 350 watt psu,

PSU's arent light bulbs. A "350 watt" psu doens't draw 350 watts. That's the theoretical maximum you can draw from it. Sure, SETI was probably drawing a higher load than idle, vs off, but I highly doubt it was drawing 100%.

Even more complicated is the heat. If they lived in a cool climate, at least some of the waste heat that was emitted would have been absorbed by the building, which they are (presumably) paying to heat at least a few months of the year.

Edit: also if it was a public school, wouldn't they be off for about 3 months over summer ?

> PSU's arent light bulbs

You are right. I was just going off the source I found, which accounted for some of the aspects you point out, but not all. These numbers simply provide a rough estimate, and detail how the sys admin certainly did cost the school district real (and potentially very substantial) sums of money.

> Even more complicated is the heat.

You make a good case for the winter months, but the opposite would hold true for the summer months when the heat becomes a real issue and additional air conditioning may be necessary, etc...

> Edit: also if it was a public school, wouldn't they be off for about 3 months over summer ?

Not necessarily. Usually over the summer months (for a traditional schedule school as opposed to a year-round schedule like many public schools are these days) some of the staff is typically on campus monday-friday doing various administrative tasks, workshops, prepping for the next school session, moving classroom furniture, cleaning, etc... IT staff would be included in this, and therefore we can reasonably assume the sys admin was still running his systems.

He lived in Arizona which is hot and a school full of space-heaters (humans) will probably run AC more often than heat.

What grandparent commenter showed was the school district's numbers are plausible. I doubt the district entered into a legal battle without first consulting an engineer or two about the electric costs.

I wonder if anybody even so much as plugged a typical computer used into a Kill A Watt and compared with/without SETI running.

I have. I tested 3 models of HP that we had on campus which are kind of aging. Here's my results in Watts:

HP 6000 Pro SFF - Core2Duo E8400 @ 3GHz - 100% CPU: 70, Idle: 26, Sleep: 2-3, Off: 0-1,

HP 6200 Pro SFF - Core i5 3470 @ 3.2GHz - 100% CPU: 83, Idle: 31, Sleep: 2-3, Off: 0-1,

HP 6300 Pro SFF - Core i5 2400 @ 3.1GHz - 100% CPU: 100, Idle: 30, Sleep: 2-3, Off: 0-1,

It's been a few months now, and I can't remember what I used to max out the CPU. It wasn't SETI @ Home, but it should give you a pretty good approximation. When it was off, it had Wake on LAN enabled, so it still drew some power. But it would fluctuate between 0 and 1.

Neat! Thanks for that. That definitely supports the estimate of cost from running SETI. You wouldn't think it'd be that much, would you?

Speaking only for myself, in Phoenix we barely ever turn on heat, even in winter, whereas the AC is currently running and will be for another month or two yet.

Even if the machines only drew 50% of that at load, it would still be over $3 000 000. That's a lot of money.

It's definitely worth pointing out, but his electricity estimate is so low that the numbers effectively work out for Tier 1 in the SF bay area. I'm sure a school is not in tier 1. But maybe they get beneficial commercial/educational/government pricing, I don't know.

They may be off for 3 months over the summer. Which might mean they were actually off, or it might mean that they ran 24/7 and NONE of that time was for 'educational' purposes.

> PSU's arent light bulbs. A "350 watt" psu doens't draw 350 watts.

Yes, it will draw even more than 350 Watts at full load. 350 is the output power, whereas incandescent lightbulbs are measured in input power. So a 350 watt power supply, depending on efficiency of the PSU anyway, is probably using around 400-500 Watts at the wall at full load.

But this completely misses the point. You can put a 1000 watt PSU in a computer and it won't draw 1000 watts at full load. That's just the max the PSU can supply. SETI is mainly CPU intensive so even if they had the absolute smallest possible PSU, SETI isn't going to cause it to produce the full 350 watts.

And that's a pretty awful efficiency you suggest. It may be that bad, but I'd sure hope not.

SETI also runs (or did, from what I remember) a 3D screensaver. It's not doing nothing on the videocard, anyway, but I don't know how close to fully maxing out anything it gets.

Talking about power load, though, I have a Kill-A-Watt pass-through power meter, and I have measured power at the wall exceeding the rated output power capacity of an 80% efficient power supply on a computer running at full load (running a graphically and computationally expensive computer game).

A lot of PSUs are pretty inefficient. You can look up the stats. It wasn't until recently that you even had a lot of them earning the 80 Plus "bronze" rating, which is 80% efficiency. There's still plenty with worse efficiency than that. A decade ago, you're probably looking at 60% efficiency on average.

Just one data point, but I have a machine I run Folding@Home on most of the time, and that is plugged into a Kill-A-Watt. The difference between idle and full folding power is about 70W, as I recall. Not trivial, but 1/5 of what you've estimated here.

15 years ago, the power difference was probably smaller, maybe quite a bit smaller -- modern CPUs are better optimized for low idle power.

Most schools don't have "modern" computers (by most definitions of the word "modern" at least).

You're probably running a much more powerful processor than the school's computers were. The report is from 2009 at the earliest based on the cited forum post. There were some very low TDP processors during the 2000-2009 time frame that were very common in education systems.

Core 2 Duo was <35W TDP and was pretty popular from what I remember. P3 was in the same neighborhood too.

https://en.wikipedia.org/wiki/List_of_CPU_power_dissipation_...

You were looking at laptop Core 2 Duo. Desktop was 65 to 105.

Ah my mistake.

> Think about all the things the school district could have spent an extra ~$7MM+ on...

That's assuming they didn't use computers. Your calculation assumes that the computer uses a full 350+ watts of electricity. If it had a 350 watt PSU and was drawing that full load - it would shutdown (I've had this happen). Even then the CPU doesn't require 350 watts of power [1]. The only device that I know of that will use and designed to use a full load is a bitcoin miner.

Taking in consideration that these computers might have been left on already. I would argue that he used MAYBE an extra 50 watts of electricity [2]. So assuming $0.10/kw/hr (which is the cost of where I live) - he wasted a whopping $0.12/day or ~$50/year.

You have to take into account if the computers were already on. He didn't waste electricity because someone left their computer on - his "waste" would be electricity use that is above that of an "idle computer". It should also be pointed out that if the computer has Mcafee or some other crappy AV the CPU usage would be 100% anyways due to poor programming of the AV software (I have personally seen this many times and you don't know how many times people complain about their slow computer because the AV is using 100% of the CPU).

There is no way he wasted $7m - and even if he did and no one noticed that is part of a larger problem. Besides they only claim he wasted $1.2m - $1.6m [3] - with no evidence of how they came up with that number. That is saying he managed to waste over $100k/year (over 9 years) in equipment purchases, electricity etc - and NO ONE noticed this? I find it hard to believe that the people managing the budget were like "$100k unaccounted for this year? no big deal...". And even if it was accounted for and signed off - there should be 2 people fired - his and the guy who approved the purchase.

Even from the article:

> would find that in a middle of a lesson, the SMART Board had turned off.

This has happened to me personally on my own laptop. Not because I was mining bitcoins or running seti@home. In fact I've seen them installed and they are such POS that no instructor I know actually uses it as a SMART board.

[1] https://en.wikipedia.org/wiki/List_of_CPU_power_dissipation_...

[2] http://i.stack.imgur.com/4HQPY.png

[3] http://www.azcentral.com/news/articles/2009/11/30/20091130se...

Even using your math:

~$50 per year per computer * 5,000 computers = $250,000 in additional power consumption per year

He ran this scheme for 9 years:

$250,000 per year * 9 years = $2,250,000 total additional electricity costs

No matter what way you run the math, he blew millions of the school district's budget.

> No matter what way you run the math, he blew millions of the school district's budget.

And the school blew millions by not turning off the computers when not in use. Power management is built into Windows - this can easily be done/managed with group policy.

> $250,000 per year * 9 years = $2,250,000 total additional electricity costs

This is all assuming he went around in a ski mask before he left every day and turned on all the computers across the school. Sure he might have used WOL but from the article I don't think he was that devious. From reports of what I read this was something the previous system admin guy did and just passed it along what to do.

I'm not saying he could have wasted money - I'm just giving him the benefit of the doubt and trying to make sure we don't frame him as some sort of criminal mastermind out to rip off this school district. There are a lot of reports out there about this and it's obvious the school wanted to spin it to look like he was the bad guy.

He was the IT guy, so for all we know (and can likely safely assume) he had a GPO to not turn off the systems, but instead schedule this app to run after students left for the day.

I do believe somewhere someone stated the district had wanted the systems put to sleep (not turned off) and believed it was happening.

All of that extra power really only amounts to tens of dollars per year.

Running a processor at an extra (grossly exaggerating here) few watts + an extra 5-10 watts for fans, LEDs, etc don't amount to much. Especially when you consider that all electricity is paid for by the Thousand-Watts/Hour.

Were the machines always on regardless? Wouldnt they have consumed some portion of the said electricity no matter what? Or was he preventing the machines from being shutdown or sleeping?

The software runs on lowest priority. So when people are using the computer the software just sits there, not doing much. After a few minutes of inactivity the seti software (which sometimes was a screensaver) would kick in. Rather than sleeping or idling the computer would be doing computationally intensive work. Some distributed stuff (especially Prime95) is used specifically because it makes the computer work hard - it's a reliable stress test. This causes extra power to be used for the processor and the fans.

http://www.mersenne.org/download/

Running Folding@Home, SETI@Home, Prime95, and other CPU intensive applications requires much more power than simply running idle. For example, I have a processor that when running Prime95 will use more than double the idle load wattage. [0]

[0] http://www.bit-tech.net/hardware/cpus/2010/04/27/amd-phenom-...

> Wouldnt they have consumed some portion of the said electricity no matter what?

Some yes, but there's a lot of difference between a machine idling at 0% CPU and one fully loaded with SETI or Folding and running 100% all the time.

Would be even more worse now with deep sleep states and GPGPU, but even back then the difference between an idle and a fully loaded machine was absolutely significant.

During the winter months, the electric heaters that were also performing computation would supplement the normal dumb heaters. The cost of electricity or fuel saved on heating should therefore be subtracted from the cost of running @Home. Likewise, additional cooling costs would have to be added for the summer months.

Once those kWh are used in the CPU, they don't just disappear. They become heat.

http://www.technologyreview.com/news/414050/computer-cluster...

http://www.bbc.com/news/magazine-32816775

Is heat turned on around the clock?

Even if they were always on running *@home increases their power load a lot during non school hours when they'd normally be at dead idle. The way @home software usually works is once the CPU has been at low utilization for a certain period it cranks up it's computation. If he was top of SETI@home then the computers were definitely up 24/7.

Running MS Windows on computers results in wasted electricity.

Reply on news.ycombinator.com