Network Effects | Part I

INTRODUCTION

Since every microsecond nanosecond counts in our life, at the present time I have to spend each microsecond to other works rather than writing here about high-frequency trading and/or the world of microwave. That’s the reason I decided to offer the opportunity to different people involved (or not) in the HFT industry to write here. I thought the motto of this blog may become “Where capital markets speak” but another organization already took the headline – too bad. Anyway, I’m happy a couple of people agreed to replace me. Here is the first part of a new series, Networks Effects, written by my first guest I’m delighted to welcome here. I only slightly edited the text (a couple of photos, a few links, etc.) and added comments within [brackets], signed SIM. Happy reading.

 

NETWORK EFFECTS | PART I

Early last month, a reporter called me with a curious question.

“I’ve learned that Jump recently purchased a 31-acre field just across from the CME’s match in Aurora. And they paid a lot. Like $14 million, apparently. Any idea what could be going on?”

Prime Illinois farmland produces of order 200 bushels of corn per acre per year. At 4 dollars a bushel, Jump Trading LLC – the opaque high-frequency powerhouse, with the blank silvery website and the Twitter account with no tweets – could be looking at a cool $25K in yearly agricultural revenue from their parcel. 

“Which side of the data center?” I asked.

“It’s due North.”

Carteret, Mahwah, Secaucus, 350 Cermak and CoreSite all lie east or slightly southeast. North is not on the geodesics.

“Could be a shortwave set up?” I ventured, after a pause, but more or less completely at a loss. “Maybe they’re encrypting the E-mini ticks on one-time pads and broadcasting them worldwide. Tokyo, London, Frankfurt, Singapore, Sao Palo. Hey, are you on-line?”

He was.

“Pull up the Google Images of the `Rampisham Short Wave Radio Station’,” I said, “You’re looking for something like that; a field growing antennas instead of corn. If you drive out to Aurora and take pictures, it’ll give a sense of what they are up to.”

The next day, the photos popped up on my screen.

Source: Brian Louis/Bloomberg

Jump’s newly purchased field was a bleak unplowed expanse, a jumble of early spring weeds. In the distance, trucks are visible, roaring by on I-88. Fermilab is just over the horizon, having produced, for tiny flickering moments in the defunct 2 kilometer-wide Tevatron, temperatures that existed during the first microsecond of the Universe’s existence.

In the extreme southeast corner of the field stood a Kohler portable generator, a short, stubby pole with a millimeter-wave radio unit, and two small dish antennas pointing northeast.

***

More than a decade ago, I did know about Getco, and indeed, had marveled at the sheer audacity of that name. A friend of mine, a fellow physicist, worked at a 1990s-era desk that pioneered statistical arbitrage and early versions of high frequency trading. In that now-distant time, split-seconds, open phone lines, and human reaction times were all still relevant. I knew that servers were migrating to co-location in data centers to gain proximity to the exchanges. I’d heard the delicious rumors of the staggering PNLs that a collection of firms, mostly located in Chicago, were generating. “$1M per day”, “Effectively infinite Sharpe.” [Once I read in a court document that a big “HFT” firm in Houston, in its early stages, around 2001, had a Sharpe ratio of a 20 to 40, “which is extraordinarily high” – SIM].  In late 2007, a person, perhaps in position to know, and after one too many margaritas, let slip what sounded like a key to a castle, “ETFs, Man, we’re unpacking the ETFs and arbing ‘em against the underlying.’ A few weeks later, a successful trader, completely sober, remarked cryptically, “Order of magnitude, you need to think in dollars per millisecond.” Another, “Read Lefevre’s book. Everything in there still works.” It was a view through an imperfect, entirely incomplete window into a mysterious exhilarating world that was completely out of reach.

Can a sprawl of clues and misinformation be arranged to form a glimpse of how HFT actually works? A few simple principles appeared to interpolate consistently through the rumors. (1) If latency is an issue, then the time scale over which one’s predictions are valid has to be short. That is, if you can somehow predict where a stock is going to be priced next week, then your concern is with order working and market impact, not with microsecond latencies or nanosecond time stamps.  (2) There’s simply no time to solve complicated equations during production. In the co-lo cage it has to be about heuristics and lookup tables. (3) The Central Limit Theorem and the Law of Large Numbers will work to one’s advantage. Diffusion scales with the square root of time, so VIX of 20 deannualizes to price-changing tick on the E-mini every few seconds. So queue positioning must be critical, and market making, or its high-tech equivalents must be a core component of HFT.

***

The infamous Forbes article about Spread Networks was published in September 2010, but I didn’t see it until the first week of November, and so it’s clear that others had long since formulated all the thoughts I immediately had upon reading it: For a fiber that was so straight and so expensive, why was the Chicago to New York round trip travel time so excruciatingly slow? The distance is about 1200 km, so the minimum one-way signal propagation time should be about four milliseconds, not seven. Then I remembered from second-semester physics that the index of refraction in glass is something like n=1.5… Confusion.

Isn’t it faster and cheaper to send the signal through the air?

 A crash Google course in radio-frequency communication – a topic that I knew literally nothing about – suggested that there would be no show stoppers to either relaying a signal with a chain of microwave radios, or, at almost zero cost, using meteor burst communication. After convincing myself that I’d done enough due diligence to not make a fool of myself, I sent an e-mail floating those two schemes to a retired communications engineer whom I knew slightly. A week or so later, he wrote back.

I think that the idea of using a radio link seems a good idea and could work, but I believe there may be some problems with doing it by meteor scatter. To my way of thinking, a standard microwave relay link chain may be the best bet, and there’s a possibility of using a two-hop tropospheric scatter link.

Keep in mind that radio has two separate issues: technical feasibility and obtaining use of the spectrum. For something like this, I’d guess cost would not be overwhelming. Another issue is whether the protocol used requires a response from the distant end, or whether that can be eliminated through some cleverness so the latency is essentially “open loop.”

First, about meteor scatter. The standard meteor scatter link has the unusual characteristic that the reflection from the ionized line doesn’t have the same geographic coverage as an ionized plane would, and so the number of useful meteor trails between two locations is greatly less than the total rate of meteors. Typical links are very narrow band, partly because they are generally low power. So a higher power link might resolve the burst rate issue, but even then, I believe there would not be sufficient aligned paths to provide real-time service.

A more sure-fire way would be to use the relay sites that were once employed in the old AT&T Long Lines 4 GHz and 6 GHz networks.

Many of these were up for sale a while ago, and I’d guess they have been acquired by outfits that lease tower and building space. In this situation, you could expect to have reasonable bandwidth, and the latency would of course be determined by the velocity of light in air, plus a little for the plumbing and circuitry. Modern equipment for this service is available, as are the antennas and feedlines. Because of the greater bandwidth and flexibility, and lower cost, of optical fiber, there hasn’t been a market for this particular technology. But I think the Chicago-New York one-way latency would be in the range of 4 ms for the 700 mile distance. See http://long-lines.net/places-routes/1st_transcon_mw/LL0949/02.html

There are unlicensed 5 GHz bands that could provide this service, or it could be in the licensed bands of 4 and 6 GHz. There would possibly be some fade outages, but the system could be engineered against that.

Another possibility is a two-hop troposcatter link. This requires big antennas and high power, but that’s all been developed for satellite earth stations. There are 20 MB military systems, but the maximum range per hop is 400 miles, so you’d need an intermediate point. Given the reliance on the troposphere, there can be some outages, but the availability might be enough to be useful.

This reply seemed encouraging. I kept digging. Being two-hop, and nearly geometrically optimal, the troposcatter scheme looked like it would be the fastest option. A survey of the troposcatter provider websites, however, showed infrastructure that invariably ran to armored Humvees, ominously large dishes, shipping container-sized generators, and camouflage netting. The prospects of orchestrating the parking of such rigs in downtown Chicago seemed remote.

Microwave dishes on communication towers, however, are ubiquitous. They are so much a part of the landscape that one hardly notices them [that’s the reason I got interested in the HFT microwave world – SIM]. I sent out more inquiries, floating various ideas. Looking back through my notes, I have a printed-out e-mail from an engineer friend, dated December 8, 2010, that was remarkably prescient:

Very interesting.  It got me thinking of other ways to pull it off.  Here’s an idea that is even less far-fetched. You could just use unlicensed spectrum, such as that used for Wi-Fi, and set up a mesh network and the number of hops you need.

I did some poking around online. There’s a company that sells spectrum license-exempt base-station equipment with up to 30km range.  The main purpose is enterprise or municipal/rural Wi-Fi networks.  But with that kind of range you could conceivably set up around 40ish hops to get from NY to Chicago.  I’m not sure how much latency would be introduced on each hop, but I’m guessing no more than 20us.  At my previous job I worked on hardware processing of the entire network stack (TCP/IP) and the internal latency was well under 5us.  Using a conservative figure of 20us per hop adds an additional 800us (1.6ms round-trip) to the 8ms speed-of-light time.  Still brings you in under 10ms total, which compared to the 13ms in the article is certainly a competitive advantage.   Also, the bandwidth should be respectable, something on the order of megabytes.  Nothing like fiber, but certainly enough to communicate trade information for some limited set of symbols.  The other advantage would be that it is somewhat scalable.  You could build multiple point-to-point hop links.  Since this stuff is line-of-sight, you could easily have parallel networks along very similar paths.  This would not only increase bandwidth, but add redundancy and reliability to the network.  The cost of this solution would be pretty minimal if the available products have reasonable performance.  I found another company, American Tower, that rents antenna tower space for cellular and radio (think of it as antenna hosting, similar to data-center hosting).  It seems like all the pieces are there to just go out and slap a solution together.  I’ve got to think the big players have done this, but again it’s one of those things that you could read about in the paper some day and think holy shit, we talked about that! After all, we talked about the tunnel to Chicago, and it turns out someone built it. And the minute it gets out that someone has done it, there would be a flood of people doing it, since it’s low-cost and straightforward.  The Wi-Fi spectrum along the straight-shot from NY to Chicago would get so crowded that nobody in rural Pennsylvania and Ohio would ever be able to use their Wi-Fi.  I’m sure at that point the government would have to step in.  So even if someone is using Microwave, I’m guessing the Wi-Fi idea is less likely.  There’s nothing liking hiding a top-secret super-low latency network in plain sight!

My last notebook entry from 2010 is a fragment with no attribution that had been forwarded, as if by a two-hop troposcatter link, from someone who was reportedly working in the industry.

Actually… There is a Chicago firm that uses microwave to go over a town/area on the way from Chicago to NYC. (apparently some telco firm controls the switches in this area and they would have to go around so they decided to go over) supposedly their times are 13.9 vs 15 for the Spread Network direct line (and 16.5 for regular line)

Something about that last entry must have seemed confusing and daunting, and my interest drifted elsewhere for the next ten weeks.

A we say in French: “à suivre…”
Advertisements

HFT in the Banana land | Part 7

This unexpected new episode of the “HFT in the Banana land” series may also be coined “HFT in my backyard | S02E01”, or “S22E45”, or in more simple terms “Going to the Signal to get signals faster”. This blog has been silent for a while as I really didn’t have the time to write here, but that doesn’t mean there is nothing to say (in fact, there is a lot to write about but as I’ll explain later, I decided to withdraw the season 2 of “HFT in my backyard” as I’ll write a whole book about all my investigation on the wireless networks the HFTs love). But I couldn’t resist to add this new episode of “HFT in the Banana land”, even if it’s not about the banana – but a consequence of what happened in Richborough where the two giant masts Vigilant and New Line Networks wanted to erect there were harshly refused by the Dover District Council (read the previous episode).

Here is the story as I understand it. Let’s start with a general view of the current situation, from Frankufrt to Slough. I take the Vigilant route as the Canadian firm was the first to bid in Richborough to get this beautiful nearly perfect straight line between Frankfurt (FR2 data center) and Slough (LD4 data center):

Entete

Here are the details of the Vigilant towers…

Click to enlarge

… one by one, starting from Frankfurt :

Capture d_écran 2017-05-23 à 10.27.37

From Frankfurt the path goes to Weibern where there is a high tower perfectly located close to the straigh line between FR2 and LD4 (the good old German industry). Then from Weibern the signal goes to Simmerath, at the Belgium-Germany border:
It seems most of the microwave players (Vigilant, Optiver, McKay Brothers, etc.) put dishes in Simmerath. The wikpiedia page about this tower is amusing. It says: “Am 6. September 2013 wurde der Sendeturm aufgestockt. Der genaue Grund für die Aufstockung sowie die aktuelle Gesamthöhe sind nicht bekannt. Bis dahin betrug die Gesamthöhe 66 Meter.” In short: “in September 2013 the tower was expanded”, as shown on this photo I found in a German newspaper:

The wiki page states that “the exact reasons why the tower was expanded are not known”… but the explanation is quite simple: there was not enough space anymore for all the HFT players, so the tower was adapted to receive all the dishes, as shown on this (wiki) picture took just after the tower was rebuilt: 

HFT dishes colocated in Simmerath

From Simmerath the Vigilant signal goes to Welkenraedt and then to Tienen:

You can watch the Vigilant dishes in Welkenraedt on the antennasite; in Tienen they are probably on this strange water tower:

From Tienen the path goes to Merchtem and Egem:

Since the Merchtem tower is located 6 kilometers from my new home in Brussels, I paid a visit to this tower months ago, and it looks like that:

Here is now the Egem tower, which is the tallest structure in Belgium (if you want to know more about Egem, read again this old post about my trip there, where I wanted to have a coffee in a place which appeared to be a sadomasochist brothel (?!).

From Egem, the ideal Vigilant route would go to a building in Oostduinkerke (ideally located very close to the straight line between FR2 and LD4), and then to Richborough (cf. the previous episodes of “HFT in the Banana land”)…

… but the Vigilant mast in Richborough was refused. The Egem-Oostduinkerke-Richborough path is a dead end (at least for now ;). That means the Vigilant route, from Egem, has to go South to Dunkerque (or to Coudekerque-Branche, where Vigilant has dishes on a tower) before going North to UK (in Tilmanstone, where Vigilant will expand a small tower to cross the Channel – from the grain silo where McKay has dishes in the port of Dunkerque?):

This is the kind of path HFTs don’t like: not a straight line but a triangle, meaning they have to make a detour and that’s not good for latency. The most important detail here is the high-frequency traders can’t cross the Channel with a path close to the straight line (Oostduinkerke-Richborough); they have to do it from (around) Dunkerque (or Houtem, where Jump purchased a now-very-famous giant tower for $5M). The consequence is: if you want to build the best/fastest network from FR2 to LD4, you need to have a straight path from Frankfurt to (around) Dunkerque (the big path in red in the picture below).

This is what the competitors name the “South route”, opposed to the “North route”. The North route was annihilated by the decision of the Dover District Council about Richborough, so the operators need to improve the “South route”. Here is the real beginning of this episode.If you want to build the best path from Frankfurt to Dunkerque (the “South route”), forget Simmerath (which is too far north). You have to find another tower further to the south, inside the square in red in the picture above:

And you know what? There is such a tower there: the famous Signal de Botrange. The Signal looks like that:

In Belgium the Signal de Botrange is very famous for many reasons. First, it’s the highest point of the country: 694 meters above the sea level. Secondly, it’s one of the few ski resorts of Belgium. Thirdly (in my opinion), this is one of the most beautiful places of Belgium. I went there a couple of times to explore the “High Fens”, or “Hautes Fagnes” in French, a very nice nature reserve. 

There is an amazing story about Botrange: for a long time there was nothing in the highest point of Belgium (actually it’s the highest point of both Belgium, Luxembourg and Netherlands), but in 1923 the local governor Herman Baltia decided to erect a 6 meter stairs :“Apparently, clocking the country’s highest point in at just shy of 700 meters was simply too tantalizing, and so in 1923, a 6-meter (18-foot) stone staircase was constructed atop the peak. From its height, on the top step of the seemingly nonsensical staircase, visitors can survey the land from exactly 700 meters above sea level.” The ridiculous thing looks like that:

Eleven years later, in 1934, an observation tower was built, reaching a height of 718 meters above sea level: the Signal de Botrange. Now here is the story of the HFTs around Botrange. I don’t remember who talked to me, three years ago, about a new project of mast in Botrange reated to HFT but last December I have been asked by a firm not to share (here on this blog) the public documents I would find about this project, as the firm didn’t know how many competitors would be aware of what was going on. I said “ok” and instantaneously started to find intel about a story I could not tell about until yesterday. I was surprised when I found this video interview of the local mayor made by a local TV:

Note that the interview as made in November 2013, and the title of the video is “A new 50-meter tower to suport financial operations between London and Frankfurt”, and I think this is the very first press coverage in Belgium on the microwave networks (too bad I missed it before). In short the mayor says that the local community have been contacted by a Canadian firm (that smells Vigilant), by an American firm (that smells Jump, now New Line Networks), by a consortium including French, Dutch and German firms (the French one should be McKay Brothers, the Dutch firm is strange – that could be Optiver or Flow Traders, but both the Dutch firms don’t work with McKay…). The mayor explains the different competitors wanted to erect a towers in Botrange because the fibre is too slow, so HFTs use microwave, etc. At least the local people know exactly why some firms need new towers… The mayor also says that a request for proposals would be made as the total budget is €M2-3. It’s not clear if the mayor states that all this money would be distributed to the local community, but he seems to be very happy with the project as the new 50-meter tower would be higher than anything in Botrange. A new record!

So, back in 2012, some HFT firms were already lurking around the Signal de Botrange. It seems it took quite a long time before the request for proposals was released… on March 3, 2017. The public documents are here. Among the usual boring administrative stuff (but where we learn that the minimum bid amount is €50,000 a year), there is a map of the Signal de Botrange with the location of the future tower:

 

I did a quick simulation with Google Street (of course, the winner will have to put all the different telco dishes which are around the Signal on the new tower). 

The competitors had until last Friday (May 19th) to submitt an offer, and yesterday the Collège communal (the municipal college) started to open the enveloppes. I called the local office to know more about what’s going with the offers, but the clerk told me that they can’t communicate now as the Conseil communal (an other administrative college) will have to discuss the offers too, etc. I was not even able to know how many offers were sent to Botrange. But the local authorities should decide soon about the next dead-lines. Who will be the winner? One can assume the Canadian firm, the American firm and the French firm submitted offers but perhaps there are other HFT lurking around Botrange too… We will see. Anyway, that’s amazing the HFTs want to be on the Signal to get signals faster. Thanks to HFT, soon the highest point of Belgium will be higher than ever, and that’s pretty cool for such a small country!

In the meantime, on the other side of the network, in Houtem, @Nuklearexperte had fun by peeing on Jump’s ground. That is what happens when a man drinks too much Belgian beers…

 

HFT in the Banana land | Part 6

 

Here is the final episode of HFT in the Banana land – or The Case of the Small English Town vs Two Large HFT Towers. No happy end here for the “high-frequency trading”/network providers firms Vigilant Global and New Line Networks who seeked to erect two giant masts in Richborough to improve the Slough-Frankfurt microwave route. Last Thursday the Dover District councilors followed the planning officer’s recommendations and refused both the masts. An interesting moment (sorry for the typos when I was live-tweeting the meeting). It’s important to understand here that the debate about the mast was not about market structure, speed, arm race, the need to save a few microseconds to do arbitrage (or whatever), etc. A comment on Twitter said that “If only SEC and int’l regulators did similar diligence as Dover city council when exchanges want to sell speed” but the meeting in Dover was only about planning applications: the councilors had to decide about the erection of a 2-metre high boundary fence, the installation of a car park charging machine, the erection of a verandah, the formation of a “Juliette-style” balcony, and in the midst of all this, they had to decide about two heavy quasi-identical construction projects in Richborough. I was right to follow the more-than-one-year story of these masts as the most interesting here is the way HFT activities could have had consequences on the daily life of the people (living around the Banana land).

img_20170126_180058

 

The Chairman (and his group) was at the centre of the scene, whereas the different councilors were around. On the left side you had chairs for the people who wanted to speak “against” an application, and on the right side you had (far less) chairs for those who wanted to speak “in favour”. For each application the DCC allows only one person to speak out against a project, and one person only to speak out for. Each of them had only 3 minutes to provide its views (that’s short). Unsurprisingly, when a person defended the erection of a 2-meter boundary fence, the nearest neighbor asked to speak against the fence. That said, expected when two women had an argument in the right corner of the room, the neighbourhood quarrels were peaceful. Everyone had precise arguments on each applications, and most of the time the councilors followed the officers recommendations. The meeting started at 6:00 pm and half an hour later the council decided to make a short break before discussing the Vigilant and NLN masts. Then the planning officer who was in charge came in the room with two large folders containing all the documents filed by the applicants (the documents I have been reading patiently for a year).

img_20170126_203037

The officer started with the Vigilant mast, which was filed first on January 2016. The room was very quiet, the Vigilant team, the McKay Brothers guy, the New Line Networks guy, me, Bass de Banaan and other neighbors – everybody was silently listening to the officer’s speech.

hdr

The DDC officer described the Vigilant application (the banana land area, the need for an optical line-of-sight to transfer data between Slough and Frankfurt, the 322-meter mast, the fact other towers would deviate too far from the direct line between Slough and Frankfurt, etc.) and quickly raised issues about the consequences of such a giant tower. In short, the officer summarized what he wrote in its report: most of the people living there are against the mast; it would be an eyesore in the landscape; the fact a lot of administrations/institutions disagreed with Vigilant’s statements regarding the ecological issues, the impacts on the Roman Fort and on the possible reopening of the Manston airport, the new National Grid project, and so on. The officer also talked about the St Peter church in Sandwich. He didn’t detail but St Peter is a good story: “The Council’s heritage officer has considered in more detail the setting of St Peter’s Church in Sandwich. It is identified that the Dover District Heritage Strategy defines churches as being of outstanding significance, and notes that such buildings have value in their contribution to the aesthetics of the historic landscape and wider rural environment; it states ‘the spires of rural churches can often be seen over long- distances and are recognised and valued local landmarks’.” For a very long time the tallest buildings, in the Western World, were church towers, and they needed a line-of-sight network. The officer: “From the viewing platform of St Peter’s Church, there are far reaching views to the north towards the Church of Saint Mary in Minster. In this view the Proposed Development would draw the eye and detract from the inter-relationship between St Peter’s and Saint Mary.” Here we have a battle between two line-of-sight paths: one between two old churches, the other between two telecommunication towers: “there would be some harm, within the less than substantial range, to the significance of the setting of these churches” wrote the officer. Don’t touch God’s network.

capture-decran-2017-01-30-a-10-43-50

Then the officer went further in his remarks: the fact Vigilant and NLN did not agree on one mast is not a good sign, and all the benefits promised by the applicant “are minor benefits compared to issues involved by such a giant mast”. At this point it didn’t sound good for Vigilant. The officer concluded and the first speaker was invited to talk. As far as I understood, he is a local resident, a retired engineer, and he did a weird 3-minute speech against the Vigilant mast, saying that HFT is “computerized gambling only”, that this kind of activity could harm the market like the subprime crisis did, at some point he cited the Knight Capital debacle in 2012 – and that is ironic: the speaker said such a mast would improve speed and would have bad consequences (a new Knightmare) but he seemed to ignore that the Vigilant competitor here, New Line Networks, is a joint venture between Knight-Getco (known as KCG now) and another trading firm, Jump Trading. Anyway. After that, a Vigilant representative had his 3-minute speech and advocated for the mast (in short: “we have been working on this project for 18 months, we proposed NLN to share but they refused, there is no problem with the Fort, this mast would be temporary [20 years though], we offer a lot of contributions, we choose the area to mitigate the consequences [true], etc.”.

First reaction of the officer: “I don’t change my opinion”. Then a first councilor said: “Seriously, what does that look like?” (i.e. such a mast is ugly), another talked about the major impacts the tower would have on the landscape, another talked about the consequences  on the Manston airport, another talked about “layering” and, in short, all the councilors said they don’t get the real local (and national) benefits such a mast could provide to the community. “National benefit is minor, public benefit is more important than private [HFT] benefits” stated the DDC chairman. The ten councilors voted and the hammer dropped: 9 against the mast, and one abstention. Boom. Vigilant had the best case, they worked hard to comply with all the issues raised by the DDC, but it was not enough. End of the story. Harsh.

Then came the New Line Networks mast. I won’t detail as the officer made more or less the same statements. The best part was when he said NLN stated “The tower in Houtem is not enough to achieve a perfect route between Slough and Frankfurt” [at this point I thought: “so, why Jump spent $5,000,000 to buy the Houtem tower if it’s not that perfect?”]. Also (and this is an important legal point), the officer emphasized the fact NLN didn’t properly understand the purposes of an obscure thing named “S106” (Bass de Bannan was 100% right on that) and the legal issues about the Community Interest Company (which would collect the money from NLN before flowing it through the local institutions). What the officer was implying is that NLN didn‘t da good job (the way NLN filed a last 40-page document two days before the meeting, answering to questions raised over the last several months, proves that).

Then came what was probably the funniest moment of the meeting (at least for the councilors). Guess who asked to speak against the NLN mast? Vigilant of course, and the councilors had a good laugh as the Vigilant representative was now speaking from the left corner (against NLN) whereas he was sitting in the right corner minutes before (advocating for its own mast). In short, the Vigilant representative criticized the NLN mast with (sometimes) the same arguments the DDC officer had against the Vigilant mast minutes before. Weird moment, even if Vigilant had good points. Then the operations director of New Line Networks, who came from Chicago, had his 3-minute speech to defend its mast. I didn’t envy his position: the Vigilant mast was just wickedly refused by the DDC and he was trying to convince that a similar mast in Richborough would be important for the local and London/FinTech communities in a post-Brexit context. Trying to save a few microseconds in 3 minutes, in front of an angry crowd, is hard. Then the DDC chairman said “we refused the Vigilant application for these reasons, we should refuse this one too”, a Labour councilor had ironic words about the fact the two applications were not exactly the same (laughs in the room), the councilors voted and a new hammer dropped: 10 councilors against NLN, and not even one abstention. Zero.

capture-decran-2017-01-30-a-14-11-22

The situation around the Channel, January 17

In a way, Vigilant won: they got at least one councilor who didn’t vote against their mast – one point for the better work they did. But overall the “HFT” firms got a slap. Vigilant Global and New Line Networks wanted to erect two giant dicks in the middle of nowhere to save microseconds, and the Dover District Council responded with a “Good bye” (I liked the smile the officer). They just don’t want this thing in their backyard. Landscape versus money. It was too much. Here is the end of the HFT in the banana land spin-off. I assume competitor McKay Brothers (and other HFT firms) are happy. Of course the firms can appeal. I don’t know if they will but with what happened in Dover last Thursday, making appeal for two masts would be indecent and they would certainly get another slap. Vigilant is now walking around Tilmanstone, other are improving their networks elsewhere – latency depends on lots of factors. It’s time to leave Lotus corniculatus alone. After the meeting I ended up in a McDonald’s restaurant (the only one around the DDC offices) with brother McKay and the Vigilant team. They are nice and smart people, I’m sure they’ll find solutions other than putting a 300-meter mast between two churches. Good luck.

img_20170127_131738

Find the HFTs