An Indiana prison inmate named Jeffrey Rowe, the plaintiff in this suit under 42 U.S.C. § 1983, charges administrators and prison staff (actually employees of Corizon, Inc., which provides medical services to the inmates at Pendleton Correctional Facility, Rowe’s prison) with deliberate indifference to a serious medical need — that is, with knowing of a serious risk to inmate health or safety but responding ineffectually (as by departing substantially from accepted professional judgment) or not at all. See, e.g., Farmer v. Brennan, 511 U.S. 825, 837, 114 S.Ct. 1970, 128 L.Ed.2d 811 (1994); Sain v. Wood, 512 F.3d 886, 894-95 (7th Cir.2008). Such conduct was held in Farmer to violate the cruel and unusual punishments clause of the Eighth Amendment, deemed applicable to state action by interpretation of the due process clause of the Fourteenth Amendment. Rowe charges gratuitous infliction of physical pain and potentially very serious medical harm — cogent examples of cruel and unusual punishment. He has a subsidiary claim of having been retaliated against for filing this lawsuit, a claim we discuss briefly toward the end of our opinion. The district judge granted summary judgment in favor of the defendants on both claims, dismissing Rowe’s suit and precipitating this appeal.
In 2009, already an inmate at Pendleton, Rowe was diagnosed with reflux esophagitis, also known as gastroesophageal reflux disease (GERD). See National Institutes of Health, “Gastroesophageal reflux disease,” www.nlm.nih.gov/medlineplus/ency/ artiele/000265.htm (visited August 17, 2015, as were the other websites cited in this opinion). The Mayo Clinic explains that “a valve-like structure called the lower esophageal sphincter usually keeps the acidic contents of the stomach out of the esophagus. If this valve opens when it shouldn’t or doesn’t close properly, the contents of the stomach may back up into the esophagus (gastroesophageal reflux)---- [GERD] is a condition in which this backflow of acid is a frequent or ongoing problem. A complication of GERD is chronic inflammation and tissue damage in the esophagus.” Mayo Clinic, “Diseases and Conditions, Esophagitis: Reflux Esophagitis,” www.mayoclinic.org/diseasesconditions/esophagitis/basics/causes/con20034313. As we explained in a recent case in which, as in this case, a prison inmate complained of failure to treat his GERD (and we reversed the grant of summary judgment in favor of the prison staff), “GERD can ... produce persistent, agonizing pain and discomfort. It can also produce ‘serious complications. Esophagitis can occur as a result of too much stomach acid in the esophagus. Esophagitis may cause esophageal bleeding or ulcers. In addition, a narrowing or stricture of the- esophagus may occur from chronic scarring. Some people develop a condition known as Barrett’s esophagus. This condition can increase the risk of esophageal cancer.’ WebMD, Heartbum/GERD Health Center, “What Are the Complications of Long-Term GERD?” www. webmd.com/heartburn-gerd/guide/refluxdisease-gerdl?page=4.” Miller v. Campanella, 794 F.3d 878, 880, 2015 WL 4523799, at *2 (7th Cir. July 27, 2015). Rowe complains of pain based on neglect of his need *624for symptomatic relief; continued neglect will endanger him more profoundly.
The prison physician who diagnosed Rowe with GERD told him to take a 150-milligram Zantac pill twice a day. Zantac inhibits the production of stomach acid and is commonly used to treat esophagitis (as we’ll abbreviate the name of Rowe’s disease). Although technically “Zantac” is merely the trade name for ranitidine manufactured by GlaxoSmithKline (in prescription strengths) and Boehringer Ingelheim (in over-the-counter strengths), it is often used as a synonym for ranitidine, see Wikipedia, “Ranitidine,” http://en.wikipedia. org/wiki/Ranitidine, because Glaxo was the first, and remains the best-known, manufacturer. “Zantac” is the only word for the drug that appears in the briefs, and so we too will call the drug that Rowe received “Zantac.”
After the diagnosis Rowe was given Zantac pills and was permitted to keep them in his cell and take them when he felt the need to. This regimen continued for more than a year. But in January 2011 his pills were confiscated and he was told that he would be allowed to take a Zantac pill only when a prison nurse gave it to him, and that would be at 9:30 a.m. and then at 9:30 p.m. He complained that he needed to take Zantac with his meals, which were, oddly enough, scheduled by the prison for 4 a.m. and 4 p.m. (why these times, we are not told). The prison had decided that inmates such as Rowe who take psychiatric medications should not be allowed to keep any pills in their cells — yet the head of health care at the prison told Rowe that he could keep in his cell (and thus take whenever he wanted) any Zantac pills that he bought at the prison commissary — which, however, as we’re about to see, he couldn’t afford. No reason has been articulated for forbidding him to keep Zantac given him by prison staff while permitting him to keep Zantac that he bought at the commissary and take it whenever he needs to in order to prevent or alleviate pain. There is no suggestion that Zantac is a narcotic or otherwise consumed for nonmedical as well as medical reasons.
The defendants question Rowe’s inability to pay for the pills. They point out that in one 13-month period he spent approximately $60 at the commissary. But the prison commissary charges $3.28 for just four 75-mg Zantac phis (and recall that Rowe was to take two 150-mg pills daily), meaning that he would have to pay almost $1300 for a 13-month supply. And he was forbidden to buy more than eight days’ worth of Zantac a month from the commissary, which was only about a quarter of the amount that he needed.
To continue the narrative of what seems a senseless series of decisions by the prison’s medical staff, as well as heartless given what the staff knew about the disease and Rowe’s continuous claims of severe pain: at the beginning of July 2011, a month after he filed suit, he ceased receiving Zantac because his “prescription” (that is, his authorization to receive over-the-counter Zantac free of charge on a continuing basis) had lapsed. He made a series of requests for the drug beginning on July 3, but the nurse defendants denied all of them because he had no prescription. When he complained he was told by the administrative director of the medical staff: “Your chronic care condition does not warrant the continued use of Zantac. The continual use of over-the-counter medications can create further health problems in many instances. You will have to purchase this off of commissary if you wish to continue taking it.” Notice the contradiction (illustrating the run around to which Rowe was continually subjected) in *625denying Rowe free Zantac because it could create “further health problems” but permitting him to buy and use it at will, though he couldn’t afford to buy it. Nor is there any suggestion that Zantac is one of the over-the-counter medications that can create health problems if taken daily for a protracted period of time. And finally, if over-the-counter medicines are to be barred, why wasn’t Rowe given a prescription for 300-mg Zantac pills;' these are not only prescription rather than over-the-counter drugs but one such pill a day may be sufficient to control one’s GERD, compared to two or more when an over-the-counter strength Zantac is prescribed.
On July 13, 2011, in response to Rowe’s continued requests for a renewed prescription for Zantac, a physician who works at the prison (though employed by Corizon) named William H. Wolfe, whose professional specialty is preventive medicine, about which see American College of Preventive Medicine: Physicians Dedicated to Prevention, www.acpm.org/, rather than gastroenterology, see healthgrades, “Dr. William H. Wolfe, MD„” www. healthgrades.com/physieian/dr-williamwolfe-2fgkl/background-check, and who is a frequent defendant in prisoner civil rights suits, reviewed Rowe’s medical records and opined that his condition didn’t require Zantac at all — this despite the fact that Rowe had been continuously prescribed Zantac for almost two years and that Wolfe himself had been the prescribing doctor for a quarter of that period. But though initially refusing to provide a new prescription for Zantac, Wolfe later relented and on August 2 prescribed it though he later stated in an affidavit that he had done so as a “courtesy” to Rowe and not out of medical necessity. (Prescribing drugs for prison inmates as a “courtesy” seems very odd; it is not explained.) The upshot was that Rowe had no access to Zantac for more than a month (between July 1 and August 3) — a significant deprivation. Even after Zantac was restored to him, he continued to be allowed to take it only at 9:30 a.m. and 9:30 p.m., both times being many hours distant from his meals.
In another affidavit Wolfe stated that “it does not matter what time of day Mr. Rowe receives his Zantac prescription. Each Zantac pill is fully effective for twelve hour increments. Zantac does not have to be taken before or with a meal to be effective.” However, according to Boehringer Ingelheim, the manufacturer of over-the-counter Zantac, while Zantac can be taken at any time “to relieve symptoms,” in order “to prevent symptoms” it should be taken “30 to 60 minutes before eating food or drinking beverages that cause heartburn.” Zantac, “Maximum Strength Zantac 150,” www.zantacotc.com/ zantac-maximumstrength.html# faqs, and this advice is repeated on the labels of the boxes in which over-the-counter Zantac is sold. Were Zantac equipotent whenever taken, the manufacturer would not tell consumers to take it 30 to 60 minutes before eating, for having to remember when to take a pill adds a complication that the consumer would rather do without. There is thus no reason for the manufacturer to be lying, and it would be absurd to think that Dr. Wolfe, a defendant who is not a gastroenterologist, knows more about treatment of esophagitis with Zantac than the manufacturer does.
Rowe’s aim was pain prevention, so having to take Zantac six and a half hours before a meal did not do the trick. It left him in pain for five and a half hours during and after the meal, until he got his next Zantac pill. Wolfe’s statement that “each Zantac pill is fully effective for twelve hour increments” is also contradicted by the Zantac website, which states that one 150-*626mg pill “lasts up to 12 hours” (emphasis added). Thus a pill taken six and half hours before a meal might not be effective in alleviating the pain caused by acid secretions stimulated by the meal.
It might be thought that a corporate website, such as that of the Zantac manufacturer, would be a suspect source of information. Not so; the manufacturer would be taking grave risks if it misrepresented the properties of its product. In any event, the Mayo Clinic’s website, as we’ll see in a moment, confirms the manufacturer’s claims.
Wolfe’s affidavit states that Rowe was complaining just of “alleged heartburn [that] was not a serious medical condition warranting a prescription for Zantac” — but if so why did he prescribe Zantac for Rowe during the very period in which, according to the affidavit, Rowe’s condition was not serious? (The affidavit fails to mention that it was Wolfe who had prescribed Zantac for Rowe, but that’s conceded.)
It’s true that the Mayo Clinic’s website, at “Drugs and Supplements: Histamine H2 Antagonist (Oral Route, Injection Route, Intravenous Route),” www. mayoclinic.org/drugs-supplements/ histamine-h2-antagonis1^oralrouteinjection-route-intravenous-route/proper-use/drg-20068584, after listing various drugs (including ranitidine) for treatment of the cluster of ailments that includes esophagitis, states that “for this class of drugs ... patients taking two doses a day are instructed: ‘Take one in the morning and one before bedtime.’ ” But this dosing, Mayo goes on to state, is appropriate “only for patients taking the prescription strengths of these medicines.” The 150-mg pills that Rowe was taking are available over the counter; a prescription is required only for the 300-mg version. Both the Boehringer Ingelheim and Mayo websites also say that the patient shouldn’t take Zantac for more than two weeks unless directed by a doctor — but Rowe was of course directed by Wolfe, as well as by other doctors earlier, to take Zantac on a continuing basis.
Not only wasn’t Rowe allowed to take Zantac with his meals; he was not, as the Mayo website recommends, allowed to take it with water a half hour or an hour before eating a meal or drinking beverages that might cause him esophageal pain. As the Mayo website explains, for “adults and teenagers-150 mg with water taken thirty to sixty minutes before eating a meal or drinking beverages you expect to cause symptoms. Do not take more than 300 mg in twenty-four hours” (emphasis added).
Stomach acid is of course integral to the digestion of food, and indeed thirty percent of total gastric acid secretion is stimulated by the anticipation, smell, and taste of food, before the food ever reaches the stomach. Thomas A. Miller, Modem Surgical Care: Physiologic Foundations and Clinical Applications 344-45 (2006). “The .foods you eat affect the amount of acid your stomach produces,” and “many people with GERD find that certain’ foods trigger their symptoms.” Healthline, “Diet and Nutrition for GERD,” www. healthline.com/health/gerd/ dietnutrition# Overviewl. So it is no surprise that Rowe experiences painful symptoms when he eats without having been allowed to take a Zantac pill shortly before the meal.
The Physicians’ Desk Reference, “PDR Search: Full Prescribing Information: Zantac 150 and 300 Tablets,” www.pdr.net/ full-prescribmginformation/zantac-150- and-300-tablets?druglabelid=241, states that a 150-mg dose of Zantac inhibits 79 percent of food-stimulated acid secretion for up to three hours after it’s taken. This implies that the drug’s efficacy decreases over time and so supports Rowe’s claim *627that a 150-mg dose does not suppress his food-stimulated acid secretions when taken six and a half hours before a meal. The Physicians’ Desk Reference also says that “symptomatic relief commonly occurs within 24 hours after starting therapy with ZANTAC 150 mg twice daily,” which could be misread to mean that it does not matter what time of day the pills are taken, but which actually means that it takes a day for the body to recognize Zantac as a source of relief from esophageal distress. This interpretation is confirmed by Mayo, which states (at the website cited earlier): “It may take several days before this medicine begins to relieve stomach pain.”
The evidence that Rowe was in pain for five and a half hours after eating is his repeated attestation — in his verified federal complaint and his declarations— that he experienced pain for that length of time when he was not allowed to take Zantac with or shortly before his meals. For purposes of summary judgment his attestations of extreme pain must be credited. See 28 U.S.C. § 1746; Fed.R.Civ.P. 56(c). There was no plausible contrary evidence. The affidavits of the only expert witness on the proper times at which to take Zantac, defendants’ witness Wolfe, were highly vulnerable. Wolfe is not a gastroenterologist. He says that Rowe didn’t need Zantac yet prescribed Zantac for him. He opined with confidence about what Rowe needed or didn’t need — yet never examined him — and offered no basis for his off-the-cuff medical opinion. A court should not “admit opinion evidence that is connected to existing data only by the ipse dixit of the expert.” General Electric Co. v. Joiner, 522 U.S. 136, 146, 118 S.Ct. 512, 139 L.Ed.2d 508 (1997); see also Finn v. Warren County, 768 F.3d 441, 452 (6th Cir.2014) (“the ‘knowledge’ requirement of Rule 702 requires the expert to provide more than a subjective belief or unsupported speculation”); Guile v. United States, 422 F.3d 221, 227 (5th Cir.2005) (“we look to the basis of the expert’s opinion, and not the bare opinion alone. A claim cannot stand or fall on the mere ipse dixit of a credentialed witness”); McClain v. Metabolife International Inc., 401 F.3d 1233, 1242 (11th Cir.2005).
Remember that Rowe had been diagnosed with esophagitis back in 2009 and that for the ensuing two years physicians had prescribed Zantac to treat his condition. Furthermore, the Indiana Department of Correction permits such continuous treatment only to treat a serious health condition, so presumably the prescribing physicians thought Rowe’s condition serious. None of this evidence or inference is undermined by Dr. Wolfe’s evidence.
A member of a prison’s staff is deliberately indifferent and thus potentially liable to an inmate if he “knows of and disregards an excessive risk to inmate health,” Williams v. O’Leary, 55 F.3d 320, 324 (7th Cir.1995), quoting Farmer v. Brennan, supra, 511 U.S. at 837, 114 S.Ct. 1970; see also Miller v. Campanella, supra, at 880, 2015 WL 4523799 at *2. Rowe makes two distinct claims of deliberate indifference; the evidence that we’ve reviewed tends to substantiate both. There is both evidence that defendants Wolfe, Deborah Dotson, Melissa Bagienski, Chris Deeds, and Lisa Gibson were deliberately indifferent to his pain when they denied him access to free Zantac for thirty-three days, and that defendants Mary Mansfield, Gibson, and Dr. Michael Mitcheff were deliberately indifferent to his pain when they insisted — for many months — on giving him Zantac only at 9:30 a.m. and 9:30 p.m., instead of at his prescribed mealtimes. Regarding the first claim, if the nurse defendants to whom Rowe complained about reflux pain were not author*628ized to give him the free Zantac they should have promptly referred the matter to a doctor.
The evidence of Wolfe’s deliberate indifference to Rowe’s pain and resulting need for Zantac is, as we’ve shown, substantial, and likewise the evidence that limiting Rowe’s taking Zantac to 9:30 a.m. and 9:30 p.m. for a protracted period exhibited deliberate indifference to a serious medical need. Wolfe never told anyone, so far as appears, when would be the best times for administering Zantac to Rowe. In very large doses Zantac will remain in your blood stream long enough to affect the stomach acid produced by meals eaten many hours later, but the Mayo and Boehringer Ingelheim timing recommendations suggest that this isn’t true for 150-mg doses. Wolfe’s assertion that “it does not matter what time of day Mr. Rowe receives his Zantac prescription” is implausible as well as vigorously contested. Rowe’s pain and the Mayo Clinic’s timing recommendations suggest that giving 150-mg doses of Zantac five and a half hours after one meal and six and a half hours before the next (and only other) meal of the day may be a substantial departure from accepted professional practice, preventing summary judgment for defendants regarding Rowe’s claim of deliberate indifference to avoidable pain caused by the timing of his medication. See Sain v. Wood, supra, 512 F.3d at 894-95. Since Rowe’s pain strongly indicated that he was experiencing reflux, the reflux could have had serious medical consequences (up to and including cancer) in addition to inflicting chronic pain on him. Prisoners aren’t supposed to be tortured.
In citing even highly reputable medical websites in support of our conclusion that summary judgment was premature we may be thought to be “going outside the record” in an improper sense. It may be said that judges should confine their role to choosing between the evidentiary presentations of the opposing parties, much like referees of athletic events. But judges and their law clerks often conduct research on cases, and it is not always research confined to pure issues of law, without disclosure to the parties. We are not like the English judges of yore, who under the rule of “orality” were not permitted to have law clerks or other staff, or libraries, or even to deliberate — at the end of the oral argument in an appeal the judges would state their views seriatim as to the proper outcome of the appeal.
We don’t insulate judges like that, but we must observe proper limitations on judicial research. We must acknowledge the need to distinguish between judicial web searches for mere background information that will help the judges and the readers of their opinions understand the case, web searches for facts or other information that judges can properly take judicial notice of (such as when it became dark on a specific night, a question we answered on the basis of an Internet search in Owens v. Duncan, 781 F.3d 360, 362 (7th Cir.2015), citing WeatherSpark, “Average Weather On September 22 For Chicago, Illinois, USA: Sun,” https://weatherspark.com/ averages/30851/9/22/Chicago-Illinois-United-States), and web searches for facts normally determined by the factfinder after an adversary procedure that produces a district court or administrative record. When medical information can be gleaned from the websites of highly reputable medical centers, it is not imperative that it instead be presented by a testifying witness. Such information tends to fall somewhere between facts that require adversary procedure to determine and facts of which a court can take judicial notice, but it is closer to the second in a case like this in which the evidence presented by the *629defendants in the district court was sparse and the appellate court need only determine whether there is a factual dispute sufficient to preclude summary judgment.
Rule 201 of the Federal Rules of Evidence makes facts of which judicial notice is properly taken conclusive, and therefore requires that their accuracy be indisputable for judicial notice to be taken of them. We are not deeming the Internet evidence cited in this opinion conclusive or even certifying it as being probably correct, though it may well be correct since it is drawn from reputable medical websites. We use it only to underscore the existence of a genuine dispute of material fact created in the district court proceedings 'by entirely conventional evidence, namely Rowe’s reported pain.
There is a high standard for taking judicial notice of a fact, and a low standard for allowing evidence to be presented in the conventional way, by testimony subject to cross-examination, but is there no room for anything in between? Must judges abjure visits to Internet web sites of premier hospitals and drug companies, not in order to take judicial notice but to assure the existence of a genuine issue of material fact that precludes summary judgment? Are we to forbear lest we be accused of having “entered unknown territory”? This year the bar associations are busy celebrating the eight hundredth anniversary of Magna Carta. The barons who forced King John to sign that notable document were certainly entering unknown territory, and risking their lives to boot. Shall the unreliability of the unalloyed adversary process in a case of such dramatic inequality of resources and capabilities of the parties as this case be an unalterable bar to justice? Must our system of justice allow the muddled affidavit of a defendant who may well be unqualified to be an expert witness in this case to carry the day against a pro se plaintiff helpless to contest the affidavit?
This is not the case in which to fetishize adversary procedure in a pure eighteenth-century form, given the inadequacy of the key defense witness, Dr.' Wolfe. Let’s review: Wolfe refused to continue Rowe’s Zantac prescription in July 2011 while Rowe was being kept waiting for three weeks before being seen by a doctor. Wolfe knew Rowe had esophagitis: he reviewed Rowe’s medical records, which contained the 2009 diagnosis and revealed nearly two years of physicians’ having prescribed Zantac for him continuously. Wolfe had personally prescribed Zantac for Rowe for six months of those two years and must have known that the Department of Correction authorizes such treatment only for a serious health condition. Rowe was complaining of continuing reflux pain; and while Wolfe denied a prescription renewal on July 13, he demonstrated his awareness that Rowe might need treatment by scheduling him for a later appointment (the August 2 appointment) to evaluate his request to resume taking Zantac.
Against this background, to credit Wolfe’s evidence that it doesn’t matter when you take Zantac for relief of GERD symptoms (evidence that may well have failed to satisfy the criteria for the admissibility of expert evidence that are set forth in Fed.R.Evid. 702) just because Rowe didn’t present his own expert witness would make no sense — for how could Rowe find such an expert and persuade him to testify? He could not afford to pay an expert witness. He had no lawyer in the district court and has no lawyer in this cpurt; and so throughout this litigation (now in its fourth year) he has been at a decided litigating disadvantage. He requested the appointment of counsel and of an expert witness to assist him in the *630litigation, pointing out sensibly that he needed “verifying medical evidence” to support his claim. The district judge denied both requests, leaving Rowe unable to offer evidence beyond his own testimony that he was in extreme pain when forbidden to take his medication with his meals.
The web sites give credence to Rowe’s assertion that he was in pain. But the information gleaned from them did not create a dispute of fact that was not already in the record. Rowe presented enough evidence to call Dr. Wolfe’s assessment into question — Rowe claims that after his medication was switched to the 12-hour schedule he was in extreme pain and Dr. Wolfe, without examining Rowe or disclosing the basis for his opinion (as we require experts to do), stated cursorily that the medicine would be effective for 12 hours. It will be up to the factfinder to decide, on a better developed record, who is right.
Nor is pain the only concern. Esophageal reflux disease can lead to serious damage of the stomach or esophagus, and even to cancer.
It is heartless to make a fetish of adversary procedure if by doing so feeble evidence is credited because the opponent has no practical access to offsetting evidence. To say for example that however implausible Dr. Wolfe’s evidence is, it must be accepted because not contested, is to doom the plaintiffs case regardless of the merits simply because the plaintiff lacks the wherewithal to obtain and present conflicting evidence. Rowe did not move to exclude Wolfe as an expert witness on the ground that Wolfe neither qualified to give expert evidence in this case (because he is not a gastroenterologist) nor, as a defendant, was likely to be even minimally impartial. But Rowe does not have the legal knowledge that would enable him to file such a motion.
We have decided to reverse the judgment. We base this decision on Rowe’s declarations, the timeline of his inability to obtain Zantac, the manifold contradictions in Dr. Wolfe’s affidavits, and, last, the cautious, limited Internet research that we have conducted in default of the parties’ having done so. We add that the judge erred not only by giving undue weight to Wolfe’s internally contradictory affidavit but also by relying on a defendant (Wolfe) as the expert witness. There are expert witnesses offered by parties and neutral (court-appointed) expert witnesses, but defendants serving as' expert witnesses?— and in cases in which the plaintiff doesn’t have an expert witness because he doesn’t know how to find such a witness and anyway couldn’t afford to pay the witness? And how could an unrepresented prisoner be expected to challenge the affidavit of a hostile medical doctor (in this case really hostile since he’s a defendant in the plaintiffs suit) effectively? Is this adversary procedure?
Esophagitis is a common disease for which Zantac is a common treatment, and it makes common sense as well as medical sense that a drug for treating symptoms of stomach acid backing up into the esophagus would be administered shortly before or shortly after meals unless the massive 300-mg pill was being administered to the patient, and it was not in this case. Rowe claimed that the Zantac he took became ineffective in treating his esophagitis pain symptoms when the prison staff decided to give it to him only long before his meals. His pain and the timing recommendation of the Mayo Clinic that we mentioned earlier suggest that giving 150-mg doses of Zantac six and a half hours before and five and a half hours after meals may be a substantial departure from accepted professional practice. But without his own expert, Rowe couldn’t counter Wolfe’s as*631sertion that Zantac does not need to be taken shortly before, or with or shortly after, a meal in order to be effective. As Rowe explained in his brief, while he “provided evidence that Zantac does not ‘prevent’ reflux during its 12 hours of effectiveness, and that it was not effective at relieving Rowe’s symptoms, the district court accepted the word of a defendant [i.e., Dr. Wolfe], who was speaking as an ‘expert,’ that the treatment Rowe received was adequate and effective. Had an expert been appointed, the expert would, have confirmed Rowe’s factual representations, and would have supported Rowe’s objection that the defendant lacks personal knowledge about the condition(s) Rowe had because Wolfe never physically examined Rowe or had diagnostic testing done on Rowe” (citations omitted).
Rowe’s allegations alone were sufficient to preclude summary judgment, and were enhanced by the defendants’ own evidence, which included both Wolfe’s contradictory evidence (among other things, he asserted that Rowe does not need Zantac and yet prescribed it for him) and the absurd opinion by the medical director that over-the-counter medications should not be provided to prisoners. Allowing Wolfe to be an expert witness in the case despite his being a defendant and not practicing the medical specialty at issue was another boost to the plaintiffs case, though again not one that an unrepresented, indigent prisoner could exploit.
We are coming to the end of this long opinion but we need to change gears for a moment: Besides arguing deliberate indifference to a serious medical need, Rowe accuses several of the defendants, in particular Dr. Wolfe and Nurse Bagienski, of retaliating against him for filing a lawsuit. He says they told him that going without Zantac for a month would make him “think twice about bringing lawsuits about inadequate medical care.” If indeed they said this — an issue that cannot be determined without a trial — Rowe has a solid claim of retaliation. The retaliation claims against the other defendants were properly dismissed, however, and likewise the deliberate-indifference claims against the following defendants, who the district court correctly found were not responsible for the failure to treat Rowe’s medical condition competently — Rose Vaisvilas, Wayne Scaife, and Kenneth Hysell. But we reverse with regard to the remaining defendants and remand the case for further proceedings consistent with this opinion.
Although reversing, we are not ordering that judgment be entered in Rowe’s favor. As we’ve explained, we are not invoking Fed.R.Evid. 201 and thus not taking judicial notice of any facts outside the district court record. The remaining defendants are entitled to try to rebut any evidence whether or not presented in the district court, including any evidence found on the Internet. Like the conventional forms of evidentiary inquiry, Internet research must be conducted -with circumspection. In particular it must not be allowed to extinguish reasonable opportunities for rebuttal.
Pure adversary procedure works best when there is. at least approximate parity between the adversaries. That condition is missing in this case, in which a pro se prison inmate, incapable of retaining an expert witness (expert witnesses usually demand to be paid — and how would this inmate even find an expert witness?), confronts both a private law firm and the state attorney general.
Because of the profound handicaps under which the plaintiff is litigating and the fact that his claim is far from frivolous, we urge the district judge to give serious consideration to recruiting a lawyer to repre*632sent Rowe, see Miller v. Campanella, supra, at 880, 2015 WL 4523799 at *2; Perez v. Fenoglio, 792 F.3d 768, 784-85 (7th Cir. July 7, 2015); appointing a neutral expert witness, authorized by Fed.R.Evid. 706, to address the medical issues in the case; or doing both. We are mindful that district courts don’t have budgets for paying expert witnesses. But the medical issues in the case are not complex; there should be no difficulty in the judge’s persuading a reputable gastroenterologist to speak to Rowe and some of the prison medical personnel (Rowe’s prison is only 30 miles from Indianapolis, and there are 128 gastroenterologists in or near Indianapolis, healthgrades, www.healthgrades.com/ gastroenterology-direetory/in-indiana/ indianapolis), to sit for a deposition, and, if necessary, to testify. Rule 706(c)(2) states that a court-appointed expert “is entitled to a reasonable compensation, as set by the court,” and that “the compensation is payable ... in any ... civil case [not involving just compensation under the Fifth Amendment] by the parties in the proportion and at the time that the court directs- — and the compensation is then charged like other costs.” In light of Rowe’s indigency, the court if it appoints its own expert witness will have to order the defendants to pay the expert a reasonable fee if the expert is unwilling to work for nothing. Most prisons are strapped for cash, and this is something for the district court to bear in mind in deciding on whether and how large a fee to order the defendants to pay a court-appointed expert witness in a case (such as this case) that has sufficient merit to warrant such an appointment.
A substantial academic literature identifies serious deficiencies in the provision of health care in American prisons and jails. See, e.g., Andrew P. Wilper et al., “The Health and Health Care of U.S. Prisoners: Results of a Nationwide Survey,” 99 Am. J. Public Health 666 (2009), and the studies posted by the Academic Consortium on Criminal Justice Health, www.accjh. org/. On the quality of treatment problems of Corizon, the employer of Dr. Wolfe and the other medical staff members sued by Rowe, see David Royse, “Medical Battle Behind Bars: Big Prison Healthcare Firm Corizon Struggles to Win Contracts,” April 11, 2015, www.modemhealthcare.com/ article/20150411/MAGAZINE/304119981; also Human Rights Defense Center, Prison Legal News, “Corizon Needs a Checkup: Problems with Privatized Correctional Healthcare,” March 2014, www.prisonlegal news.org/news/2014/mar/15/eorizon-needsa-checkup-problems-with-privatizedeorrectionalhealthcare/. The present case illustrates the problems that this literature has identified.
Affirmed in Part, Reversed in Part, and Remanded
Appendix
We respectfully suggest that the dissenting opinion is misleading in certain respects that require a response; page references are to pages in the dissent.
Page 29: The dissenting opinion states that “the reversal is unprecedented, clearly based on ‘evidence’ this appellate court has found by its own internet research.... When the opinion is read as a whole, the decisive role of the majority’s internet research is plain.” No, the majority opinion endeavors to make clear that Rowe’s allegations alone, coupled with the affidavit of Dr. Wolfe and other defense evidence, would be enough without any reference to the Internet to preclude summary judgment for the defendants, and doubtless would have precluded summary judgment had Rowe been represented. The dissent ignores this part of the majority opinion.
*633Page 29: The reader is told that “the majority writes that adherence to rules of evidence and precedent makes a ‘heartless ... fetish of adversary procedure.’ ” That is not what the majority opinion says; it says: “It is heartless to make a fetish of adversary procedure if by doing so feeble evidence is credited because the opponent has no practical access to offsetting evidence” (emphasis added). Nowhere does the majority opinion deny the validity of the federal rules of evidence or of procedure.
Page 32: The proposition in the dissent that the prison’s response was adequate as long as it “provided at least some treatment for pain” overlooks the fact that a 150-mg Zantac pill given six and a half hours before one’s next meal provides, according to Rowe, no alleviation of pain caused by stomach acid backing up into the esophagus, which is the pain of which Rowe complains. Also, it can’t be correct that providing “some” treatment of pain always gets a prison doctor off the hook. Suppose Rowe were in agony from a slipped disk; would it be enough for Dr. Wolfe to give him an aspirin? To tell him, if he broke his leg, that it would heal by itself, in time?
Page 35: The statement that the majority opinion “holds in essence that the district judge erred by not doing such independent factual research” is mistaken. There is no such holding or suggestion in the opinion. The opinion merely suggests that the district judge should have appointed, and on remand should appoint, an expert witness who is a gastroenterologist (as Dr. Wolfe, the defendants’ principal witness, is not) and who also is not a defendant.
Pages 35-36: The dissent’s citation of Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), as a celebration of traditional adversary procedure misses the significance of Daubert, which is that it enlarged the role of the judge in policing expert testimony. The district judge in this case failed to play the role envisaged in Daubert by treating Dr. Wolfe as an expert on GERD despite his being a defendant accused of neglecting Rowe’s GERD and also his not being a gastroenterologist. A Daubert hearing would doubtless have led to his exclusion from an expert-witness role.
Page 39: The dissent says that “when a prisoner brings a pro se suit about medical care, the adversary process that is the foundation of our judicial system is at its least reliable. Few prisoners have access to lawyers or to expert witnesses needed to address medical issues.” Right on! (And Rowe is not one of the few who does have the necessary access.) But affirmance of a quite possibly incorrect decision cannot be the correct solution to the problem thus correctly stated by the dissent. The majority opinion offers a modest solution — a remand to enable a competent, impartial evidentiary exploration of Rowe’s claim.
Page 40: On this page the dissent repeats its contention that the majority is insisting that district judges conduct Internet research: “The majority clearly implies, while denying it is doing so, that the district judge herself should have done the independent factual research the majority has done on appeal, questioning an unchallenged expert affidavit....” No; the district judge should have recognized the existence of a substantial issue of material fact, barring summary judgment. Rowe’s evidence of pain contradicted Dr. Wolfe’s affidavit.
Page 41: The dissent expresses concern that the defendants may have to pay most or all of an expert witness’s fee in a case *634brought by an indigent prisoner, such as Rowe. But it seems unlikely that a gastroenterologist would charge more than a nominal fee merely to testify thafi — what appears to be obvious — in order to prevent serious esophageal pain (and the even more serious consequences that can ensue from untreated GERD) a 150-mg Zantac pill should be taken no more than an hour before eating — not six and a half hours. One has only to read the label on a box of 150-mg Zantac pills to learn when the pill should be taken to prevent pain — 30 to 60 minutes before eating. In addition, an expert’s fee, if any, would in a case such as this, with its numerous defendants, be split many ways or, more likely, be paid for by the Indiana Department of Correction, the State of Indiana, Corizon or its liability insurer, or individual defendants’ malpractice insurance (depending on the contractual arrangements between Corizon and the state, as well as the parties’ insurance arrangements), or some combination of these well-heeled entities.
Page 42: The dissent states: “Without an expert witness qualified to present the facts and opinions the majority finds persuasive, that information does not come into evidence.” This implies that without an expert witness, a party cannot defeat a motion for summary judgment. That isn’t true. If a jury believed Rowe, he would win. It would be more likely to believe him than to believe Dr. Wolfe.
Page 42: The parade of horribles on this and other pages of the dissent (such as page 35, discussed earlier in this Appendix) is based on a belief that the majority is ordering that the district judge on remand do her own Internet research. Not so. It is unlikely that any Internet research by anyone will be necessary. All that should be necessary is testimony by a qualified, impartial expert witness who is a gastroenterologist and is not a defendant in this litigation.
Page 43: The dissent again states that we are requiring judges to conduct their own factual research. No. We are even accused by the dissent of trying to turn judges into substitutes for physicians. Again no.
Page 45: The dissent appears to misunderstand the Mayo Clinic’s advice to “take one [Zantac pill] in the morning and one before bedtime.” As pointed out in the majority opinion, this advice is intended “only for patients taking the prescription strengths,” whereas Rowe was taking the 150-mg strength that is available over the counter. The Mayo Clinic provides different advice for the 150-mg pill: that it should be taken 30 to 60 minutes before meals to prevent heartburn symptoms (the mildest GERD symptoms). The dissent does not mention Boehringer Ingelheim’s advice, also quoted in the majority opinion, that while Zantac can be taken at any time “to relieve symptoms,” in order “to prevent symptoms” it should be taken “30 to 60 minutes before eating food or drinking beverages that cause heartburn.” That is, if you have pain, you take a pill right away to alleviate the pain; if you foresee pain as a result of eating or drinking, you take the pill before you eat or drink — but not six and a half hours before.
Page 45: The dissent’s reference to taking Zantac for more than “two weeks” without a doctor’s permission is irrelevant to the case because Rowe had a doctor’s permission — indeed Dr. Wolfe’s permission — to take Zantac and had begun taking it long ago, always with permission.
Page 45: The reference to symptomatic relief beginning “24 hours” after taking Zantac could be understood to mean that Zantac can prevent pain that far in advance. Not so. As explained in the majority opinion, “24 hours” is the time it *635takes for Zantac when first taken to begin to have á therapeutic effect.