From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp86.iad3a.emailsrvr.com (smtp86.iad3a.emailsrvr.com [173.203.187.86]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id F1F323B29D for ; Sun, 12 Apr 2020 12:31:43 -0400 (EDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=g001.emailsrvr.com; s=20190322-9u7zjiwi; t=1586709103; bh=oU/pLjSXEgqo1nOY9SCRP7prOlgsU29A6hbbppa8NpM=; h=Date:Subject:From:To:From; b=gtZmKUu4ShSQokhiUNWPvKDYKsWCztLUnRej1g1P8z5jzu5CL826GPGfujn3MIeCC amtlN0zV9zev2sQntb0JmKnNI+BM6XIpO9ZiHyW5UZ/Dz9b1FzSiV5LNdrRxjgJ9zi pdkNgmTrVjF7YiWnwX5MyU0mrurbNUs90LK4fCfA= Received: from app37.wa-webapps.iad3a (relay-webapps.rsapps.net [172.27.255.140]) by smtp19.relay.iad3a.emailsrvr.com (SMTP Server) with ESMTP id 9A6FC378A; Sun, 12 Apr 2020 12:31:43 -0400 (EDT) X-Sender-Id: dpreed@deepplum.com Received: from app37.wa-webapps.iad3a (relay-webapps.rsapps.net [172.27.255.140]) by 0.0.0.0:25 (trex/5.7.12); Sun, 12 Apr 2020 12:31:43 -0400 Received: from deepplum.com (localhost.localdomain [127.0.0.1]) by app37.wa-webapps.iad3a (Postfix) with ESMTP id 8753D6004A; Sun, 12 Apr 2020 12:31:43 -0400 (EDT) Received: by apps.rackspace.com (Authenticated sender: dpreed@deepplum.com, from: dpreed@deepplum.com) with HTTP; Sun, 12 Apr 2020 12:31:43 -0400 (EDT) X-Auth-ID: dpreed@deepplum.com Date: Sun, 12 Apr 2020 12:31:43 -0400 (EDT) From: "David P. Reed" To: "Dave Taht" Cc: "Make-Wifi-fast" MIME-Version: 1.0 Content-Type: text/plain;charset=UTF-8 Content-Transfer-Encoding: quoted-printable Importance: Normal X-Priority: 3 (Normal) X-Type: plain In-Reply-To: References: <1586653525.57076088@apps.rackspace.com> Message-ID: <1586709103.55072209@apps.rackspace.com> X-Mailer: webmail/17.3.5-RC X-Classification-ID: 6e20393c-5a16-4900-834f-6838bea687ba-1-1 Subject: Re: [Make-wifi-fast] =?utf-8?q?Anyone_got_a_view_on_this=3F?= X-BeenThere: make-wifi-fast@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 12 Apr 2020 16:31:44 -0000 Thanks, Dave. In explanation, rather than defense of MIT research in this c= ase, I'd like to say one thing.=0A=0AThe pressure to hype ones results toda= y is astronomical for professors, especially pre-tenure, but even post-tenu= re. Compared to my time at MIT's Project MAC that became the Laboratory for= Computer Science, and then much later, CSAIL, you could get tenure and res= pect merely by publishing one really good paper a year at SOSP, ASPLOS, ...= Today, you have to have your name on 5-10 student-produced papers per year= , get grants or contracts to support all your students renewed every year, = ... and hopefully find grad students to teach all your classes for you.=0A= =0AThe NIT News Office is huge today. It's people maintain their jobs by fi= nding and promoting any little timely idea. (a few years ago, MIT publicize= d "wireless power" as a breakthrough, without any technical details of the= limitations imposed by physical reality on the idea of transmitting power = using a 1 MHz RF radiator and pickup in a room. But hey, a lot of reporters= picked it up and put MIT's name in the papers.=0A=0AThis is because there = is no funding for academic research, basically. Neither industry nor govern= ment thinks they need to pay for this stuff. So instead, a travesty has bee= n made of much research in academia by the need to self-promote. And "every= one does it".=0A=0AThe result is that zillions of results are published und= er the good name of MIT or other institutions, but the reader has to evalua= te for themselves whether the results are useful.=0A=0AI hate this. But I h= ave a lot of sympathy, a great deal of sympathy, for those trapped in that = system.=0A=0AI'm lucky. I can do my own research in my own home lab. If I w= ant, I can publish on the Internet, for free. But I only feel a need to do = that if the result is worth publishing. I don't have to prematurely publish= a half-baked idea and get it through skeptical referees. Instead, I can ju= st polish and give a finish to the ideas so it is worth reading about.=0A= =0AThat was what publishing a computer science paper used to be about. Ther= e was about one conference that mattered in one's field - in my case SIGOPS= . And only really good papers got published. If you couldn't publish there,= you didn't. And you could still get tenure.=0A=0AOn Saturday, April 11, 20= 20 10:05pm, "Dave Taht" said:=0A=0A> On Sat, Apr 11, = 2020 at 6:05 PM David P. Reed wrote:=0A>>=0A>> https:= //techxplore.com/news/2020-04-wireless-networks.html=0A> =0A> MIT still is = good at PR, unlike us.=0A> =0A> Yer the one that handed them their ass on y= our prior paper review.=0A> (and me, privately, I handed 'em their ass on t= he quality of their=0A> result vs a vs modern fq_codel on wifi (post "endin= g the anomaly"),=0A> since they didn't compare against the real deal, seemi= ngly=0A> specifically picking a kernel that pre-dated our work - and I view= the=0A> trace results they use as entirely useless). the actual paper=0A> = presented is probably this one:=0A> https://www.usenix.org/system/files/nsd= i20-paper-goyal.pdf=0A> =0A> I only skimmed it briefly. no running code, no= n-reproduceable results, traces...=0A> =0A> Rather than compare against fq_= codel, they ripped out a real wifi=0A> emulation entirely, and did codel wi= th cubic rather than fq_codel.=0A> They got a glow on from getting publishe= d after multiple tries. And=0A> then there's the totally artificial test se= tup...=0A> =0A> "Our emulated cellular network experiments used a mini-mum = RTT of 100=0A> ms and a buffer size of 250 MTU-sized pack-ets. Additionally= , ABC=E2=80=99s=0A> target rate calculation (Equation(1))used=CE=B7=3D0.98a= nd=CE=B4=3D133ms. Our Wi-Fi=0A> implementationuses the link rate estimator = from =C2=A75, while our emulated=0A> cel-lular network setup assumes the re= alistic scenario that=0A> ABC=E2=80=99srouter has knowledge of the underlyi= ng link capacity [1"=0A> =0A> Uh, huh.=0A> =0A> PLEASE NOTE: That said, I l= ike the simplifying conceptualization of=0A> ECT(1) equals accellerate in a= bc, *with* a transport that thinks that=0A> way. it's sort of inbetween the= SCE concept and the L4S DCTCP concept=0A> and may well have a bit of promi= se. But lacking code...=0A> =0A> One of these days having the war with real= results with real tech vs=0A> academia might seem worthwhile, but, me, I'm= still got a glow on from=0A> actually seeing the ath10k finally behave.=0A= > =0A> https://forum.openwrt.org/t/aql-and-the-ath10k-is-lovely/59002=0A> = =0A> only cost us 4 years of life, that.... so many other chips to fix...= =0A> =0A> I really, really, really used to have way more respect for MIT an= d=0A> academia than I do today. I really wish they'd produced results=0A> a= gainst our more modern openwrt implementation and published source=0A> code= . As it is it feels like they are not advancing or being true to=0A> the go= als of real science. For all I know, by us knocking it so far=0A> out of th= e park with "ending the anomaly", no academic will cite it or=0A> compare a= gainst it because they can't do better. Just ignore it. Am I=0A> grumpy abo= ut this? you betcha... do I feel like writing a paper or=0A> letter to the = editor? I don't really know the right way to fight an=0A> academic battle. = Never learned. What's the right response? "Dear=0A> USENIX, you do your rea= ders a disservice by allowing a paper that made=0A> claims about wifi to no= t do any coherent comparison with "ending the=0A> anomaly", published a few= years prior, containing code that seems to=0A> outperform it by a large ma= rgin, and the current default in the linux=0A> kernel"?=0A> =0A> Prefer jus= t to ship code....=0A> =0A> I kind of feel that way also when stanford keep= s insisting on still=0A> teaching CS244 with a ludicrously small buffersize= relative to what's=0A> being observed in the field, also. And then people = clone that course=0A> thinking it represents some reality. And the recent b= uffersizing=0A> workshop... oh, don't get me started. It's makework. Keepin= g a problem=0A> alive to keep the grant money coming.=0A> =0A> =0A>>=0A>> _= ______________________________________________=0A>> Make-wifi-fast mailing = list=0A>> Make-wifi-fast@lists.bufferbloat.net=0A>> https://lists.bufferblo= at.net/listinfo/make-wifi-fast=0A> =0A> =0A> =0A> --=0A> Make Music, Not Wa= r=0A> =0A> Dave T=C3=A4ht=0A> CTO, TekLibre, LLC=0A> http://www.teklibre.co= m=0A> Tel: 1-831-435-0729=0A> =0A